ID
63
Pathway
Clinical, Community, and Translational Research
Description
ChatGPT is just the beginning. A multitude of large language models are rapidly emerging, each with unique capabilities. In this project, we are exploring how these models perform on tasks such as summarizing clinical information, analyzing sentiment, and extracting structured insights from unstructured data. Currently, we are looking at how AI models analyze patient sentiment to understand how accurate these models are in recognizing patient emotions, which is especially needed as AI becomes more prevalent in medicine, such as in scribing physician-patient encounters
Goals
1) Create a protocol used to evaluate AI models and compare their responses to human raters
2) As human raters, analyze emotional sentiments of patients' posts on reddit
3) compare AI ratings of patient sentiment to human ratings to understand how AI models compare to humans in recognizing sentiment in a medical context
2) As human raters, analyze emotional sentiments of patients' posts on reddit
3) compare AI ratings of patient sentiment to human ratings to understand how AI models compare to humans in recognizing sentiment in a medical context
Mentor
Sudipta Tripathi and Rajesh Anumolu
Mentor E-mail
Sudipta.Tripathi@umassmed.edu Rajesh.Anumolu@umassmed.edu
Department
Renal and Transplant Medicine
Prior Students
Isha Gupta
Rithvik Sandiri
Shyam Rana
Rithvik Sandiri
Shyam Rana
Availability
Yes