Jehan Patel

Background

About Jehan Patel

As a research associate at the Michigan Medicine Biomedical & Clinical Informatics Lab (BCIL), I used generative AI for developing image-to-text based models to automate the creation of colonoscopy reports and chest X-ray reports. My project also used transfer learning and self supervision to identify thoracic diseases in pediatric chest X-rays. I also implemented and optimized BCIL’s proprietary Tropical Geometry Fuzzy Neural Net machine learning algorithm to assess a patient’s risk of heart failure and rank their eligibility for advanced therapies. Previously, I worked at BCIL as part of a UROP project for the MRADS program, where I developed ways to improve the processing of large data sets for MCI and Alzheimer’s disease using novel tensor-based methods (Canonical Polyadic Decomposition) incorporated into Machine Learning algorithms. In addition to my multidisciplinary engineering coursework at UMich, I’ve completed Google’s prestigious ‘Data Analytics Certificate’ course and Oracle’s ‘OCI Generative AI Professional Certificate’, and was selected for the NSA-funded GenCyber Warrior Academy summer program, hosted by the Pennington Military Leadership Center at the University of North Georgia. I’ve taken several LLM classes and have researched a variety of models, including transformer-based architectures like GPT-3 and Llama 2. I’m proficient in a wide range of programming languages including C/C++, Java, Python, Matlab, Julia, SQL, HTML, JavaScript, ARM Assembly, and Verilog.