research

My research focuses on clinical NLP and computational approaches to healthcare, with particular interest in speech and language technologies for neurocognitive disorder detection.

Conversational Behavior Modeling

Berkeley Speech Group | ICML 2026 (Submitted)
with Prof. Gopala Anumanchipalli

We introduce a framework for modeling human conversation as multi-level perception, reasoning over conversational behaviors via a Graph-of-Thoughts (GoT). The approach formalizes the intent-to-action pathway with a hierarchical labeling scheme, predicting high-level communicative intents and low-level speech acts to learn their causal and temporal dependencies. The GoT framework structures streaming predictions as an evolving graph, enabling a transformer to forecast the next speech act and generate concise justifications for its decisions.

[arXiv]

Speech Biomarkers for Neurocognitive Disorders

Berkeley Speech Group | Interspeech 2026 (Under Review)
with Prof. Gopala Anumanchipalli

Developing automated speech-based methods for detecting and classifying neurocognitive disorders from naturalistic speech samples. Details withheld for double-blind review.

Sex Differences in Multimorbidity Burden

Computational Precision Health (UC Berkeley & UCSF) | JAMA (Submitted)
Advisors: Prof. Irene Y. Chen, Prof. Yulin Hswen

Analyzing multimorbidity patterns across 344,038 adults from the NIH All of Us Research Program. Our work reveals that women experience consistently higher morbidity burden across the lifespan, with the largest disparities in midlife (ages 40–59). This research has implications for understanding health equity and developing targeted interventions.

Computational Precision Health (UC Berkeley & UCSF)
Advisors: Prof. Irene Y. Chen, Prof. Yulin Hswen

Characterizing multimorbidity accumulation patterns around menopause onset using EHR data from 203,247 adults. This work examines race-stratified trajectories to understand health disparities in midlife women.

Past Research

Berkeley School of Law | 2023–2024
Advisor: Prof. Tejas N. Narechania

Built an interactive NLP platform for comparing Supreme Court majority and dissent opinion drafts using semantic similarity and token-level analysis. Presented at the UC Berkeley Data Science Discovery Symposium (2024).