AMIA 2025: Clinical AI & Data Science
Check out my latest post on clinical AI and data science at AMIA 2025.
View on LinkedIn
About
Advancing data-informed care
I design and deploy AI systems that are grounded in clinical workflows and co-developed with care teams. My Ph.D. in Computational Medicine centered on early warning systems for critical events in intensive care units. Today, I apply those lessons as a Data Scientist and Postdoctoral Fellow at Duke, focusing on perioperative medicine and Generative AI for Medicine.
My work combines statistical rigor, prospective clinical evaluation, and multidisciplinary collaboration. I enjoy partnering with clinicians, informatics teams, and industry collaborators to translate AI research into sustainable delivery models that improve outcomes and equity.
Featured
Latest LinkedIn Post
Focus Areas
Where clinical expertise meets machine intelligence
I build systems that are technically sound, interpretable, and ready for deployment in high-stakes settings.
Trustworthy Clinical AI
Design LLM-powered agents and RAG pipelines that surface context-rich insights in perioperative care, balancing performance with safety and guardrails.
Predictive Monitoring
Develop time-series and multimodal models that predicts hemodynamic shock, hypothermia, and respiratory events hours ahead of deterioration.
Clinical Impact
Lead prospective validations and randomized trials that demonstrate measurable improvements in value-based care and clinical decision-making.
Recent Highlights
Selected milestones and collaborations
-
2025
Randomized controlled trial of a transthoracic echocardiography decision tool published in the Journal of the American Society of Echocardiography.
-
2025
Contextual phenotyping of pediatric sepsis cohorts using LLMs selected for an oral presentation at AMIA 2025.
-
2024
Launched the Pediatric Sepsis Data Challenge leaderboard and open-sourced evaluation tooling for global teams.
-
2024
Open-sourced code and benchmarks for the Pediatric Sepsis Challenge to accelerate reproducible clinical AI research.
Research Spotlights
Impactful collaborations across the translational pipeline
Agentic Hybrid RAG
Multi-agent retrieval augmented generation framework for scientific literature review, future work will combine structured EHR data with LLM reasoning to support perioperative care teams.
Prospective Shock Prediction
Prospectively validated prediction of hemodynamic shock in resource-limited pediatric ICUs.
Thermal Imaging for Shock
Early detection of hemodynamic shock using thermal videos and deep computer vision models to enable critical care monitoring.
Digital Health Equity
Evaluated how electronic health literacy and social determinants of health are connected.
Randomized Trial of a TTE Value Tool
Demonstrated how decision support improved value-based imaging in a multicenter randomized controlled trial. Journal of the American Society of Echocardiography, 2025.
Contextual Phenotyping with LLMs
Introduced an LLM-driven pipeline for phenotyping pediatric sepsis cohorts using chart context and registry data. AMIA Annual Symposium, 2025.
Early Warning for Hemodynamic Shock
Leveraging the potential of thermal imaging and machine learning to predict hemodynamic shock in critically ill patients. Scientific Reports, 2019.
Digital Health Literacy & SDOH
Assessment of how social determinants influence digital health literacy among perioperative populations. Preventive Medicine Reports, 2024.
Hypothermia Prediction with Physiological Vitals Time-Series Data
Combined physiologic vitals time-series including temperature time-series signals with machine learning to predict hypothermia. Frontiers in Physiology, 2022.
Predicting Childhood Asthma Exacerbations
Leveraged longitudinal EHR and lung impulse oscillometry data to prognosticate the risk of acute asthma attack events in pediatric populations. European Respiratory Journal suppl_64, 2020.
Environmental Drivers of Respiratory Events
Quantified the impact of ambient air pollution on pediatric emergency visits for acute respiratory illness. Environmental Science and Pollution Research, 2021.
Prospective Validation of Shock Prediction
Demonstrated real-world applicability of an early warning system for hemodynamic shock in collaboration with critical care teams. SSRN Working Paper, 2023.
Media & Outreach
Sharing work across clinical and public audiences
Connect
Let’s build the next generation of clinical AI
Collaborations & Speaking
I enjoy partnering on clinical deployments, translational research, and mentorship for emerging data scientists.