Abstract
Psychiatry today must gain a better understanding of the common and distinct pathophysiological mechanisms underlying psychiatric disorders in order to deliver more effective, persontailored treatments. To this end, it appears that the analysis of ‘small’ experimental samples using conventional statistical approaches has largely failed to capture the heterogeneity underlying psychiatric phenotypes. Modern algorithms and approaches from machine learning, particularly deep learning, provide new hope to address these issues given their outstanding prediction performance in other disciplines. The strength of deep learning algorithms is that they can implement very complicated, and in principle arbitrary predictorresponse mappings efficiently. This power comes at a cost, the need for large training (and test) samples to infer the (sometimes over millions of) model parameters. This appears to be at odds with the as yet rather ‘small’ samples available in psychiatric human research to date (n < 10,000), and the ambition of predicting treatment at the single subject level (n = 1). Here, we aim at giving a comprehensive overview on how we can yet use such models for prediction in psychiatry. We review how machine learning approaches compare to more traditional statistical hypothesisdriven approaches, how their complexity relates to the need of large sample sizes, and what we can do to optimally use these powerful techniques in psychiatric neuroscience.
Introduction
Current diagnostic and prognostic schemes in psychiatry need improvement. While current diagnostic methods are optimized for reliability, the underlying neurobiology is complex and variable [1, 2], both because the etiology of psychiatric disorders is highly diverse [3], and because the brain and behavior are per se highly complex systems involving multiple levels of temporal and spatial granularity and millions of nonlinear feedback loops. It has been argued that a diverse array of biophysical and biochemical factors may give rise to similar functionality at the level of the neuronal dynamics underlying behavior and, vice versa, that the same changes in neural dynamics may produce different behavioral outputs depending on the context [4]. These observations may partly explain why only a subgroup of patients respond to drug or psychotherapeutic treatment approved for any given disorder [5,6,7]. Personalized forms of therapy therefore require a different characterization that supplements categorical conventional diagnoses. This approach could be based on the analysis of large patient cohorts that include sufficient heterogeneity covering a wider range of personal disease histories, and on transdiagnostic and multilevel approaches to the identification of pathological mechanisms underlying mental illness [8], that is, on the integration of many different data modalities [9,10,11], from genetic and molecular information (‘omics’) to brain and behavioral data. On the other hand, individual subjectlevel information needs to be integrated with such an account when designing personalized therapies.
There have been high hopes recently that artificial intelligence (AI) algorithms, in particular from the field of deep learning (DL), can meet these challenges. DL algorithms excel in processing highly complex data within which data features may interact at multiple levels and in highly nonlinear ways. In consequence, when combined with large amounts of data, they may have an enormous potential for healthcare services (see [9] for review). For instance, deep neural networks (DNNs) are remarkably successful at tasks requiring object or scene recognition [12,13,14] and natural language processing [15, 16]. DNNs have shown human to superhuman performance in challenging board games by inferring rules mostly from ‘own experience’, playing the game against themselves, rather than from expert knowledge [17]. This ability to automatically learn relevant higherlevel representations from raw data, also referred to as automatic feature extraction, is one central aspect which makes the application of DNNs in biomedical areas attractive [9, 18, 19]. DNNs have for instance already been successfully adopted to automatize skin and breast cancer detection [20, 21]. In the sector of mental health, recent studies have begun to harness the potential of DNNs and ‘big data’, especially in domains which are particularly datarich, such as online social media platforms or smartphone and mobile sensor based data (e.g., [22,23,24,25,26]). Attempts to collect big data in other data domains, including for instance data on brain structure and function, genetics, or behavior on cognitive tasks, have been actively pushed by different consortia and funders (e.g., ENIGMA, ABIDE, ADNI, ADHD200, OASIS, ABCD).
Building big multimodal data bases is certainly an important step in identifying coherent patient subgroups in an unsupervised manner, gaining a better mechanistic understanding by acknowledging interactions and connections between different levels of analysis, and for personalizing treatments. But how much data do we need, when is ‘big’ big enough? Scientific data sets are often relatively small, carefully harvested in thoughtfully designed experiments, even when many of them are combined into common data bases. Even methods that generate a high volume of data, such as transcriptomics or neuroimaging, are often applied to a limited number of human subjects. Can DNNs efficiently be used on such comparatively small data bases? And how is the apparent conflict resolved between the need to process huge data sets on the one hand side, yet to construct subjectlevel models taking very individual information into account on the other hand? The present article attempts to address some of these questions, from a statistical and machine learning (ML) perspective, and discusses some of the factors that play a role both on the data side, as well as on the side of the models used for analysis.
Models in statistics and ML
As discussed in the previous section, psychiatric research needs to address a variety of related challenges, including the identification of biomarkers for robust diagnostics, the identification of subgroups with shared disease characteristics (biological and psychological features) and common therapeutic response profiles, and personalization of treatments through subjectlevel predictions of potential outcomes and disease trajectories. These efforts are underpinned by the search for a deeper understanding of the neurobiological mechanisms underlying aberrant cognitive and emotional function across disorders, and the design of effective medication and intervention strategies based upon these insights (e.g., [10, 27]). From a statistical point of view, the former set of challenges may be formulated in terms of regression or classification problems, or in terms of unsupervised detection of structure (clustering) [28], while the latter, deeper scientific, questions may be supported by combining statistical and ML techniques with computational modeling [29, 30].
For a classification problem, for instance, we may want to predict treatment responses or symptom severity from brain activation during a cognitive task, or to distinguish between individuals diagnosed with different psychiatric disorders based on structural features of the brain (e.g., [31]). In both of these cases we can express our problem in terms of a relationship between an output variable \(y\) (e.g., clinical diagnosis), or a set of output variables \(y\) (e.g., severity of different symptoms), and a set of input variables or features x (e.g., functional activation in different brain areas). In principle, these input variables or features may come from different modalities (e.g., measures of structural connectivity vs. functional activation vs. polygenetic risk scores). If the outputs \(y\) are categorical class labels, e.g., clinical diagnoses, we call this a classification problem, while a regression problem refers to the situation where outcomes \(y\) are continuously (real) valued, or at least bear ordinal relationships (like the natural numbers) as, e.g., ratings on scales of symptom severity. Both regression and classification problems are examples of supervised settings, for which not just the input data x but also the outputs y are known for a socalled training set.
In contrast, if, for instance, we question current diagnostic schemes and would like to identify novel types of clinically relevant groupings in feature space x unbiased by current nosological knowledge, we call this an unsupervised setting, a domain of ML and computational modeling.
In both types of settings we often formulate the problem in terms of a mathematical model of the data, either in terms of a functional relationship f_{θ} between x and y in the supervised case, where θ denotes parameters of this function (e.g., regression weights), or just in terms of the data itself in the unsupervised case. An (point) estimate of this function (denoted \(\hat f\)), or more specifically of its parameters (denoted \(\hat \theta\)), is obtained by changing these parameters such that some form of loss function is minimized (or some optimality criterion is maximized), a process called model training or, in statistical terms, model estimation or inference (in Bayesian inference, we would seek to determine the full [posterior] distribution across parameters θ, not just a point estimate). Such a loss function could be, for instance, the meansquarederror (MSE) in the Gaussian case, i.e., the average sum of squared deviations between true and predicted outputs, or the negative loglikelihood of the data, which quantifies how likely it is to observe the current data given some estimate of the parameters. In the unsupervised case, a loss function may, for instance, be some type of measure that formalizes an idea of structure in the data, e.g., specifying the betweengroup vs. withingroup distances for any possible assignment of data points to groups.
Hypothesis testing vs. prediction
There is no principal difference between models in statistics and ML, and both can be used, in principle, for either hypothesis testing or prediction (see Fig. 1). This distinction between hypothesis testing and prediction, however, is indeed important: while traditionally statistics has been more concerned with hypothesis testing, machine learners have been more interested in prediction [32]. In classical statistical hypothesis testing, we evaluate a probabilistic statement about the data, often formulated in terms of parameters of the model (e.g., that certain regression coefficients are equal to zero), and aim to obtain a probability for how likely a certain state of affairs (related to the null hypothesis) holds in the whole population of potential observations (which may be finite or infinite) given the model assumptions. This probabilistic inference is based on the observed training data alone. In prediction, in contrast, we aim to forecast future previously unobserved outcomes, e.g., the likely output \(y\)^{(new)} given a new observation x^{(new)} (also called test data if used to formally evaluate the prediction error (PE), see Section ‘Model complexity, sample size, and generalization’).
Another, related issue here is whether the stated models are probabilistic or deterministic: For hypothesis testing at some level there are always random variables and probability distributions involved, such that relationships between input and output variables, for instance, are formulated in terms of moments of probability distributions, e.g., \(\mu _y := E\left[ {y{\mathrm{}}{\boldsymbol{x}}} \right] = f_\theta ({\boldsymbol{x}})\), where f_{θ} is the function that maps variables x onto the conditional mean (expectation value) \(\mu_{y}\) of the distribution of \(y\), and θ are its parameters (e.g., regression coefficients). For prediction, the function f_{θ} does not necessarily have to express a probabilistic relationship, i.e., we may just have \(y\) = f_{θ}(x), expressing outcomes \(y\) directly as some (deterministic) function of the features x. However, in modern ML probabilistic models are getting more and more popular (sometimes termed statistical ML), as they also provide a sense of the uncertainty associated with predictions. Although this enables formal hypothesis testing in principle as well, these models and their associated probability distributions are often tedious and difficult to handle.
Finally, while in statistics the functions or models f_{θ} are usually quite simple and/or allow for precise and unique analytical solutions, meaning that we can obtain an exact and unique solution to the optimization problem through ‘paper & pencil’ derivations, in ML the functional relationship f_{θ} may be quite complex, like a DNN. While the latter are potentially much more powerful in detecting and utilizing complex, higherorder nonlinear feature combinations for prediction, they are unfortunately also often much less interpretable than if f_{θ} were a simple linear function.
DNNs and the universal function approximation theorem (UAT)
DNNs likely constitute the most powerful class of ML models, at least from a mathematicalcomputational perspective, and in their most basic and most commonly employed form are deterministic. In visual terms, they can be understood as networks of artificial neurons, units, or nodes arranged in a feedforward chain of layers, termed a feedforward neural network (FNN; Fig. 2a), with each node computing some nonlinear function f (the socalled activation function) on the weighted sum of its inputs. Symbolically, this corresponds to a function f_{θ} that may be written as a deep nesting of multiple nonlinear functions \(y = f(f\left( {f\left( {f \ldots f\left( {\boldsymbol{x}} \right) \ldots } \right)} \right))\) arranged in a chain.
In a first stage, an input layer receives the information about the inputs or predictors x (e.g., regional gray volume, or connectivity between areas [33,34,35]), which is then propagated forward via ‘synaptic’ connections with specific connection weights through one or multiple hidden layers, up to an output layer which represents the prediction \(\hat y\) of the true (but unknown) outcome \(y\). A neural network (NN) is often called ‘shallow’ if it contains just one or two hidden layers, and ‘deep’ if there are more of them. A DNN is trained by adjusting all its connection weights (the model’s parameters θ) such that the error between predicted (\(\hat y\)) and true (\(y\)) outcomes is minimized across a training set for which the true \(y\) are known (for reviews see [12, 15, 36]), a process in which successive hidden layers of the network tend to learn more and more abstract representations of the data (e.g., edges and corners on early layers for visual images and fully segmented object representations on deeper layers, cf. [13, 37, 38]), much like the ventral visual processing stream of the human brain [39]. There are several different types of DNN architectures and models, such as multilayer perceptrons, convolutional neural networks (CNN), or deepbelief networks, to some of which we will return below (see [36]).
Often in neuroscience and psychiatry we deal with sequential or time series data, where either sequences of inputs and outputs may have to be mapped onto each other (as in language), or some informative characteristics of the temporal structure are to be extracted. Not only do measurements in psychiatric and neuroscientific research often come as time series, e.g., in the form of functional magnetic resonance imaging (fMRI), electroencephalography (EEG), or mobile sampling data, or as sequential behavioral responses across trials of an experiment, but mental illness is a temporally dynamic and evolving phenomenon per se [4, 40], with quite heterogeneous temporal trajectories across individuals [3]. Just like classical statistics extends the class of regression models to the time series domain by also regressing values y onto their own past, as in autoregressive movingaverage (ARMA) models or ARMA models with exogenous inputs (ARMAX), NNs can be extended to the time series domain by incorporating previous function outputs, \({\boldsymbol{y}}_t = f_\theta ({\boldsymbol{y}}_{t  1},{\boldsymbol{x}}_t)\). These devices are called recurrent neural networks (RNN), since they do not only include feed forward but also recurrent connections between units, i.e., activity may propagate back and forward among units. This means activity can reverberate in RNN just as in the real brain, and they can produce sequences of outputs completely on their own, like giving answers to questions as in common virtual assistants. Mathematically, RNN constitute discretetime dynamical systems and they come with a whole set of novel properties that pure feedforward NNs lack (see, e.g., [4]). Some researchers refer with ‘depth’ in RNN more to their temporal depth (in contrast to the ‘spatial depth’, i.e., the number of layers), by which one means the temporal lags or the time scales across which dependencies among observations and temporal structure can be detected by the system [41]. Some in this sense, deep RNN architectures have been purpose designed to bridge long temporal delays, such as long shortterm memory (LSTM) [15] or gated recurrent unit (GRU)based networks [42].
NNs with just one nonlinear hidden layer have a surprising mathematical property that all simpler statistical models, like the classes of general or generalized linear models, lack: In principle they can represent or approximate arbitrarily closely any continuously valued function \(y\) = f(x) between predictors and outcomes, according to the much celebrated universal approximation theorem (UAT) due to Cybenko [43], Hornik et al. [44], and Funahashi [45] (and similar theorems exist for noncontinuous mappings, like binary outcomes; [46]). That is, whatever the true underlying functional relation \(y\) = f(x) in the real data is, a NN with just one hidden layer would be able to represent it (see Fig. 2a)! This of course raises the question why including more than one hidden layer, as in DNNs, is a sensible thing to do. It turns out that shallow NNs with just one layer and deep NNs fundamentally differ in how the number of units required to approximate a given function grows with the required accuracy of the approximation: While under some conditions the number of units required to achieve a given level of accuracy may only grow algebraically in the number of layers, exponentially more units may be required within a given layer to achieve that same level [47,48,49,50]. Besides these computational reasons, it has been observed that DNNs are capable of representation learning or automatic feature extraction, i.e., can construct the most useful representations of data themselves directly from the raw data across successive layers. For instance, when trained on facial images, a DNN will learn to represent simple features such as edges and nodes in early layers, then eyes and noses in later layers, and finally entire faces [37]. Without prior knowledge the model identifies noses and eyes as predictive features of faces.
We conclude by pointing out that similar theorems as for function representation in feedforward NNs exist for RNN as well: RNN can approximate, in principle, arbitrarily closely any dynamical system that may have generated the true time series observed [45, 51, 52], and could represent any Turing machine [53]. While these theorems ascertain that in principle any feedforward or timedependent (dynamical) function could be implemented in terms of NNs with as little as one hidden layer, they make no statements about the difficulties involved in finding that implementation, or how much data are required to achieve an approximation of satisfying accuracy. Increasing the number of units or layers, or more generally the complexity of the function f_{θ}, will enable to approximate more complicated functions to the degree of accuracy desired, but it will generally also increase the sample size needed for model estimation or training. This is because model complexity and sample size are intimately related, as we will discuss next.
Model complexity, sample size, and generalization
In healthcare, when we try to determine a diagnosis or prognosis, or when we seek to identify novel biomarkers, we ultimately care less about hypothesis testing but more about prediction. We demand that a model trained on a set of training data also performs well if we apply it to new observations not contained in our previous training set, i.e., helps with the right diagnosis or prognosis, or with determining the best form of therapy. In other words, the goal is to select the model which will minimize the error when predicting outcomes for unseen individuals based on the learned relationship in the training data. As with the criteria used for model training, the PE may be based on different types of loss functions, for instance the MSE loss or a likelihoodbased criterion. There are at least three different types of prediction we need to distinguish (see Fig. 3): The insample PE refers to the situation where we keep one part of the data fixed, e.g. the predictors, and aim to determine the expected deviation between a new set of true and predicted outcomes for this given set of predictor values (Fig. 3c). A more interesting quantity is the outofsample PE, where we train the model on some data and then draw a new sample to evaluate the PE (Fig. 3c; [28, 54]). In this process, we often assume that the new sample has the same statistical properties as the training sample, i.e., was drawn from the same probability distribution. This may not be the case, however, with important implications for clinical practice. Here we call this the ‘outofdomain’ PE (Fig. 3d), which can only be determined realistically if we have data from different domains, or if we have a good mechanistic model of the processes underlying our sample.
When we ask ‘how big is big enough?’, we are really asking how large a sample should be, and which properties it should have, in order to be able to infer a model with acceptably low PE. Except for the simplest types of statistical models, like linear Gaussian models (GLM), we cannot simply compute the sample size required to achieve a given PE, since the probability distributions and expectation values involved in this computation are analytically intractable. This is because in ML we are usually dealing with (highly) nonlinear models and consequently more complex probability distributions. Alternatively, one may think of numerical sampling (MonteCarlo) techniques to evaluate the required expectations, but even these are often out of the question since in machine and specifically DL we are commonly dealing with such highdimensional variable and parameter spaces that sampling is not computationally feasible [55]. So explicit determination of required sample sizes is not possible for most problems of practical relevance, but what we can do is try to obtain a PE estimate for a given model.
Biasvariance tradeoff and model complexity
Why is the training error, which we can compute directly from the sample at hand, not a good measure for the quality of our model? For a given model with a given number of parameters it indeed is, and so it is completely justifiable to determine model parameters such that the training loss (e.g., the negative likelihood) is minimized. However, it is not a good estimate of the loss we could expect when applying our model to a new sample, and hence is not suitable for selecting among different models with different numbers of parameters. Reasonably complex models such as polynomial basis expansions or multilayer NNs can fit (i.e., approximate) any function, and hence any given set of training data, to an arbitrary degree, making the training error in fact zero for a sufficient number of parameters (see Fig. 4b, c). This is even true for simple linear models if the number of predictor variables and parameters are as large as or larger than the number of observed outputs (simple example: if you have observed just one predictor/ output pair \(\{ x,y\}\) and consider the linear model \(y\) = β_{0} + β_{1}x, then you can find infinitely many solutions for parameters \(\{ \beta _0,\beta _1\}\) that result in an exact fit with zero error; however, only one of these will be the one that describes the true relation between predictors and outputs in the population, as illustrated in Fig. 4c). A regression model with the same number of outcomes as predictors, or with a sufficiently large number of parameters in a reasonably powerful model, can produce a curve that goes through every single data point. At some point such a model will capture the entire variability in the data including noise, implying that it will ‘interpret’ pure noise as systematic and meaningful fluctuation. This phenomenon is also known as overfitting. Overfitting implies large variance in the predictions as each time we draw a new sample, we will obtain a new model, as illustrated in Fig. 4c.
In classical hypothesis testing, we assume that we have a reasonably accurate model of the data to begin with, and perform all probability calculations under this assumption (Fig. 1). In many empirical situations, and particularly in psychiatric research, this approach has not brought on the desired progress [27, 56]. Recently the focus has therefore shifted toward adopting ML approaches to infer more complex models directly from data (see e.g., [9, 27, 31, 56,57,58] for reviews). A complex model is capable of learning a broader range of functional relationships between features and outcomes (Fig. 2b), and is therefore more likely to fit the training data well (see Fig. 4b). In statistical terms, the model will exhibit low bias, by which we mean the systematic deviation between the true datagenerating function and the best possible model estimate of the function, i.e., \(( {f( x )  {\mathrm{E}}[ {\hat f_\theta ( x )} ]} )^2\). For squared error loss and identically and independently distributed (i.i.d.) data, the expected test error can be precisely decomposed into this bias, the above mentioned variance, and an irreducible noise term [54]. One can roughly think of model complexity as a measure of how versatile and flexible a model may align to the data (sometimes referred to its capacity [36]; see Fig. 2b and Fig. 4b). Ideally, we want to select a model as powerful and flexible as possible, yet balancing bias and variance in an optimal manner. This is where sample size considerations and ‘big data’ come in. Larger data sets enable learning ever more intricate relationships from the data, as they allow for more complex models with lower bias while keeping the variance down (Fig. 4a). Sample size effectively shifts the tradeoff between bias and variance such that more complex models could be inferred without compromising the PE (see Fig. 4a).
While the biasvariance tradeoff is a core concept in traditional statistical learning theory and determines model selection, recent empirical observations with DL models surprisingly suggest that once models are strongly overfit, beyond the point where a perfect match to the training data is obtained, they may actually lead to even better generalization [59,60,61,62]. That is, quite counterintuitively, after the test error first reaches a maximum within the overfitting regime (Fig. 4a), it tends to decline again as model complexity is further increased, leading to a ‘double descent’ curve [63]. In this regime all models almost perfectly fit, or interpolate, the training data. However, importantly, this only occurs if a socalled regularization term is included in the optimization function, which leads to ‘smooth’ function fits by implicitly biasing the training process toward simpler models which exhibit a smaller norm [63]. However, the precise mathematical mechanism underlying this phenomenon is still not fully understood (e.g. [61, 64, 65]).
Model complexity moreover, is not that straightforward a concept, and different definitions exist [54]. It is not merely related to the nonlinearity of a model, or the number of its parameters (except if we stay within a certain model class like GLMs). One can easily give examples of two models with equal number of parameters but with different expressiveness in the sense that one can approximate a larger set of functions than the other. Often statisticians talk about the effective number of parameters or degrees of freedom to express the idea that the true degrees of freedom may depend on the functional relationships and constraints involved ([54], although this is in itself a tricky topic [66]). One quite intuitive concept of model complexity proposed in statistical learning theory, is the Vapnik–Chervonenkis dimension, which quantifies how many data points a model function can neatly separate (or ‘shatter’; [67]).
Model selection
To select a specific model, or its number of parameters, among a larger class of models, we need an estimate of the outofsample PE. Attempts have been made to derive analytical formulae to obtain such an estimate [68, 69], but mostly numerical methods based directly on the data need to be used to produce a reliable estimate. Here we will not review this topic in its entirety, but rather focus on a few commonly used methods that illustrate how sample size and model complexity come into play.
Analytical formulae for model selection usually estimate the PE based on the training error adjusted or penalized by a term that expresses the (average) optimism of the training error. The idea is that the training error is an overly optimistic estimate of the expected test error (as noted above), and so by approximating this optimism and adding it to the training error we should obtain a better PE estimate. In fact, the methods around usually only provide an estimate of the insample PE (see above), i.e., the error evaluated across new outcomes sampled at the same data points used for training [54].
Popular analytical formulae are the Akaike information criterion (AIC; [68]) and the Bayesian information criterion (BIC). Denoting by L_{MLE} the value of the data loglikelihood evaluated at the maximum likelihood estimator \(\hat \theta _{{\mathrm{MLE}}}\), the AIC is given by AIC = −2L_{MLE} + 2k, from which we immediately see how the number of parameters k penalizes model complexity. The sample size N affects the AIC indirectly through the loglikelihood: For a constant number of parameters k, the likelihood term will gain importance as N increases (its numerical size will increase) such that the second, penalizing term becomes less relevant. This example illustrates quite directly how data fit (training loglikelihood) and model complexity (number of parameters) are traded off against each other. Unfortunately, these methods often provide only relatively crude PE approximations (e.g., [70, 71]), and AIC and BIC have been observed to overfit and underfit, respectively [72, 73].
Probably the most popular numerical method for estimating the outofsample PE, and somewhat of the current ‘gold standard’, is cross validation (CV). In CV one trains a model on a larger fraction of the available data, say 90%, and then tests model performance on the 10% leftout data that have not been used for model training, thus obtaining an outofsample PE. In Kfold CV, this process is repeated for each of the K = 10 × 10% data fragments in turn [74], i.e., each 10% section is put aside once for testing while training the model on the remaining 90%, in this way making full use of all the data available for both training and testing across ten iterations. The final PE estimate is then the average across all ten runs. Other ratios of training to test set size are of course possible. The extreme case where only a single data point is left out for testing is called ‘leaveoneout’ CV. It turns out that CV is itself subject to the biasvariancetradeoff with the proportion of leftout test data as a free parameter [28, 54, 75].
CV could be used to either select a model or assess the outofsample PE [54, 76], but cannot be used for both simultaneously. When we use Kfold CV to determine the model with lowest CV error which we select for further use among, say, M tested models, we need to be aware that its associated CV error will likely be an overly optimistic PE estimate: Using the CV error both for model selection and model assessment represents a kind of ‘doubledipping’ which will lead to an estimated PE lower than the actual one just by chance, as we had M different attempts to compute a CV error [28, 54, 77,78,79]. To compute a true outofsample PE estimate in this case, we should really split the data into three segments, one for training (training set), one for model selection (validation set), and put one aside purely for PE assessment (test set). Alternatively, when data are scarce a nested CV scheme may be applied in which model assessment and selection are carefully separated by an outer (assessment) loop which separates the data into training, validation, and test set, and an inner (selection) loop, which shuffles only the training and validation set [79]. Nonetheless, leaking data used for model selection into the model assessment step is perhaps the most common mistake made in the literature (see also [77]).
If we have too few data, we may not be able to afford a separate test set or just a small test set. The larger the test set, of course, the lower will be the uncertainty about the mean PE estimate, i.e., its variance or standard error (see also [80]). In fact, too small test (and training) samples may be one explanation why we observe a counterintuitive negative correlation between model classification accuracy and sample size across psychiatric studies [81,82,83]. Large uncertainty in the PE estimate in combination with publication bias may have resulted in the predominant reporting of high accuracy estimates for small samples (apart from reasons related to larger sample homogeneity within small sample size studies).
To summarize, as our sample increases in size, the variance (standard error) of estimated model parameters will decrease. As a consequence, we can afford more complex models which come with lower bias. Where exactly this tradeoff is optimized, needs to be determined for the specific data and class of models at hand by formal procedures like CV. Other properties of the data, like the amount of irreducible noise or the type of distribution the data were drawn from, will also affect the required sample sizes. If the distribution is very broad, multimodal or with long heavy tails, we may need larger samples.
Crosssite and outofdomain prediction
When we obtain an outofsample PE estimate, we assume that any new data we would like to use our model on comes with the same statistical/ distributional properties. Essentially this means that all sources of variability in the data need to be the same across samples, that is, we need truly random samples from the whole population to which we would like to generalize. Variability in a feature (e.g., reduction in brain volume) may emerge from multiple sources such as disease heterogeneity (e.g., reduction is not present across all individuals), biological variability (e.g., brain volume is itself quite variable and may even correlate with other confounds such as age), or from measurement noise (e.g., the assessment of brain volume is noisy). Differences in inclusion criteria may constrain disease related or biological variability, while different measurement devices (for example different MRI machines in multisite studies) may generate systematically different errors across samples and can thereby result in violations of common distributional assumptions (i.i.d., see also Fig. 3). This in turn may lead to one of the biggest dangers involved in building clinically relevant prediction models, as amply illustrated in the recent literature [81,82,83,84]: If, for instance, models are trained and tested on data from the same clinical site or group, or a consortium which employs common procedures among its members, the model may learn predictive but disease irrelevant sitespecific characteristics [82, 85]. Of course this reasoning also applies to inferences drawn through hypothesisdriven approaches and could explain heterogeneities across studies by differences in distributional properties of the investigated samples.
Particular challenges for time series and sequential data
Our goals in time series analysis may be twofold: On the one hand, we may just want to extract temporal features from a time series, like the power in different frequency bands or the functional connectivity, which we would then like to use as predictors in a classification or regression model. In that case, assuming we have time series from N independent subjects, we could simply proceed as outlined earlier, since the model ultimately used for prediction is not itself a time series model, but a feedforward model that simply uses features extracted from N independent time series as inputs. Often, however, our goal is to forecast a time series, for instance, we may want to predict stock market shares, or a future patient trajectory from medical records with sequential entries across time [86], or from mobile data like various sensors and ecological momentary assessments [87]. In these cases, we have to consider that time series and sequential data come with their own specific problems since consecutive measurements across time are usually highly dependent, violating the assumption of i.i.d. data that underlies most of statistical testing. Due to these autocorrelations (and potentially nonstationarity) in the data, it is not as straightforward to split the data into K folds and perform CV [88]. For instance, we cannot just randomly leave out some fraction of data points, since this would destroy the temporal contingencies on which time series models rest (they are built to detect the temporal structure and use it for forecasting). Even if we leave out temporally contingent time series segments, the question arises how to train the model across the resulting temporal gap. Finally, any leftout segment will be highly correlated with other segments, at least with the directly preceding one, implying that it will not constitute an independent test set as is the basis for determining the PE by CV. If time series from N different subjects (or reasonably independent trials) are available, we could instead run the same strategy as above and train models on, say, 90% of the subjects, fix the parameters, and test their prediction performance on the 10% leftout subjects. This comes with additional issues that we will only briefly touch upon here: First, time series data in biology and psychology are generated by some larger underlying dynamical system, which we only partially observe [4, 89, 90]. When we apply our trained time series model to new observations, our estimate of the initial condition (which we need to run the time series model) may therefore be highly ambiguous, often implying a prohibitively large variance in the predictions. Second, especially for time series generated by dynamical systems it is in fact an open question, which metric would be most suitable for assessing prediction performance: In a chaotic dynamical system, for instance, temporal trajectories quickly diverge even when we have captured the true underlying system with our model, rendering conventional MSE or likelihoodbased measures directly evaluated on the time series unsuitable [4, 89,90,91].
Model training, computational efficiency, and searching complex optimization landscapes
Another point to consider is computational and numerical issues involved in inferring statistical and ML models from data. Unlike simple statistical models for which analytical or straightforward and fast numerical procedures with unique solutions often exist, the optimization landscapes for many ML algorithms, DNNs in particular, may be highly complex, high dimensional, and rugged, such that optimization becomes a serious challenge. Potentially, even if we knew that model A would in principle be the one that optimizes the biasvariancetradeoff for a given data set (Fig. 4a), the specific point in its multivariate loss function that optimizes this tradeoff may be extremely difficult to find in practice, requiring a lot of computational resources. In general, more complex models take (much) longer training times, and while on the one hand side, big data may be required to sufficiently specify some complex model, on the other hand, they come with a particular computational burden especially for complex models. Hence, one needs to be aware that the applicability of more complicated models like DNNs is not merely limited by the sample sizes required to meet the biasvariancechallenge, but also by the additional issues involved in finding a nearoptimal solution (which is usually not that unique in DNNs, see [92]), and the computational costs that come with it.
In sum, regression and classification problems in psychiatry will likely require learning complex mappings between features and outcomes, to integrate across data from multiple domains, and to combine both temporal and spatial information. However, sample size may curtail inferring models of the required complexity by virtue of the biasvariance tradeoff and computational issues involved in both finding minima of loss functions for complex models and in computing hardware and temporal resources.
DL from big and small data in psychiatry
While DNNs are capable of revealing complex but highly predictive feature combinations, they commonly have a large number of parameters, somewhere between hundreds into the many millions. From the discussion in the previous section one may deduce that for such models a huge amount of data is needed to battle the biasvariance tradeoff. For instance, in image processing tasks where DNNs with eight hidden layers and over 60 million parameters (and more units) are commonly employed, over 15 million labeled images are used for training [14]. These are sample sizes which are just not available in psychiatry, in particular when expensive and laborious techniques, such as neuroimaging during cognitive tasks, are involved. However, this does not mean that we cannot employ DNNbased methods in psychiatry. A DNN framework for data analysis consists of (1) the model architecture, (2) a loss function, (3) a training algorithm, and (4) the data itself on which the DNN is to be trained. We can in fact tune all four of these components to make DNN approaches applicable to small or medium sized data sets, as will be briefly reviewed next.
Network architecture
The specific network architecture determines which class of functions can be computationally efficiently approximated [36]. Choosing a suitable model architecture, by easing the training process, may therefore help to reduce the demands on sample size. In a sense, we are making use of prior domain knowledge to offset potential limitations of the data, a similar strategy as in Bayesian approaches to model inference.
CNNs are an example of networks deliberately designed for processing image information in computer vision [36]. Inspired by the primate visual system [12], CNNs are set up to exploit spatial invariances in an image for extracting feature maps, using units with spatial ‘receptive fields’ (i.e., localized spatial filters). Each feature map is learnt by combining information across multiple receptive fields, using the same set of shared connection ‘weights’. This weightsharing principle, exploiting the insight that images could be decomposed into features reoccurring at multiple spatial positions, substantially reduces the number of tobetrained parameters. CNNs are probably the most popular class of DNN models in medicine when developing classifiers on the basis of imaging data (see [9, 10] for reviews). In psychiatry and neurology they have been used, for instance, to classify disorders based on anatomical brain images obtained through MRI (e.g., [93]), functional brain images or measures derived from it like functional connectivity (e.g., [94,95,96]), or for combining structural and functional neuroimaging data (e.g., [97,98,99]).
Similar to what CNNs are for the visual domain, in the temporal domain LSTMs are specifically ‘engineered’ systems that enable to extract longterm dependencies in the time series through special ‘memory cells’ and multiplicative gates, which control the information flow into and out of these memory cells [15]. In psychiatry, LSTMs have for instance been used to forecast depressive and manic states in bipolar patients based on mobile data [26], to detect mental disorders from speech [100,101,102,103], to discriminate between psychiatric patients and healthy controls [104, 105] (see [106] for GRU based approach), or to process text passages from social media platforms to identify subjects at high risk for drinking alcohol [22].
Another central property of the NN design is the specific form of the units’ activation function (Fig. 2c top). For instance, for many if not most problems rectifiedlinear unit (ReLU) activation functions represent a particularly efficient choice because they facilitate the training process for specific mathematical reasons (see below, sect. on training algorithms). Luckily, the UAT holds for ReLU functions as well [107]. In fact, the choice of activation function, ReLUs in particular, may be more important than any other network design feature [36, 108].
Choice of loss function and regularization techniques
The choice of loss function is primarily determined by the scale level of the data (e.g., continuous, ordinal, or categorical data), and by whether we are working within a statistical framework or in a more deterministic ML framework. In ‘conventional’ deterministic ML we often simply go with the MSE criterion, which may be interpreted as a Gaussian loglikelihood under the assumption of a constant identity covariance matrix (thus yielding no insight into true uncertainty). In a statistical framework, we usually desire to model distributions across data which comes with measures of uncertainty, and hence use likelihoodbased criteria or Bayesian approaches for model training. Bayesian criteria come with special benefits that affect required sample sizes, but they also make model training more tedious such that most NN optimization is based on likelihoodbased approaches (in ML often phrased in terms of the negative loglikelihood, also referred to as crossentropy for categorical data). However, a statistical approach often entails that we have to treat the network’s hidden activation states as random variables as well, socalled latent variables [109], which implies that we often can only use approximations to the loglikelihood such as the socalled evidence lower bound [55, 110, 111]. While statistical approaches and criteria often take (much) longer for model training, they provide full probability distributions across the data and, in fact, may capture important relations within the data much better [36].
Besides the general consideration of whether we would like to work in a statistical or a deterministic ML framework, the loss function may be modified in particular ways to encourage the training algorithm to find solutions, which reduce the effective number of parameters or model complexity in a specific way. This is called regularization, and more generally has been defined as ‘any modification we make to a learning algorithm that is intended to reduce its generalization error but not its training error’ [36]. The most popular techniques are L1 and L2 regularization, the latter also known as ‘weight decay’, ridge or Tikhonov regularization. L1 regularization adds the sum of absolute parameter values, \(\lambda {\sum }_{j = 1}^P \left {\theta _j} \right\), to the loss function, while L2 regularization adds the squared parameter values, \(\lambda {\sum }_{j = 1}^P \theta _j^2\), where the weight λ > 0 controls the relative importance of the regularization (or penalty) term in the loss function, e.g. \({\rm{Loss}} =  \log p\left( {{\mathbf{X}}{\mathbf{\theta }}} \right) + \lambda {\sum }_{j = 1}^P \left {\theta _j} \right\) for an L1 penalty added to the negative loglikelihood. L1 and L2 regularizations are also often used even for simple linear regression models, where the former is known as the ‘least absolute shrinkage and selection operator’ [112], and the combination of L1 and L2 regularizatiosn has been termed ‘elastic net’ [113]. For high values of λ, model parameters will be forced toward 0, where the L1 penalty will make some of them exactly 0 eventually (hence they drop out from the model), while L2 regularization on the other hand tends to shrink parameters associated with features which show low covariation with the outcomes (see e.g., [33, 114] for examples in psychiatry).
L1 and L2 regularizations are common in all types of statistical and ML models, including FNNs and RNN. For DNNs, however, also more specific techniques have been developed to prevent overfitting and encourage sparser and less complex solutions. One particularly effective method is parameter dropout. Here, a fraction of units is temporarily removed at random from the network (e.g., by multiplying their output with 0 and effectively dropping them from the loss function) such that only a ‘thinned’ network is trained at any one time [115]. For testing, all units are then reinstated with their outputs weighted by the probability with which these units were present during training. Dropout seems to drive units to learn more robust representations [115], is computationally inexpensive, and works for both FNNs and RNN [36]. Additional more purpose tailored regularization solutions have been proposed to solve the longterm dependency problem in ‘vanilla’ RNN, e.g., by directly modifying the loss function (e.g., [89]) or indirectly via specific parameterizations of the weight matrices (e.g., [116,117,118]). Importantly, many of these approaches achieve performance comparable or superior to that of LSTMs with often far less parameters, and are on top more easily interpretable, for instance in terms of the underlying dynamical system and its properties. An intuitive understanding of how networks represent information, e.g., how RNN store memory and implement dynamical systems, can sometimes guide such tasktailored regularization schemes. For instance, by regularizing only a fraction of RNN parameters toward line attractor configurations, we can force a network to store both short and longrange dependencies [89]. Recovering features encoded at both fast and slow temporal frequencies is imperative to identify signatures of aberrant brain function, as different frequencies carry different information (e.g., α or γwaves in EEG [119]). Likewise, mental health related features inferable from mobile devices and sensors can be found at both fast and slow frequencies (e.g., typing dynamics [24] vs. sleepwake cycles [120]). Regularization approaches gain importance as sample size decreases and can remarkably improve generalization, as seen for instance in terms of improved dynamical system reconstruction [89].
Training algorithm
We have already identified the training (optimization or inference) algorithm as another potential bottleneck that could also affect the required sample size. There are several steps that we can take to improve finding an acceptable local minimum or solution. First, any training algorithm begins with an initial draw of parameter estimates, also referred to as initialization. Naively, one may think of randomly sampling initial parameter estimates which representatively cover the entire (high dimensional) parameter space, but in most cases this is computationally highly demanding and infeasible. A great deal of research has therefore focused entirely on developing efficient initialization procedures in DNNs [36]. Hinton et al. [121] introduced a clever training technique in which layers are pretrained one after another such that network parameters are already sensibly initialized before the full training of the whole network (see Fig. 5). While this sounds like a rather minor modification, this insight contributed strongly to the groundbreaking success of DNN algorithms. Another larger body of research focuses on socalled annealing approaches [122, 123]. Here, the loss function is gradually modified throughout training such that the training algorithm is first guided into regions of parameter space where generally higher likelihood (lower loss) solutions are to be found, which are then iteratively refined. For example, in socalled Boltzmann machines, a specific type of generative NN model, the ‘energy landscape’ is made initially very flat to encourage the system to escape from local minima, and then gradually steepened, called simulated annealing [123].
As another example, for fully probabilistic models, meaning models which treat both the observations and latent (hidden) variables as random variables, the variational annealing approach proposes to gradually increase the ratio between observation and latent variable noise in the loss function, that is, to decrease the relative noise in the hidden variables across training iterations [124, 125]. The idea is that initiating the latent variable mappings with very high noise (i.e., low precision), essentially makes the optimization criterion (in the limit) a quadratic and convex function of the observations, and thus easy to solve. As the ratio is slowly increased, putting a stronger emphasis on the latent variable model fit, more and more hidden configurations inconsistent with the data slowly ‘freeze’ out. Rather than steepening the overall ‘energy landscape’ as in simulated annealing (i.e. cooling the overall temperature, or variance), this approach gradually decreases the relative temperature of the hidden variable loss.
In addition, the specific procedure by which parameters are updated can have an important impact (cf. section 'Model complexity, sample size, and generalization'). Perhaps the most defining aspects of a training procedure are (1) how it scales with data size and parameters, (2) which information in the data it exploits, (3) which steps it takes to escape local minima, and (4) how it deals with regions of differing slope in the loss function. Perhaps the most popular training scheme for DNNs is stochastic gradient descent (SGD) [126]. The idea behind gradient descent in general is that in order to move toward a local minimum of a function, we simply need to follow steps proportional to the negative gradient of that function. SGD makes use of this principle, but rather than computing the gradient across the entire data set, it computes the gradient from a small subset of (randomly drawn) samples, or minibatches, thus injecting some noise into the training process that may help to avoid local minima. Particularly for large amounts of data, SGD is computationally efficient and comparatively fast. As noted above, SGD training further profits from certain choices of a neuron’s activation function, like the ReLU: ReLUs have a (piecewise) constant gradient everywhere which eases issues with widely differing slopes in a model’s loss function during gradientdescent training, while sigmoidal activation functions do not only have gradients varying quite strongly across their range of inputs (Fig. 2c bottom), but in particular tend to saturate for very small or very large inputs, making gradientbased training more difficult [36].
Widely varying gradients may, however, also be compensated by taking higherorder information (that is, e.g., 2ndorder derivatives) into account: Through this, approaches such as expectation–maximization or Gauss–Newton methods [28, 109], although computationally (much) more challenging, may be more efficient in finding minima and may thus benefit function approximation based on smaller samples. Another strategy to deal with varying slopes is to adapt the learning rate, a factor with which the gradient is scaled during each step of SGD, locally or across training iterations. Various such algorithms with stepsize regulation have been proposed (e.g., Adam or AdaGrad, see also [127]).
Lastly, we point out that modern DNN research has come up with some general procedures to promote model generalizability independent of inference frameworks, which are now part of many standard protocols. These approaches include early stopping or adversarial training procedures (see [36] for more examples). In early stopping, training is stopped when the validation (rather than the training) set error ceases to decrease for some time. The validation error is evaluated every couple of steps and a copy of the associated latest parameter settings are stored such that it can be returned to as training continues [36]. Adversarial training, on the other hand, directly attempts to find weak spots in a model by searching for slight input perturbations which will cause large deviations in the output, and thus could ‘fool’ the model [128, 129]. Szegedy et al. [129] show how such slight (undetectable to humans) distortions of, for example, a dog image can make a NN falsely predict an ostrich. Deliberately searching for and training networks against such flaws renders them more robust to small perturbations around the neighborhood of the training data.
There are also various steps we can take with the data itself to ease the burden on the model side and encourage solutions that better generalize to new observations [36]. One idea is to reduce the dimensionality of the data or preprocess it in some way such that fewer model parameters are needed and the burden on the model training framework to discover most useful representations of the data by itself is reduced. Based on our own domain knowledge, we may preselect features, which we deem to be highly informative. For instance, using the average regional gray matter volume from sMRI images as features will spare the NN the work of identifying and representing distinct segregated regions based on individual voxel values. Most psychiatric studies handselect features beforehand, for instance, by computing functional connectivity values from the BOLD time series, saving the NN the work of learning which temporal representations are relevant [33], or summarizing crucial genetic information in terms of polygenic risk scores and therefore bypassing the need to locate or detect significant genetic variations or polymorphisms based on the entire genome [130]. In fact, any data processing step may be seen as a kind of feature selection, including preprocessing, rescaling, or selection of regions of interest, since each of these steps involves certain decisions about what is important about the data.
However, the downside of such preprocessing and feature selection based on domain knowledge could be that we overlook important and highly predictive aspects of the data, integrate them away or average them out in some way. In some sense this contradicts the spirit of DL which is supposed to find useful data representations and features on its own [12]. It may therefore be more fruitful to provide the entire voxel level data to DNNs (e.g., [114, 131]).
An alternative and perhaps less biased approach to manual feature engineering could be to ‘automatize’ the process. This may include anything from popular linear dimensionality reduction techniques like principal component analysis, metric or nonmetric multidimensional scaling [28, 109], or simple latent variable models like factor analysis, to nonlinear dimensionality reduction techniques like locally linear embedding [132] and Isomap [133], or more recent methods like ‘tdistributed Stochastic Neighbor Embedding’ (tSNE) [134] or autoencoders [135, 136]. Autoencoders (AE), for instance, are NNs which project higherdimensional input data to a lowerdimensional latent space (encoder part), where this lowerdimensional data representation is optimized such that the original input is reconstructed from it with least loss at the output layer (decoder part; Fig. 5, left panel). Hence this whole design can be thought of as a highly nonlinear dimensionality reduction technique that aims to produce a latent representation of most informative nonlinear feature combinations [135, 136]. Gupta et al. [137] were among the first to use an AE combined with a CNN for classification of a neurological disorder. The AE effectively extracted lowlevel image features later successfully used to assess Alzheimer’s Disease. Pinaya et al. [138] moreover trained an AE to extract features from brain volume data on a large sample of healthy individuals (n > 1000). Interestingly, this trained (unsupervised) AE could predict brain volume alterations in patients suffering from schizophrenia or autism (n < 100) as compared to control, suggesting that the AE did indeed extract mentalhealth related features. The approach demonstrates a clever way of making use of (relatively) large already available and openly accessible data sets from healthy individuals for disease classification, or potentially even to gain insight into pathological mechanisms in smaller samples.
Rather than reducing the input dimension, we could also artificially increase the sample size and variation within the sample, an approach termed data augmentation. For image data, this includes rotations, translations, rescaling, flipping, shearing, or stretching of the original images, or simply adding noise (see also [139,140,141] for speech recognition examples). The idea behind these operations is that they will assist the network in learning invariant, more general representations, robust under certain transformations and conditions where data are only partially observable or noisy. For recognizing a smile on a face, for instance, it should not matter whether the image is blurry, or the face is presented upside down (see [142,143,144] for examples in neuroimaging and psychiatry).
Another data augmentation strategy involves generative models, that is, models which contain probabilistic latent variables and by virtue of that can—if properly trained—generate data with the same distributional properties as the original data. One such framework that recently became popular for this purpose are socalled generative adversarial networks (GANs) [145]. GANs attempt to approximate the true datagenerating distribution by training two networks in competition with each other, a generator and a discriminator network. The generator attempts to create data samples as similar as possible to the true data while the discriminator strives to distinguish true from fake (generated) samples. The two networks coevolve throughout training, and by attempting to fool the discriminator, the generator, if successful, learns to approximate the datagenerating distribution from which new (simulated) data samples may be generated and used for training [146]. Similarly, one could use generative models like GANs to fill in missing values in multimodal data sets, a common problem in psychiatry, rather than discarding an entire multivariate data point (see e.g., [147]). Along another line, Nguyen et al. [148] used GANs to unbias MRI images from different sites by successfully transforming images from one site into those from another. Such approaches could help to make more efficient use of larger crosssite data sets which often suffer from sitespecific heterogeneities. In cases in which we have access to a large unlabeled data set, rather than simulating data, we may also choose to augment the data set by semisupervised learning approaches such as pseudo labeling [149]. Here a network is first trained on labeled data, then unlabeled data is fed through it to obtain predictions (pseudo labels), and finally the network is trained on the entire (augmented) data set.
Transfer learning is another technique for improving the data situation by transferring knowledge gained in one data domain to a current problem setting that we expect to share some statistical characteristics with the transfer domain [150]. For instance, rather than training a DNN from scratch on an object recognition task each time one faces a new problem setting, machine learners frequently make use of already publicly available trained DNN models like AlexNet [14] or VGGNet [151], and simply finetune parameters on their current task. Lu et al. [152] have extended this approach to structural brain recordings and apply AlexNet to identify pathological images. In another example, Thomas et al. [153] train a DNN to decode cognitive states of participants during a working memory task. They demonstrate how pretraining their network on six other unrelated cognitive tasks considerably improves network performance when compared to random initialization. This sort of pretraining or transfer learning saves data resources and training time that would otherwise be required to learn common (often low level) features which the network would have needed to extract anyway (like edges and nodes in an image, [154]). It can also be understood as a feature selection step conducted by another model rather than the data analyst, and that is not fixed but will further be adapted to the current setting through training. There are examples where a CNN designed to classify a neurological condition based on sMRI images has been shown to perform even better when pretrained on natural images rather than on sMRI data itself, perhaps because natural images are in a sense richer in lowlevel features also needed to classify sMRI images more effectively [137]. In general, pretraining on any datarich domain that could be expected to share some statistical distributional properties with the target data set may profoundly help in using complex DNN even for smaller samples. Here, open access data may be of huge assistance. Models could be pretrained on similar, or ideally perhaps even the same disorder as targeted in a current application, and then merely finetuned on the target data (see also [93, 155]).
Pretraining could prove particularly useful when individualizing models for single subjects in order to provide patientcentered (treatment) predictions (Fig. 5). Transferring knowledge already gained from other data sets could prove very valuable in building complex and nonetheless robust individualized models. For instance, we could first train models on data sets across individuals and use the inferred parameters as efficient initializations, which would help finetune quite complex predictive models at the single subject level (Fig. 5).
Another way to transfer knowledge from other domains or tasks to the current problem setting is metalearning. While different definitions exist, metalearning is most commonly understood as a paradigm where the system ‘learns to learn’, that is, which optimizes the learning process of an algorithm itself via multiple learning episodes or tasks (see [156] for a recent thorough review). For instance, Andrychowicz et al. [157] demonstrate how the optimization procedure of a model can itself be optimized via gradient descent and thereby outperforms handcrafted optimization algorithms in many different settings. The optimization process of the metalearning algorithm can refer to almost any part of the model, including architecture, parameter initialization, and many more, and may be realized through different optimization procedures such as gradient descent, reinforcement learning, or evolutionary algorithms [156]. One particularly interesting aspect is that such algorithms have succeeded in designing classifiers, which are capable of learning in only a few shots, i.e., from few data instances [158]. It is conceivable that along similar lines pooling multiple psychiatric data sets and using metalearning principles could yield sets of classifiers which learn quickly on new problem settings.
Many of the approaches discussed, like manual feature engineering, transfer learning, or specific network designs tailored to particular tasks, may be seen as different ways of utilizing prior knowledge to facilitate NN training and reduce the sample size requirements. This aligns with the more general idea in Bayesian frameworks for model training, where previous knowledge is incorporated in a statistically principled way through prior distributions on the parameters [28, 159]. While such approaches come with the danger of biasing the resulting model or parameter estimates in the wrong direction, they are on the other hand known to potentially strongly reduce the variance in the resulting parameter estimates and to protect to some degree against overfitting [160]. It turns out, in fact, that some of the common regularization approaches, like ridge regression, can be derived within a Bayesian approach that places certain priors on the parameters [161]. Hence, Bayesian inference strategies can be used to both import prior knowledge from the same or different data domains into the current parameter estimation, and to regularize models.
In sum, the sample size needed to successfully train a DNN will depend on multiple factors such as the type of data, network size and architecture, type of stochasticity in the data, dimensionality of the feature space, regularization schemes, and the actual target function the DNN is supposed to learn, to name but a few. Power calculations are simply not available for highly nonlinear models like DNNs with complicated likelihood functions and probability distributions, and hence any suggestions regarding sample size could only be based on examples in the literature employing very similar architectures. Perhaps the largest bundle of work to date has been performed in the field of classification based on neuroimaging data. In this field, samples of several hundred participants appear to provide a good starting point for successfully training DNNs, with accuracies roughly around 70% for multisite studies and binary classification problems (see [162] for a recent review). These results raise hope that future DNN applications may prove valuable for sample sizes available in psychiatry. Samples below N = 200 produce very heterogeneous performance results, and usually do not contain data from multiple sites, making it difficult to judge models in this range [162]. Recommendations for sample sizes for other applications are more difficult to provide. In general, we strongly recommend that authors employing DL techniques conduct thorough evaluation on prediction errors and their standard error themselves, e.g., by iteratively increasing the test set size (as done e.g., in [80]).
Future research directions
Psychiatry is in urgent need of approaches that enable tailored precision therapies. For designing efficient treatments, we also require a better understanding of the neurobiological mechanisms underlying pathology at a transdiagnostic level. While more traditional hypothesisdriven statistical approaches to these issues have not brought the necessary breakthroughs, modern ML algorithms like DNNs provide new hope given their outstanding performance in other medical domains. At first sight, the complexity (and thus computational strength) of DNNs comes at a cost—large sample sizes. However, as we tried to discuss here, there are several ways to make DNNs suitable even for much smaller sample sizes. We have discussed various concrete steps to enable the development of efficient schemes using complex models for individualized personcentered predictions (see also [9, 87]). Models first trained on group data may provide one future avenue (Fig. 5), if it can be achieved that these capture sufficient particularities at the (individualized) singlesubject level to yield meaningful forecasts, and not just reflect common group characteristics.
A deeper understanding of hidden network representations in DNNs, i.e. ‘opening the black box’, could on the other hand reveal new insights or generate new hypotheses regarding pathological neurobiological mechanisms. Indeed, several studies have already demonstrated that DNN representations may yield interpretable features (e.g., [33, 94, 99, 163]). For instance, by examining the weights of their DNN, Zeng et al. [94] observed that corticalstriatalcerebellar functional connectivity features were most relevant to the classification of schizophrenia. After training a deep AE on brain volume data from a large set of healthy individuals, Pinaya et al. [138] assessed the region specific reconstruction error made by the network when predicting psychiatric patients to pinpoint the most relevant brain regions involved in separating patients from controls. Li et al. [163] developed a visualization framework to decipher regions of interest important in the detection of individuals with autism spectrum disorder compared to controls based on fMRI recordings. Visualization approaches for assessing DNNs are currently a hot topic in ML, and future developments in this direction may help uncover interpretable multimodal biomarkers of psychiatric disease. The interplay between the bench and the bedside, pathophysiological understanding and tailored treatment, continues in the age of AI, aided by the new tools discussed in this paper.
Funding and disclosure
GK received funding from the German Research Foundation (DFG; TRR265: A06 & B08). AML received funding from the German Research Foundation (DFG; ME 1591/41; grant 5485966 Collaborative Research Center 636 subproject B7; grant 255156212 Collaborative Research Center 1158 subproject B09; grant GRK 2350/1 Research Training Group 2350 subproject B2; grant TRR 265 Collaborative Research Center subproject S02) and from the German Federal Ministry of Education and Research (BMBF; grant 01EF1803A, and grant 01ZX1904A). DD received funding from the German Research Foundation (DFG; Du 354/82; TRR265: A06; Du 354/101). AML has received consultant fees from Boehringer Ingelheim, Elsevier, Brainsway, Lundbeck Int. Neuroscience Foundation, Lundbeck A/S, The Wolfson Foundation, Bloomfield Holding Ltd, Shanghai Research Center for Brain Science, Thieme Verlag, Sage Therapeutics, v Behring Röntgen Stiftung, Fondation FondaMental, JanssenCilag GmbH, MedinCell, Brain Mind Institute, Agence Nationale de la Recherche, CISSN (Catania Internat. Summer School of Neuroscience), Daimler und Benz Stiftung, American Association for the Advancement of Science. In addition, he has received speaker fees from Italian Society of Biological Psychiatry, MerzStiftung, Forum Werkstatt Karlsruhe, Lundbeck SAS France, BAG Psychiatrie Oberbayern, Klinik für Psychiatrie und Psychotherapie Ingolstadt, med Update GmbH, Society of Biological Psychiatry, Siemens Healthineers. Open access funding provided by Projekt DEAL.
References
McTeague LM, Huemer J, Carreon DM, Jiang Y, Eickhoff SB, Etkin A. Identification of common neural circuit disruptions in cognitive control across psychiatric disorders. Am J Psychiatry. 2017;174:676–85.
Goodkind M, Eickhoff SB, Oathes DJ, Jiang Y, Chang A, JonesHagata LB, et al. Identification of a common neurobiological substrate for mental illness. JAMA Psychiatry. 2015;72:305–15.
Kendler KS, Zachar P, Craver C. What kinds of things are psychiatric disorders? Psychological Med. 2011;41:1143–50.
Durstewitz D, Huys QJ, Koppe G. Psychiatric illnesses as disorders of network dynamics. Biol Psychiatry: Cogn Neurosci Neuroimaging. 2020. https://doi.org/10.1016/j.bpsc.2020.01.001.
Rush AJ, Trivedi MH, Wisniewski SR, Nierenberg AA, Stewart JW, Warden D, et al. Acute and longerterm outcomes in depressed outpatients requiring one or several treatment steps: a STAR* D report. Am J Psychiatry. 2006;163:1905–17.
Wong EH, Yocca F, Smith MA, Lee CM. Challenges and opportunities for drug discovery in psychiatric disorders: the drug hunters’ perspective. Int J Neuropsychopharmacol. 2010;13:1269–84.
Hofmann SG, Asnaani A, Vonk IJ, Sawyer AT, Fang A. The efficacy of cognitive behavioral therapy: a review of metaanalyses. Cogn Ther Res. 2012;36:427–40.
Insel T, Cuthbert B, Garvey M, Heinssen R, Pine DS, Quinn K, et al. Research domain criteria (RDoC): toward a new classification framework for research on mental disorders. Am J Psychiatry. 2010;167:748–51. https://doi.org/10.1176/appi.ajp.2010.09091379.
Durstewitz D, Koppe G, MeyerLindenberg A. Deep neural networks in psychiatry. Mol Psychiatry. 2019;24:1583–98.
Vieira S, Pinaya WH, Mechelli A. Using deep learning to investigate the neuroimaging correlates of psychiatric and neurological disorders: Methods and applications. Neurosci Biobehav Rev. 2017;74:58–75.
Calhoun VD, Sui J. Multimodal fusion of brain imaging data: a key to finding the missing link (s) in complex mental illness. Biol Psychiatry: Cogn Neurosci Neuroimaging. 2016;1:230–44.
LeCun Y, Bengio Y, Hinton G. Deep learning. Nature. 2015;521:436–44.
Farabet C, Couprie C, Najman L, LeCun Y. Learning hierarchical features for scene labeling. IEEE Trans Pattern Anal Mach Intell. 2013;35:1915–29.
Krizhevsky A, Sutskever I, Hinton GE. Imagenet classification with deep convolutional neural networks. Advances in neural information processing systems. Curran Associates, Inc.; 2012. p. 1097–105.
Schmidhuber J. Deep learning in neural networks: an overview. Neural Netw. 2015;61:85–117.
Graves A, Mohamed Ar, Hinton G. Speech recognition with deep recurrent neural networks. Acoustics, speech and signal processing (icassp), 2013 ieee international conference on. 2013. p. 6645–49.
Silver D, Schrittwieser J, Simonyan K, Antonoglou I, Huang A, Guez A, et al. Mastering the game of go without human knowledge. Nature. 2017;550:354.
Litjens G, Kooi T, Bejnordi BE, Setio AAA, Ciompi F, Ghafoorian M, et al. A survey on deep learning in medical image analysis. Med image Anal. 2017;42:60–88.
Mamoshina P, Vieira A, Putin E, Zhavoronkov A. Applications of deep learning in biomedicine. Mol Pharm. 2016;13:1445–54.
Esteva A, Kuprel B, Novoa RA, Ko J, Swetter SM, Blau HM, et al. Dermatologistlevel classification of skin cancer with deep neural networks. Nature. 2017;542:115–18.
Cireşan DC, Giusti A, Gambardella LM, Schmidhuber J. Mitosis detection in breast cancer histology images with deep neural networks. International conference on medical image computing and computerassisted intervention. 2013. p. 411–18.
Hassanpour S, Tomita N, DeLise T, Crosier B, Marsch LA. Identifying substance use risk based on deep neural networks and Instagram social media data. Neuropsychopharmacology. 2019;44:487–94.
Gkotsis G, Oellrich A, Velupillai S, Liakata M, Hubbard TJ, Dobson RJ, et al. Characterisation of mental health conditions in social media using Informed Deep Learning. Sci Rep. 2017;7:45141.
Cao B, Zheng L, Zhang C, Yu PS, Piscitello A, Zulueta J, et al. Deepmood: modeling mobile phone typing dynamics for mood detection. Proceedings of the 23rd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. 2017. p. 747–55.
Sathyanarayana A, Joty S, FernandezLuque L, Ofli F, Srivastava J, Elmagarmid A, et al. Sleep quality prediction from wearable data using deep learning. JMIR mHealth uHealth. 2016;4:e125.
Suhara Y, Xu Y, Pentland AS. Proceedings of the 26th International Conference on World Wide Web. International World Wide Web Conferences Steering Committee; 2017. p. 715–24 .
Dwyer DB, Falkai P, Koutsouleris N. Machine learning approaches for clinical psychology and psychiatry. Annu Rev Clin Psychol. 2018;14:91–118.
Durstewitz D. Advanced data analysis in neuroscience: integrating statistical and computational models. NY, USA: Springer; 2017.
Durstewitz D, Koppe G, Toutounji H. Computational models as statistical tools. Curr Opin Behav Sci. 2016;11:93–99.
Huys QJM, Maia TV, Frank MJ. Computational psychiatry as a bridge from neuroscience to clinical applications. Nat Neurosci. 2016;19:404–13.
Orru G, PetterssonYeo W, Marquand AF, Sartori G, Mechelli A. Using support vector machine to identify imaging biomarkers of neurological and psychiatric disease: a critical review. Neurosci Biobehav Rev. 2012;36:1140–52.
Bzdok D, Altman N, Krzywinski M. Statistics versus machine learning. Nat Methods. 2018;15:233–34.
Kim J, Calhoun VD, Shim E, Lee JH. Deep neural network with weight sparsity control and pretraining extracts hierarchical features and enhances classification performance: Evidence from wholebrain restingstate functional connectivity patterns of schizophrenia. Neuroimage. 2016;124:127–46.
Guo X, Dominick KC, Minai AA, Li H, Erickson CA, Lu LJ. Diagnosing autism spectrum disorder from brain restingstate functional connectivity patterns using a deep neural network with a novel feature selection method. Front Neurosci. 2017;11:460.
Plis SM, Amin MF, Chekroud A, Hjelm D, Damaraju E, Lee HJ, et al. Reading the (functional) writing on the (structural) wall: multimodal fusion of brain structure and function via a deep neural network based translation approach reveals novel impairments in schizophrenia. Neuroimage. 2018;181:734–47.
Goodfellow I, Bengio Y, Courville A. Deep learning. Massachusetts, USA: MIT Press; 2016.
Lee H, Grosse R, Ranganath R, Ng AY. Unsupervised learning of hierarchical representations with convolutional deep belief networks. Commun ACM. 2011;54:95–103.
Zeiler MD, Fergus R. European Conference on Computer Vision. Springer; 2014. p. 818–33.
Güçlü U, van Gerven MA. Deep neural networks reveal a gradient in the complexity of neural representations across the ventral stream. J Neurosci. 2015;35:10005–14.
Bystritsky A, Nierenberg A, Feusner J, Rabinovich M. Computational nonlinear dynamical psychiatry: a new methodological paradigm for diagnosis and course of illness. J Psychiatr Res. 2012;46:428–35.
Pascanu R, Gulcehre C, Cho K, Bengio Y. How to construct deep recurrent neural networks: Proceedings of the Second International Conference on Learning Representations (ICLR 2014). 2nd International Conference on Learning Representations, ICLR 2014. 2014.
Cho K, Van Merriënboer B, Gulcehre C, Bahdanau D, Bougares F, Schwenk H, et al. Learning Phrase Representations using RNN Encoder–Decoder for Statistical Machine Translation. Proceedings of the 2014 Conference on Empirical Methods in Natural Language Processing (EMNLP). 2014:1724–34.
Cybenko G. Approximation by superpositions of a sigmoidal function. Math Control, Signals Syst. 1989;2:303–14.
Hornik K, Stinchcombe M, White H. Multilayer feedforward networks are universal approximators. Neural Netw. 1989;2:359–66.
Funahashi KI. On the approximate realization of continuous mappings by neural networks. Neural Netw. 1989;2:183–92.
Cybenko C. Continuous Valued Neural Networks with Two Hidden Layers are Sufficient, Technical Report, Department of Computer Science, Tufts University, 1988.
Bengio Y. Learning deep architectures for AI. Found Trends® Mach Learn. 2009;2:1–127.
Montufar GF, Pascanu R, Cho K, Bengio Y. On the number of linear regions of deep neural networks. Advances in neural information processing systems. Curran Associates, Inc.; 2014. p. 2924–32.
Chatziafratis V, Nagarajan SG, Panageas I. Better depthwidth tradeoffs for neural networks through the lens of dynamical systems. arXiv preprint arXiv:200300777. 2020.
Delalleau O, Bengio Y. deep sumproduct networks. Advances in neural information processing systems. Curran Associates, Inc.; 2011. p. 666–74.
Trischler AP, D’Eleuterio GM. Synthesis of recurrent neural networks for dynamical system simulation. Neural Netw. 2016;80:67–78.
Kimura M, Nakano R. Learning dynamical systems by recurrent neural networks from orbits. Neural Netw. 1998;11:1589–99.
Koiran P, Cosnard M, Garzon M. Computability with lowdimensional dynamical systems. Theor Computer Sci. 1994;132:113–28.
Hastie T, Tibshirani R, Friedman JH. The elements of statistical learning: data mining, inference, and prediction. 2nd ed. New York: Springer; 2009.
Blei DM, Kucukelbir A, McAuliffe JD. Variational inference: a review for statisticians. J Am Stat Assoc. 2017;112:859–77.
Bzdok D, MeyerLindenberg A. Machine learning for precision psychiatry: opportunities and challenges. Biol Psychiatry: Cogn Neurosci Neuroimaging. 2018;3:223–30.
Woo CW, Chang LJ, Lindquist MA, Wager TD. Building better biomarkers: brain models in translational neuroimaging. Nat Neurosci. 2017;20:365.
Wolfers T, Buitelaar JK, Beckmann CF, Franke B, Marquand AF. From estimating activation locality to predicting disorder: a review of pattern recognition for neuroimagingbased psychiatric diagnostics. Neurosci Biobehav Rev. 2015;57:328–49.
Belkin M, Ma S, Mandal S. To Understand Deep Learning We Need to Understand Kernel Learning. International Conference on Machine Learning. 2018:541–9.
Zhang C, Bengio S, Hardt M, Recht B, Vinyals O. Understanding deep learning requires rethinking generalization. arXiv preprint arXiv:161103530. 2016.
Montanari A, Ruan F, Sohn Y, Yan J. The generalization error of maxmargin linear classifiers: Highdimensional asymptotics in the overparametrized regime. arXiv preprint arXiv:191101544. 2019.
LeCun Y, Bottou L, Bengio Y, Haffner P. Gradientbased learning applied to document recognition. Proc IEEE. 1998;86:2278–324.
Belkin M, Hsu D, Ma S, Mandal S. Reconciling modern machinelearning practice and the classical bias–variance tradeoff. Proc Natl Acad Sci. 2019;116:15849–54.
Bartlett PL, Long PM, Lugosi G, Tsigler A. Benign overfitting in linear regression. Proceedings of the National Academy of Sciences. 2020:201907378.
Geiger M, Jacot A, Spigler S, Gabriel F, Sagun L, d’Ascoli S, et al. Scaling description of generalization with number of parameters in deep learning. J Stat Mech: Theory Exp. 2020;2020:023401.
Janson L, Fithian W, Hastie TJ. Effective degrees of freedom: a flawed metaphor. Biometrika. 2015;102:479–85.
Vapnik V. The nature of statistical learning theory. New York, USA: Springer Science & Business Media; 2013.
Akaike H. Information Theory and an Extension of the Maximum Likelihood Principle. In E. Parzen, K. Tanabe, & G. Kitagawa (Eds). Selected Papers of Hirotugu Akaike. New York, NY: Springer New York; 1998 p. 199–213.
Schwarz G. Estimating the dimension of a model. Ann Stat. 1978;6:461–64.
Penny WD. Comparing dynamic causal models using AIC, BIC and free energy. NeuroImage. 2012;59:319–30.
Qi M, Zhang GP. An investigation of model selection criteria for neural network time series forecasting. Eur J Operational Res. 2001;132:666–80.
Hurvich CM, Tsai CL. Regression and time series model selection in small samples. Biometrika. 1989;76:297–307.
Kuha J. AIC and BIC: Comparisons of assumptions and performance. Sociolo Methods Res. 2004;33:188–229.
Geisser S. The predictive sample reuse method with applications. J Am Stat Assoc. 1975;70:320–28.
Varoquaux G, Raamana PR, Engemann DA, HoyosIdrobo A, Schwartz Y, Thirion B. Assessing and tuning brain decoders: crossvalidation, caveats, and guidelines. NeuroImage. 2017;145:166–79.
Stone M. Cross‐validatory choice and assessment of statistical predictions. J R Stat Soc: Ser B. 1974;36:111–33.
Cearns M, Hahn T, Baune BT. Recommendations and future directions for supervised machine learning in psychiatry. Transl Psychiatry. 2019;9:1–12.
Kaufman S, Rosset S, Perlich C, Stitelman O. Leakage in data mining: formulation, detection, and avoidance. ACM Trans Knowl Discov Data. 2012;6:1–21.
Varma S, Simon R. Bias in error estimation when using crossvalidation for model selection. BMC Bioinform. 2006;7:91.
Flint C, Cearns M, Opel N, Redlich R, Mehler D, Emden D, et al. Systematic overestimation of machine learning performance in neuroimaging studies of depression. arXiv preprint arXiv:191206686. 2019.
Neuhaus AH, Popescu FC. Sample size, model robustness, and classification accuracy in diagnostic multivariate neuroimaging analyses. Biol Psychiatry 2018;84:e81–e82.
Schnack HG, Kahn RS. Detecting neuroimaging biomarkers for psychiatric disorders: sample size matters. Front Psychiatry. 2016;7:50.
Pulini AA, Kerr WT, Loo SK, Lenartowicz A. Classification accuracy of neuroimaging biomarkers in attentiondeficit/hyperactivity disorder: effects of sample size and circular analysis. Biol Psychiatry Cogn Neurosci Neuroimaging. 2019;4:108–20.
Moyer D, Ver Steeg G, Tax CMW, Thompson PM. Scanner invariant representations for diffusion MRI harmonization. Magn Reson Med. 2020;84:2174–89. https://doi.org/10.1002/mrm.28243.
Yahata N, Morimoto J, Hashimoto R, Lisi G, Shibata K, Kawakubo Y, et al. A small number of abnormal brain connections predicts adult autism spectrum disorder. Nat Commun. 2016;7:1–12.
Rajkomar A, Oren E, Chen K, Dai AM, Hajaj N, Hardt M, et al. Scalable and accurate deep learning with electronic health records. NPJ Digital Med. 2018;1:18.
Koppe G, Guloksuz S, Reininghaus U, Durstewitz D. Recurrent neural networks in mobile sampling and intervention. Schizophrenia Bull. 2019;45:272–76.
Bergmeir C, Benítez JM. On the use of crossvalidation for time series predictor evaluation. Inf Sci. 2012;191:192–213.
Schmidt D, Koppe G, Beutelspacher M, Durstewitz D. Inferring dynamical systems with longrange dependencies through line attractor regularization. arXiv preprint arXiv:191003471. 2019.
Koppe G, Toutounji H, Kirsch P, Lis S, Durstewitz D. Identifying nonlinear dynamical systems via generative recurrent neural networks with applications to fMRI. PLoS Comput Biol. 2019;15:e1007263.
Wood SN. Statistical inference for noisy nonlinear ecological dynamic systems. Nature 2010;466:1102–04.
Draxler F, Veschgini K, Salmhofer M, Hamprecht FA. Essentially No Barriers in Neural Network Energy Landscape. International Conference on Machine Learning. 2018:1309–18.
Payan A, Montana G. Predicting Alzheimer’s disease: a neuroimaging study with 3D convolutional neural networks. arXiv preprint arXiv:150202506. 2015.
Zeng LL, Wang H, Hu P, Yang B, Pu W, Shen H, et al. Multisite diagnostic classification of schizophrenia using discriminant deep learning with functional connectivity MRI. EBioMedicine 2018;30:74–85.
Heinsfeld AS, Franco AR, Craddock RC, Buchweitz A, Meneguzzi F. Identification of autism spectrum disorder using deep learning and the ABIDE dataset. NeuroImage Clin. 2018;17:16–23.
Riaz A, Asad M, Alonso E, Slabaugh G. DeepFMRI: endtoend deep learning for functional connectivity and classification of ADHD using fMRI. J Neurosci Methods. 2020;335:108506.
Sarraf S, Tofighi G. DeepAD: Alzheimer′ s disease classification via deep convolutional neural networks using MRI and fMRI. https://www.biorxiv.org/content/10.1101/070441v4. 2016.
Zou L, Zheng J, Miao C, Mckeown MJ, Wang ZJ. 3D CNN based automatic diagnosis of attention deficit hyperactivity disorder using functional and structural MRI. IEEE Access. 2017;5:23626–36.
Suk HI, Lee SW, Shen D. Hierarchical feature representation and multimodal fusion with deep learning for AD/MCI diagnosis. NeuroImage. 2014;101:569–82.
Huang K, Wu C, Su M, Fu H. Mood detection from daily conversational speech using denoising autoencoder and LSTM. 2017 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). IEEE. 2017. p. 5125–29.
Ma X, Yang H, Chen Q, Huang D, Wang Y. Proceedings of the 6th International Workshop on Audio/Visual Emotion Challenge; 2016. p. 35–42.
Yang T, Wu C, Huang K, Su M. in 2016 10th International Symposium on Chinese Spoken Language Processing (ISCSLP). IEEE; 2016. p. 1–5.
Naderi H, Soleimani BH, Rempel S, Matwin S, Uher R. Multimodal deep learning for mental disorders prediction from audio speech samples. arXiv preprint arXiv:190901067. 2019.
Dakka J, Bashivan P, Gheiratmand M, Rish I, Jha S, Greiner R. Learning neural markers of schizophrenia disorder using recurrent neural networks. arXiv preprint arXiv:171200512. 2017.
Mao Z, Su Y, Xu G, Wang X, Huang Y, Yue W, et al. Spatiotemporal deep learning method for ADHD fMRI classification. Inf Sci. 2019;499:1–11.
Yan W, Calhoun V, Song M, Cui Y, Yan H, Liu S, et al. Discriminating schizophrenia using recurrent neural network applied on time courses of multisite FMRI data. EBioMedicine. 2019;47:543–52.
Leshno M, Lin VY, Pinkus A, Schocken S. Multilayer feedforward networks with a nonpolynomial activation function can approximate any function. Neural Netw. 1993;6:861–67.
Jarrett K, Kavukcuoglu K, Ranzato MA, LeCun Y. 2009 IEEE 12th international conference on computer vision. IEEE; 2009. p. 2146–53.
Bishop CM. Pattern recognition and machine learning. New York, USA: Springer; 2006.
Kingma D, Welling M. Autoencoding variational bayes. The 2nd International Conference on Learning Representations. 2014.
Rezende DJ, Mohamed S, Wierstra D. Stochastic Backpropagation and Approximate Inference in Deep Generative Models. International Conference on Machine Learning. 2014:1278–86.
Tibshirani R. Regression shrinkage and selection via the lasso. J R Stat Soc: Ser B. 1996;58:267–88.
Zou H, Hastie T. Regularization and variable selection via the elastic net. J R Stat Soc: Ser B. 2005;67:301–20.
Plis SM, Hjelm DR, Salakhutdinov R, Allen EA, Bockholt HJ, Long JD, et al. Deep learning for neuroimaging: a validation study. Front Neurosci. 2014;8:229.
Srivastava N, Hinton G, Krizhevsky A, Sutskever I, Salakhutdinov R. Dropout: a simple way to prevent neural networks from overfitting. J Mach Learn Res. 2014;15:1929–58.
Arjovsky M, Shah A, Bengio Y. Unitary evolution recurrent neural networks. International Conference on Machine Learning; NY, NY, USA: PMLR; 2016. p. 1120–28.
Helfrich K, Willmott D, Ye Q. Orthogonal recurrent neural networks with Scaled Cayley transform. arXiv preprint arXiv:170709520. 2017.
Henaff M, Szlam A, LeCun Y. Recurrent Orthogonal Networks and LongMemory Tasks. Proceedings of The 33rd International Conference on Machine Learning. 2016;48:2034–42.
Başar E, BaşarEroglu C, Karakaş S. Schürmann M. Gamma, alpha, delta, and theta oscillations govern cognitive processes. Int J Psychophysiol. 2001;39:241–48.
Aledavood T, Torous J, Hoyos AMT, Naslund JA, Onnela JP, Keshavan M. Smartphonebased tracking of sleep in depression, anxiety, and psychotic disorders. Curr Psychiatry Rep. 2019;21:49.
Hinton GE, Osindero S, Teh YW. A fast learning algorithm for deep belief nets. Neural Comput 2006;18:1527–54.
Kirkpatrick S, Gelatt CD, Vecchi MP. Optimization by simulated annealing. Science. 1983;220:671–80.
Aarts EHL, Korst J H M. Simulated annealing and Boltzmann machines: a stochastic approach to combinatorial optimization and neural computing. (WileyInterscience series in discrete mathematics and optimization). Chichester: Wiley; 1989.
Ye J, Kadakia N, Rozdeba P, Abarbanel H, Quinn J. Precision variational approximations in statistical data assimilation. Nonlinear Process Geophysics Discuss. 2014;1:1603–20.
Abarbanel HD, Rozdeba PJ, Shirman S. Machine learning: deepest learning as statistical data assimilation problems. Neural Comput. 2018;30:2025–55.
Bottou L. Stochastic gradient learning in neural networks. Proc NeuroNımes. 1991;91:12.
Ruder S An overview of gradient descent optimization algorithms. arXiv preprint arXiv:160904747. 2016.
Goodfellow IJ, Shlens J, Szegedy C Explaining and harnessing adversarial examples. arXiv preprint arXiv:14126572. 2014.
Szegedy C, Zaremba W, Sutskever I, Bruna J, Erhan D, Goodfellow I, et al. Intriguing properties of neural networks. arXiv preprint arXiv:13126199. 2013.
Martin AR, Daly MJ, Robinson EB, Hyman SE, Neale BM. Predicting polygenic risk of psychiatric disorders. Biol Psychiatry 2019;86:97–109.
Cole JH, Poudel RPK, Tsagkrasoulis D, Caan MWA, Steves C, Spector TD, et al. Predicting brain age with deep learning from raw imaging data results in a reliable and heritable biomarker. NeuroImage. 2017;163:115–24.
Roweis ST, Saul LK. Nonlinear dimensionality reduction by locally linear embedding. Science. 2000;290:2323–26.
Tenenbaum JB, Silva VD, Langford JC. A global geometric framework for nonlinear dimensionality reduction. Science. 2000;290:2319–23.
Maaten LVD, Hinton G. Visualizing data using tSNE. J Mach Learn Res. 2008;9:2579–605.
Ballard DH in AAAI 27984 (1987).
Rumelhart DE, Hinton GE, Williams RJ. Learning representations by backpropagating errors. Nature. 1986;323:533–36.
Gupta A, Ayhan M, Maida A in International Conference on Machine Learning; 2013. p. 987–94.
Pinaya WH, Mechelli A, Sato JR. Using deep autoencoders to identify abnormal brain structural patterns in neuropsychiatric disorders: A large‐scale multi‐sample study. Hum Brain Mapp. 2019;40:944–54.
Jaitly N, Hinton GE. Vocal tract length perturbation (VTLP) improves speech recognition. Proc ICML Workshop on Deep Learning for Audio, Speech and Language. 2013;117.
Schlüter J, Grill T. ISMIR; 2015. p. 121–26.
Park DS, Chan W, Zhang Y, Chiu CC, Zoph B, Cubuk ED, et al. Specaugment: A simple data augmentation method for automatic speech recognition. arXiv preprint arXiv:190408779. 2019.
Wang SH, Lv YD, Sui Y, Liu S, Wang SJ, Zhang YD. Alcoholism detection by data augmentation and convolutional neural network with stochastic pooling. J Med Syst. 2018;42:2.
Shin HC, Tenenholtz NA, Rogers JK, Schwarz CG, Senjem ML, Gunter JL, et al. Medical image synthesis for data augmentation and anonymization using generative adversarial networks. International workshop on simulation and synthesis in medical imaging. 2018. p. 1–11.
Nguyen KP, Fatt CC, Treacher A, Mellema C, Trivedi MH, Montillo A. Anatomicallyinformed data augmentation for functional MRI with applications to deep learning. Medical Imaging 2020: Image Processing. 2020;11313:113130T.
Goodfellow I, PougetAbadie J, Mirza M, Xu B, WardeFarley D, Ozair S, et al. Advances in neural information processing systems; 2014. p. 2672–80.
Bowles C, Chen L, Guerrero R, Bentley P, Gunn R, Hammers A, et al. Gan augmentation: Augmenting training data using generative adversarial networks. arXiv preprint arXiv:181010863. 2018.
Shang C, Palmer A, Sun J, Chen KS, Lu J, Bi J. VIGAN: Missing view imputation with generative adversarial networks. 2017 IEEE International Conference on Big Data (Big Data). 2017. p. 766–75.
Nguyen H, Morris RW,oHarris AW, Korgoankar MS, Ramos F. Correcting differences in multisite neuroimaging data using Generative Adversarial Networks. arXiv preprint arXiv:180309375. 2018.
Lee DH. The simple and efficient semisupervised learning method for deep neural networks. Workshop on challenges in representation learning, ICML. 2013:3.
Pan SJ, Yang Q. A survey on transfer learning. IEEE Trans Knowl Data Eng. 2009;22:1345–59.
Simonyan K, Zisserman A Very deep convolutional networks for largescale image recognition. arXiv preprint arXiv:14091556. 2014.
Lu S, Lu Z, Zhang YD. Pathological brain detection based on AlexNet and transfer learning. J Computat Sci. 2019;30:41–47.
Thomas AW, Müller KR, Samek W. Deep transfer learning for wholebrain FMRI analyses. OR 20 contextaware operating theaters and machine learning in clinical neuroimaging. NY, USA: Springer; 2019. p. 59–67.
Yosinski J, Clune J, Bengio Y, Lipson H. How transferable are features in deep neural networks? Advances in neural information processing systems. Montreal, Canada: MIT Press; 2014. p 3320–28.
Thodoroff P, Pineau J, Lim A. Learning robust features using deep learning for automatic seizure detection. Machine learning for healthcare conference. 2016. p. 178–90.
Hospedales T, Antoniou A, Micaelli P, Storkey. A Metalearning in neural networks: a survey. arXiv preprint arXiv:200405439. 2020.
Andrychowicz M, Denil M, Gomez S, Hoffman MW, Pfau D, Schaul T, et al. Learning to learn by gradient descent by gradient descent. Advances in neural information processing systems. Curran Associates Inc.; 2016. p. 3981–89.
Ravi S, Larochelle H. Optimization as a model for fewshot learning. 5th International Conference on Learning Representations, ICLR. OpenReview.net; 2016.
Box GE, Tiao GC. Bayesian inference in statistical analysis. Hoboken, New Jersey, USA: John Wiley & Sons; 2011.
Ghahramani Z. Bayesian nonparametrics and the probabilistic approach to modelling. Philos Trans R Soc A. 2013;371:20110553.
Obenchain R. Classical Ftests and confidence regions for ridge regression. Technometrics. 1977;19:429–39.
Quaak M, van de Mortel L, Thomas RM, van Wingen G. Deep learning applications for the classification of psychiatric disorders using neuroimaging data: systematic review and metaanalysis. https://www.medrxiv.org/content/10.1101/2020.06.12.20129130v1. 2020.
Li X, Dvornek NC, Zhuang J, Ventola P, Duncan JS. Brain biomarker interpretation in asd using deep learning and fmri. International Conference on Medical Image Computing and ComputerAssisted Intervention. 2018. p. 206–14.
Author information
Authors and Affiliations
Contributions
Substantial contributions to the conception or design of the work (GK and DD). Drafting the work or revising it critically for important intellectual content (all authors). Final approval of the version to be published (all authors). Agree to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved (all authors).
Corresponding authors
Additional information
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Koppe, G., MeyerLindenberg, A. & Durstewitz, D. Deep learning for small and big data in psychiatry. Neuropsychopharmacol. 46, 176–190 (2021). https://doi.org/10.1038/s413860200767z
Received:
Revised:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1038/s413860200767z
This article is cited by

AIDEDNet: antiinterference and detail enhancement dehazing network for realworld scenes
Frontiers of Computer Science (2023)

Creating sparser prediction models of treatment outcome in depression: a proofofconcept study using simultaneous feature selection and hyperparameter tuning
BMC Medical Informatics and Decision Making (2022)

Metamatching as a simple framework to translate phenotypic predictive models from big to small data
Nature Neuroscience (2022)

Transdiagnostic inflammatory subgroups among psychiatric disorders and their relevance to role functioning: a nested casecontrol study of the ALSPAC cohort
Translational Psychiatry (2022)

Accelerated functional brain aging in major depressive disorder: evidence from a large scale fMRI analysis of Chinese participants
Translational Psychiatry (2022)