My work focuses on developing statistical and machine learning methods to realize data-driven personalized/individualized decision-making.
The exciting part about this is that by efficiently using data, we can accelerate scientific discoveries and improve personalized diagnoses and treatments. This not only enhances our ability to tailor interventions to individual patients but also has the potential to optimize health care outcomes at scale by identifying patterns and insights that might otherwise remain hidden.
Recently, my student and I developed a new framework for AI-powered medical decision-making. We have shown that by using our proposed framework, AI tools can significantly reduce the required sample sizes in chart reviews — an essential step to ensure reliable model predictions. This advancement not only streamlines the data collection process but also accelerates the validation of models, ultimately enabling faster and more accurate deployment of AI in clinical settings.
I am working on multiple proposals to advance the field of personalized medical decision-making. I am excited about extending our framework to other fields, including causal inference. Expanding into this area offers the opportunity to uncover deeper insights into cause-and-effect relationships, which can further refine personalized treatment strategies and enhance the overall impact of data-driven health care solutions.
I am particularly intrigued by using large language models to incorporate electronic health records data into the traditional medical research pipeline, which typically includes patient recruiting and data adjudication. Leveraging these models has the potential to automate and enhance these processes, making it easier to identify eligible patients, extract relevant information and ensure data accuracy — ultimately accelerating research timelines and improving the quality of medical studies.