Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain
the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in
Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles
and JavaScript.
Singh et al. perform a breath-metabolomics study on patients with epilepsy taking antiseizure medications. They find that systemic valproic acid concentrations, along with risk estimates for drug responses and side effects, can be predicted by measuring metabolites in the breath, which might help to guide therapeutic dosing and manage side effects.
Berg et al. establish a panel of patient-derived endometrial cancer organoids and xenograft models. They show that their models recapitulate the genetic profile of the donor tumor and can be used for drug testing and development of a prognostic gene signature.
Mascheroni et al. develop a method for individual clinical predictions by combining mathematical modelling and machine learning in a Bayesian framework (BaM3). By using both synthetic and real clinical datasets, they show the potential of the method to predict tumour growth in the context of clinical data sparsity and limited knowledge of disease mechanisms.
De Salazar et al. quantify the impact of BNT162b2 mRNA vaccination on COVID-19 transmission and deaths in residents of long-term care facilities in Catalonia, Spain using statistical modelling. They find that high vaccination coverage results in a substantial reduction in transmission amongst residents, preventing around 3 in 4 documented infections and COVID-19-related deaths.
Gamble and Jaroensri et al. develop deep learning systems to predict breast cancer biomarker status using H&E images. Their models enable slide-level and patch-level predictions for ER, PR and HER2, with interpretability analyses highlighting specific histological features associated with these markers.
Wagner et al. carry out a longitudinal seroepidemiological study of SARS-CoV-2 antibodies in a cohort of adults from a large company in Vienna, Austria. In individuals positive for S1-reactive antibodies at baseline, RBD-specific antibodies are most likely to persist for six months and correlate most closely with SARS-CoV-2 neutralizing ability.
Mu et al. utilize a deep learning natural language processing model as part of an active learning approach to extract diagnostically relevant semantic information from bone marrow pathology synopses. Their findings demonstrate the potential for artificial intelligence in assisting clinicians in assessing, cataloging and triaging medical text datasets such as pathology synopses.
Jun et al. evaluate sex-stratified clinical outcomes in two cohorts of patients hospitalized with COVID-19 in New York. While male sex risk is a risk factor for poor outcome in both cohorts – one from earlier and one from later on in the pandemic – some of the sex-specific risk factors observed initially are not observed later on.
MacLeod et al. evaluate the mechanical safety of 3D-printed personalised high tibial osteotomy (HTO) plates in an in silico clinical trial. Using this novel methodology, they find no increased risk of mechanical failure for personalised devices compared to conventional plates, supporting further studies to assess clinical outcomes in patients treated with personalised HTO.
Knabl et al. perform a seroepidemiological study in the Austrian ski resort Ischgl, where a super-spreader event lead to a SARS-CoV-2 outbreak. Through mathematical modelling, they find that the subsequent decline in viral transmission was most likely a combined effect of high seropositivity and the implementation of non-pharmaceutical interventions.
Pereda-Loth et al. study the dynamics of SARS-CoV-2 in France in the Autumn of 2020. They find that the government-implemented “state of emergency” and curfew measures were the initial triggers of viral suppression, rather than the more restrictive lockdown that followed.
Wulczyn et al. utilise a deep learning-based Gleason grading model to predict prostate cancer-specific mortality in a retrospective cohort of radical prostatectomy patients. Their model enables improved risk stratification compared to pathologists’ grading and demonstrates the potential for computational pathology in the management of prostate cancer.