Psychiatry research in the COVID-19 era and beyond: A role for mathematical models

A recent editorial discussed the implications of COVID-19 for psychiatry research (Galletly, 2020) and concluded that researchers must find new ways of working while social distancing remains in force. We argue that the need for innovation in research methods extends beyond the period of the pandemic and propose and invite enhanced collaboration between psychiatrists and mathematical modellers. Mathematical modelling research does not require human subjects, and as such is of particular use during times of social distancing, but also deserves broader attention beyond this. Collaborations between clinicians and mathematical modellers have been important in areas such as public health and immunology (Eftimie et al., 2016; Keeling and Rohani, 2011). Now would be an opportune time to open up new collaborations between mathematical modellers and mental health researchers.

Mathematical modelling is most useful when used to appraise existing health data or when used to explore questions where the collection of clinical data is not possible. The value of mathematical modelling is enhanced by its adaptability – models can be developed to explore clinical problems at multiple levels of complexity (Keeling and Rohani, 2011). Models at the level of individual behaviours, organs or cells have been well explored under the banner of ‘computational psychiatry’ (Anticevic and Murray, 2017), but population-level models have seen most use in epidemiology, where they have been important in understanding the transmission and management of infectious diseases, for example, by vaccination (Keeling and Rohani, 2011). In the COVID-19 pandemic, such models have guided public policy on social distancing and have been important in understanding the likely extent, severity and consequences of the pandemic for health systems (Moss et al., 2020) and the requirements of candidate vaccines. While all of these model types may be useful in mental health research in the COVID-19 era and beyond, the potential benefits of population-level modelling are most noteworthy. Population-level modelling methods are well established for communicable diseases and their sequelae (Keeling and Rohani, 2011), in comparison with non-communicable diseases, which is one factor that has hampered the translation of mathematical modelling work into policy and practice in mental health thus far.

Key questions that should be explored in future research using population-level models include estimating how the pandemic will impact the incidence of conditions such as post-traumatic stress disorder (PTSD) and depression (Galletly, 2020) and to plan the profession’s response to this. While models at the individual or cellular level might yield valuable insights into potential neuropathology associated with COVID-19, these results should ultimately be linked to population-level modelling to estimate the long-term psychiatric morbidity arising from the pandemic. Therefore, we propose that population-level mathematical modelling should be a priority in psychiatric research during COVID-19, to address these and other problems. Such modelling work will have implications beyond COVID-19, as the methods developed will have application to other public health emergencies resulting in psychiatric sequelae, such as natural disasters and armed conflicts.