Dr. Jing Li’s research is data fusion and statistical machine learning intersecting with systems or domains having complex data structures such as high dimensionality (e.g., 3D/4D images), multi-modality, and heterogeneity. The objectives of the methodological developments are to provide capacities for monitoring & change detection, diagnosis, and prediction & prognosis. The methodological developments emphasize the integration of domain knowledge and data-driven learning to design knowledge-infused or knowledge-constrained machine learning algorithms that are more efficient, interpretable, and generalizable. The application domains mainly include health and medicine, focusing on brain diseases and cancer, using patient-centric data analytics to support a broad spectrum spanning from basic scientific discovery (e.g., disease characterization, mechanistic understanding) to Precision Medicine empowered clinical decision-making in diagnosis, prognosis, and treatment.
Methodologies
- Knowledge-infused/constrained machine learning & deep learning
- Sparse learning
- Transfer/multitask learning & domain adaptation
- Semi-supervised learning & weakly-supervised learning
- Uncertainty quantification & active learning
Health and Medicine
- Precision medicine for brain cancer
- Early detection of Alzheimer’s Disease
- Classification and prognosis related to concussion, traumatic brain injury, migraine & post-traumatic headache
- Telemonitoring of Parkinson’s Disease
- Dental CBCT segmentation, lesion detection, and diagnosis of oral diseases
Engineering Systems
- Process data mining for manufacturing quality improvement
- Monitoring and anomaly detection in large communication networks.