Precision Subclassification of Mental Health in Diabetes: Digital Twins for Precision Mental Health to Track Subgroups

Mental conditions and disorders (e.g. distress, depressive, anxiety, and eating disorders) are more prevalent in people with diabetes (PWD) and associated with reduced quality of life and impaired glycaemic outcomes. Evidence supports a complex network between psychosocial factors and glycaemic control that can be highly variable between persons. It is assumed that subgroups exist that show different trajectories of glycaemia and mental health. Belonging to a particular subgroup may be linked with a higher risk of developing mental health problems compared to others. This suggests that it is possible to treat individuals in different subgroups in a manner that optimizes their treatment and can improve health outcomes. Accurate characterisation can inform more individualized care. This calls for a more personalised approach considering the idiosyncrasies of different subgroups. Over 3 years, the investigators have established the basis of a precision mental health approach for diabetes using n-of-1 analyses. By utilizing combined ecological momentary assessment (EMA: repeated daily sampling of psychosocial factors in everyday life) and continuous glucose monitoring (CGM), intensive longitudinal data per person could be collected. This enables the analysis of individual associations between glycaemic parameters and psychosocial variables and identification of individual sources of diabetes distress in each person. The objective of the present study is to use of the n-of-1 approach to identify subgroups of PWD who share common characteristics in the associations between glucose and psychosocial variables. The identified subgroups shall be used to develop a digital twin for precision mental health in diabetes. The digital twin serves as representation of a real person, allowing to make simulations and predictions of the course of mental health and glycaemia. These predictions can inform diabetes care and lead to more precise, personalised treatment decisions. To achieve this, a longitudinal panel including over 1,400 PWD who continuously complete EMA and questionnaire surveys and measure glucose levels using CGM was developed. Over 1000 clinical interviews to diagnose mental disorders have been conducted to identify major mental health conditions and map mental outcomes. To identify subgroups and develop the digital twin, the sampling will be expanded aiming at a total of 1,809 PWD. Incidence and remission of mental disorders will be determined via repeated interviews. The complex networks between clinical, metabolic, and psychosocial data will be analysed using machine learning, leading to new insights with the potential to shape future guidelines. These results will be used by the digital twin to predict courses of glycaemic control and mental health, translating the individual evidence into direct treatment suggestions.