AI Powered Remote Sleep Fragmentation and Functional Brain States Analysis for the Monitoring of Treatment Effectiveness in Major Depressive Disorders

• Applicants: Professor Dr. Thomas Frodl; Professorin Dr. Ute Habel
• Subject Area: Biological Psychiatry ; Clinical Psychiatry, Psychotherapy, Child and Adolescent Psychiatry ; Human Cognitive and Systems Neuroscience
• Term: since 2024
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 538749936

Depression is among the most frequent psychiatric disorders. Due to its relative frequency, complications, comorbidity, and consequences, depression has an eminent clinical, health economic and political relevance. As it is associated with a high rate of chronic illness courses and therapy non-responders, the search for biomarkers for prediction of therapy response is of high priority. Sleep disturbances are, although unspecific, a hallmark symptom of depression and patients with sleep disorders have a higher risk for depression. Building on this evidence, we aim to use the sleep patterns for monitoring and predicting therapy response in depression. In more detail, clinical improvement will lead to improvement of sleep fragmentation, normalization of neuroimaging data (amygdala, hyperreactivity and hyperconnectivity) and improvement in cognitive functions. Sleep fragmentation and neuroimaging markers will predict treatment response earlier than clinical symptoms. Combining interdisciplinary expertise in clinical, neuroimaging and artificial intelligence (AI) we join forces to reveal the potential of a new physiological sleep fragmentation assessment (SleepAI). SleepAI consists of a wrist worn pulse oximeter continuously streaming raw physiological data that are subsequently analyzed by the deep learning SleepPPG-Net algorithm. SleepPPG-Net estimates sleep stages and enables to compute sleep fragmentation biomarkers. In a longitudinal, observational and prospective study design we monitor sleep patterns before and several weeks after multidimensional treatment in depressed patients and use additional neuropsychological, neuroimaging and psychopathological markers to validate the therapy effects and reveal the association of sleep fragmentation biomarkers with cognitive, brain functional and psychopathological measurements over a period of 42 days. We expect that sleep fragmentation biomarkers are characterizing different patient biotypes, hence correlate with cognitive dysfunctions, neuroimaging markers and specific psychopathological symptom profiles that show predictive potential for therapy response. The application of SleepAI may hence be a useful clinical and easily applicable tool assisting the therapeutic approach and efficiency.

• DFG Programme: Research Grants
• International Connection: Israel
• International Co-Applicant: Professor Joachim Behar, Ph.D.