This research aims to identify biomarkers from multi-modal data sources, including genetic information, neuroimaging, cognitive assessments, lifelog data, and histories of substance use, such as alcohol or drugs. By analyzing these diverse data types, the study seeks to uncover biological, cognitive, and behavioral indicators associated with mental health conditions, including the effects of addiction on mental well-being.
It also explores how deep learning and AI techniques can process multi-modal data to enhance classification models for diagnosing mental health disorders. The focus is on improving model accuracy and identifying patterns across complex data sources, including the impact of external environmental factors on mental and physical health.
The integration of multi-modal data is critical for developing explainable AI models that provide transparent and interpretable insights for mental health diagnosis and prediction. Identifying effective methods for data fusion and feature selection is essential to ensure these models are both accurate and understandable.
Finally, this research examines the validation and optimization of AI models for use in digital healthcare devices, enabling real-time diagnosis and prediction of mental health conditions. The study also investigates how external environmental factors, such as stressors and lifestyle influences, contribute to mental health outcomes, ensuring these models are robust, reliable, and applicable across diverse populations.
