Sraavya Sambara

Welcome to my site! I recently graduated from Harvard University, where I studied Computer Science and History & Science, and graduated magna cum laude, Phi Beta Kappa. My interests lie broadly in advancing machine learning methods for healthcare, with a particular focus on developing multimodal AI systems that emulate clinical diagnosis and developing methods to enhance safe patient–AI interactions. Check out my research below.

Research & Publications

• RadFlag: A Black-Box Hallucination Detection Method for Medical Vision Language Models
  ML4H @ NeurIPS 2024
  Proposes a sampling-based, black-box method to detect hallucinations in medical vision–language models using entailment-based uncertainty estimation and conformal prediction.
• 3DReasonKnee: Advancing Grounded Reasoning in Medical Vision Language Models
  Pacific Symposium on Biocomputing 2026 (Oral Presentation)
  Introduces the first large-scale benchmark for grounded diagnostic reasoning over 3D MRI volumes, with region-specific supervision and evaluation metrics for localization and reasoning accuracy.
• Evaluation of Large Language Models as Emergency Department Revisit Predictors
  Pacific Symposium on Biocomputing 2026 (Oral Presentation)
  Studies the use of large language models for predicting emergency department revisits and demonstrates performance gains using retrieval-augmented and embedding-based approaches.
• Understanding the Use of Mobility Data in Disasters: Exploratory Qualitative Study of COVID-19 User Feedback
  JMIR 2024
  Examines public perceptions and concerns around the use of human mobility data during the COVID-19 pandemic, with implications for privacy-aware disaster response.
• MedRedFlag: Investigating How LLMs Redirect Misconceptions in Real-World Health Communication
  Submitted to ACL 2026
  Analyzes how clinicians redirect unsafe or false patient assumptions in practice and evaluates whether large language models appropriately correct or instead accommodate these misconceptions.

Public Outreach & Leadership

I am passionate about public outreach and fostering critical conversations about healthcare and AI. I have served as a Harvard Student Leader in Artificial Intelligence and a United Nations Millennium Fellow, working at the intersection of technology, policy, and social impact. I was also the President of the Harvard Undergraduate Global Health Forum, where I led programming and discussions on global health equity and emerging technologies, and a Managing Editor of the Harvard Political Review, where I edited long-form writing on science, policy, and ethics.

In my free time, I enjoy singing Carnatic music, playing tennis, and running.