In recent years, digital phenotyping and mobile sensing technologies have advanced, allowing the study of the digital traces that people leave through their interactions with smartphones and other devices connected to the internet-of-things. Digital phenotyping and mobile sensing provide insights into the mental health of populations by allowing the evaluation of patterns of technology use, including screen time, app usage, and GPS data (people carry their phones everywhere)1. These technological tools constitute promising ways of providing policymakers with real-time insights, with the COVID-19 pandemic offering a powerful example2. COVID-19 mitigation policies enacted during the height of the pandemic — including social distancing and severe lockdowns — were among the measures that affected the mental health and wellbeing of populations3. But in most cases these effects were known only retrospectively, limiting policymakers’ access to data that could inform cost–benefit analyses and allow nimble policymaking, enabling them to weigh the burden of mental ill health arising from the implemented policies against the costs of contracting COVID-19 (ref. 4).

How else can the study of digital footprints supplement policymaking? Mounting evidence suggests that when, for instance, a person is living with depression, they withdraw socially, and that this is mirrored objectively by changes in GPS signal activity5. This can also manifest in the more-frequent use of negative words in digital writing6, and behaviors on phone calls might provide insights into when a person is in the prodromal stage of depression, thereby predicting the onset of major depression7. Although much of the research predicting mental-health states from digital footprints is still at the feasibility stage, meaningful associations between mental health and digital data — such as from smartphone log data — have been observed in several samples8.

How can digital phenotyping and mobile sensing inform policymaking to advance the global mental health (GMH) agenda? In short: digital assessment approaches could become more accurate and objective. At present, the field relies largely on self-report scales to assess mental disorders; such scales can lack cultural relevance or validity, increasing information bias in psychiatric epidemiology and limiting work among at-risk populations, including migrants. Digital approaches are durable cost- and labor-saving solutions. In low- and middle-income countries, specialist mental-health providers are rare, and although lay healthworkers have been trained to assess mental disorders, their precarious employment often limits their long-term work engagement. Digital approaches can advance prevention. Given the promise of early detection, these approaches can improve efforts to prevent the onset of common mental disorders — a crucial yet elusive goal of GMH. Digital approaches can also aid in tailored mental-health treatments. The WHO advises an optimal mix of services for mental health, which includes intervention at the community to individual levels, with increasing degrees of intensity. Moreover, the assessment of treatment effectiveness can be aided by continued digital surveillance, and booster sessions can be better timed when applied to maintain treatment gains. Finally, the use of digital phenotyping and mobile sensing can also more objectively evaluate mental-health promotion, including leveraging the built environment in shaping GMH.

Given the potential invasiveness of collecting such data, ethical issues should be considered and incorporated in these approaches9. For example, the privacy of beneficiaries must be protected and rely on best practices, including privacy-by-design principles. In a real-time panel one might, for instance, study only the distance traveled by a person, and not specific travel routes. Aggregated-use statistics regarding the frequency of negative or positive words and the number of people being contacted could be collected (instead of raw data), ensuring anonymity.

Aside from this, questions arise around who will have access to the data. This provides opportunities for government–academic partnerships, wherein trained scientists who can read and interpret such data would translate the information for policymakers. Furthermore, as the use of digital phenotyping expands, regulatory control will be needed to clearly define for what kind of purposes the data can be used. For example, commercial applications to enhance targeted marketing and to drive user engagement can undermine efforts to improve the mental health of populations.

Notwithstanding from these key ethical issues, which clearly need to be tackled, we believe that digital phenotyping and mobile sensing technologies can be leveraged to improve GMH. It would be useful to establish representative samples around the world, with participants agreeing to their phone behaviors and GPS signals being tracked in real time. Data from these real-time panels could be combined with insights from ecological experience sampling, for example from prompts regarding current wellbeing or daily stress levels. Such data could also be used to understand how climate change and the built environment shape GMH by tracking access to green-space10 and urbanicity variables. These real-time data could provide relevant information about the wellbeing of large populations, and can be brought together with decisions by policymakers, for example during pandemics or other crises, including disasters and humanitarian emergencies. We believe that such data are also relevant in non-crisis periods, to gain insights into GMH in different regions and at different points in time.