Approved Research

Deep learning based on data integration identifies potential therapeutic targets in neurodegenerative disease

Lay summary

For hard-to-treat neurodegenerative diseases, accurate classification and risk prediction will no doubt benefit diagnosis and improve patient life quality. Despite considerable efforts made, it remains difficult to effectively integrate different types of patient data available. To begin to solve this challenge, we plan to develop a deep learning-based, risk-associated prediction framework on neurodegenerative disease patients using coupled genomics and health-related data that include mutation profiles and MR imaging. We expect that our framework will achieve promising predictive performance in distinguishing high-risk versus low-risk patients, even with incomplete data available. Our study here will have major clinical implications as it may yield novel biomarkers or drug targets for a new generation of therapeutics in neurodegenerative disease.