Zhen He group

We develop deep learning based methods that unlock the diagnostic potential of high-resolution digital pathology. Our research focuses on the automated identification and spatial characterization of novel biomarkers within the complex tumor microenvironment. By leveraging state-of-the-art computer vision models, we aim to transform traditional histopathology into a high-throughput, reproducible science that can identify subtle morphological patterns and molecular indicators that are hard to detect by pathologists.

Our primary aim is to establish a robust "digital biology" pipeline that transitions from in silico prediction to validated biological leads. In collaboration with our colleagues in LIMS, we develop scalable AI systems that can screen tens of thousands of potential candidates to identify those with the highest therapeutic potential. Through the use of automated screening and precision molecular modeling, we strive to create targeted therapies that are tailor-made for specific disease profiles, ultimately advancing the frontier of precision medicine.

We apply advanced multimodal deep learning to quantify the prevalence and nature of alcohol-related content within digital and physical environments. Our research involves the development of automated systems capable of detecting alcohol consumption, branding, and promotional materials across diverse data streams, including social media imagery, video, and text. By utilizing state-of-the-art object detection and natural language processing, we provide public health researchers with the high-throughput tools necessary to monitor marketing trends and assess their impact on population-level health behaviors.