Investigator: Vaibhav Rajan.
The increasing availability of digitized clinical data presents an unprecedented opportunity to study and gain deeper understanding of diseases, develop new treatments and improve healthcare ecosystems, and to ultimately reduce the growing worldwide clinical and economic burden of healthcare.
Broadly, our aim is to develop accurate and scalable algorithms that can effectively model health information of individuals at different levels – genomic, physiological and social – and from disparate sources. We develop data mining and machine learning models for clinical data analysis that can improve clinical decision support systems and hospital workflows. We are particularly interested in unsupervised learning and modeling challenges arising due to heterogeneity, high-dimensionality and temporality of clinical data.
Models and Algorithms
- Deep Multi-View Learning from Heterogeneous Data Sources
- Combined Knowledge-based and Data-driven Models
- Unsupervised Learning from High-Dimensional Data
- Models for Genomic Data Analysis
Applications
- Phenotype and Drug Discovery
- Adverse Drug Event Detection
- Clinical Complication Prediction
- Personalized Treatment Planning