Digital Twin

Many diseases afflict individuals not only with acute symptoms, but also with long-term, chronic burdens that extend far beyond single episodes. This chronicity not only impacts personal well-being, but also imposes significant economic costs on society. Examples of such chronic, episodic diseases span various medical domains, including autoimmune disorders like rheumatoid arthritis and multiple sclerosis, neurological conditions such as epilepsy and migraine headaches, and affective disorders such as major depressive disorder and bipolar disorder, characterized by recurring episodes of symptoms alternating with periods of remission.

Understanding the chronic course of these diseases requires gathering detailed information about disease trajectories. Methods such as smartphone-based assessment enable continuous monitoring of symptoms, shedding light on the nuanced patterns of symptom recurrence and remission.

Casting the evolution of symptoms as a dynamical system, in this project, we derive a quantitative framework for symptom dynamics - with and without intervention. In collaboration with Prof. Hamidreza Jamalabadi , we show in a large group of MDD patients that 1) we can predict future symptom change in individual patients from the dynamical system’s theoretical resistance to reach a symptom-free state and 2) that this relationship strengthens with genetic risk of MDD and childhood maltreatment while it weakens with resilience. Also, we show that Control Theoretic models of outperform classic Machine Learning approaches in predicting therapeutic response to Electroconvulsive Therapy.

Our results support the notion that the evolution of depressive symptoms and the effects of interventions can be understood analogous to a physical dynamical system governed by the interaction between symptoms and configured by genetic risk and environmental factors. This opens the door to a mechanistic understanding of individual symptom development over time and makes testable predictions for theoretically optimal therapeutic interventions.

More generally, we explore the methodologies employed across clinical and computational research domains, highlighting the significance of gathering longitudinal data and applying complex systems approaches to model disease course over time. Through interdisciplinary collaborations, we aim to elucidate the intricate dynamics of chronic illnesses and develop more effective strategies for intervention and management.