National and international health authorities have brought to the world’s attention the need to mitigate threats to mental health posed by the COVID-19 pandemic and the economic crises that ensued in its wake. This recognition comes at a pivotal moment for the field of mental health epidemiology. There is growing acknowledgement that the mental health (along with what may be called mental wealth) of nations relies on dynamic interrelationships among physical, social, economic and health systems, giving rise to a complex web of variables that challenge traditional analytic methods.

The pandemic thus represented a catalyzing event in which the infectious disease research community exemplified the role of valued scientists in the battle against a public health threat. Researchers rapidly provided decision makers with systems models for a range of possible response options prior to implementing strategies to control transmission. In several countries, these decision support tools guided effective responses to COVID-19 despite imperfect knowledge and uncertainties that characterize any evolving crisis.

Until recently, mental health epidemiology has been dominated by backward-focused examinations of past events with little capability to forecast future trends and estimate the impacts of proposed actions. Over the past 12 months, however, a series of regional, state and national systems models of the social and economic impacts of COVID-19 on mental health were developed in Australia to simulate five- and 10-year trajectories of psychological distress, the use of mental health services, waiting times and suicidal behavior. Similar efforts have been launched in Colombia, the U.K., the U.S. and elsewhere.

The Australian models provided decision makers with estimates of the potential impacts of mitigation strategies, simulated individually and in combination. The models simulated a range of social protection measures, such as employment, education and income support programs, as well as mental health and suicide prevention programs. They also encompassed awareness campaigns, service capacity expansion and care following suicide attempts. These simulations helped identify where the most timely and effective investments lie. The work informed public discourse and prompted substantial additional investment in mental health and suicide prevention by the Australian government: $2.3 billion in Australian dollars ($1.72 billion in U.S. dollars) over five years.

Systems modeling techniques are also being used to support design of flexible clinical trials and program evaluation. As part of the HOPE trial in Victoria, Australia, systems modeling is informing the design of a new program to support young people who have sought emergency care after attempting self-harm. The modeling allows researchers to simulate alternative scenarios involving the inclusion of different combinations of program components such as safety planning, the engagement of peer workers and various configurations of preventive outreach efforts. The model also allows the testing of variations in key program parameters such as the minimum duration, scale and uptake of the program. These considerations provide vital insights into the best program design ahead of any large-scale implementation.

Development in two key directions would further advance the emerging field of computational mental health epidemiology. The first is adequate and timely data—particularly changes over time in the prevalence of psychological distress, in levels of mental health service use and in the outcomes experienced by people receiving such services. Improving the quality and accessibility of data could provide feedback for improving both systems models and real-world care. Mobile phones and other digital data-collection platforms that combine sensor data and microsurveys of people in local communities are being used in several countries to secure more detailed information over time on factors that influence mental health outcomes—including sleep, physical activity, social connection, psychological distress, and alcohol and drug misuse. Such platforms could be deployed in “sentinel” surveillance sites across key communities for real-time monitoring of self-harm and other mental health outcomes. The stream of information from these data sources, coupled with machine-learning methods and systems models, could support the creation of continuously updated decision support systems that facilitate more rapid responses to community mental health crises.

In addition, ongoing advances in accessible software interfaces allow for nonexperts to contribute to model design and testing without having to be literate in computer code. Recent work using a transparent and inclusive approach to develop systems models designed to support regional-level mental health service planning and suicide prevention in Australia has helped to counter misunderstandings and distrust of modeling and garner community support for strategic actions to improve mental health.

Systems models constructed without input can be prone to idealized and invalid representations of complex, real-world systems. Additionally, a “backroom” approach to modeling can result in policy and planning decisions that lack popular support or ignore important features of the context in which a prediction applies. In contrast, when mental health systems modeling, and the learning it empowers, is achieved collectively, it has the potential to enhance regional self-determination in securing improvements to mental health services, to build trust among decision makers and communities, and to help catalyze lasting change.

If you or someone you know is struggling or having thoughts of suicide, help is available. Call the National Suicide Prevention Lifeline at 1-800-273-8255 (TALK), use the online Lifeline Chat or contact the Crisis Text Line by texting TALK to 741741.

This is an opinion and analysis article, and the views expressed by the author or authors are not necessarily those of Scientific American.