Trustworthiness in Medical Product Question Answering by Large Language Models
Date:
I gave an invited talk at the Machine Learning for Healthcare Roundtable during the Amazon Machine Learning Conference 2025, presenting our work on evaluating the trustworthiness of large language models (LLMs) in medical product question answering.
Drawing from our evaluation framework for generative shopping assistants, the talk introduced a systematic method for assessing whether LLM-generated answers about medical products adhere to FDA-approved labeling, thereby mitigating the risks of hallucinations, misinformation, and off-label promotion.
