Research Area
Clinical Decision Support
Developing AI-assisted tools that complement human expertise in clinical settings, with rigorous evaluation of real-world impact.
Overview
As AI systems are increasingly deployed in clinical settings, rigorous evaluation methods are essential to ensure these tools genuinely improve patient outcomes. Our research focuses on developing and testing AI-assisted clinical decision support systems that complement—rather than replace—human expertise.
Key areas of investigation include:
- Clinician-in-the-loop AI: Understanding how clinicians interact with AI recommendations and designing interfaces that support appropriate trust calibration
- Soft ground truth methods: Developing statistical approaches for evaluating AI when gold-standard labels are uncertain
- Implementation science: Studying how workflow integration, training, and population factors affect AI effectiveness
- Equity evaluation: Ensuring clinical AI performs equitably across demographic groups
Active Projects
Evaluating AI-Assisted Clinical Decision Support
Rigorous experimental methods for understanding how AI recommendations affect clinician decision-making, accuracy, and behavior in clinical risk assessment.
AI for Primary Care Diagnostics
Developing and evaluating generative AI systems to improve diagnostic accuracy in resource-limited primary care settings, using reinforcement learning with expert feedback.
AIM-AHEAD
Advancing clinical decision support that complements human expertise through the NIH AIM-AHEAD consortium
Clinical AI Evaluation Methods
Developing rigorous methodologies to assess AI-assisted clinical decision support systems, from algorithm performance to real-world clinical impact.
Related Publications
View all publicationsEffect of an mHealth-Supported Healthy Future Programme to Improve Type 2 Diabetes Management in Nairobi, Kenya: A Cluster Randomised Controlled Trial
Chen, Huanhuan, Ndegwa, Stephen, Kwaro, Daniel, Otieno, Walter, Oyugi, Elizabeth, Sylvia, Sean (2023)
The Lancet Global Health
Dynamic Information Sub-Selection for Decision Support
Huang, Jingdong, Galal, Galal, Anderson, Erik, Chiang, Sharon, Goldstein, Benjamin, Marks, Michael, Sylvia, Sean (2024)
Proceedings of Machine Learning Research
Collective Intelligence-Based Participatory Surveillance for Infectious Disease: Mixed Methods Pilot Study in Ghana
Marley, Gifty, Dako-Gyeke, Phyllis, Nepal, Prajwol, Rajgopal, Rohini, Koko, Evelyn, Chen, Elizabeth, Nuamah, Kwabena, Osei, Kingsley, Hofkirchner, Hubertus, Marks, Michael, Tucker, Joseph D., Eggo, Rosalind, Ampofo, William, Sylvia, Sean (2024)
JMIR Infodemiology
DOI: 10.2196/50125Quality of Telemedicine Consultations for Sexually Transmitted Infections in China: A Standardized Patient Study
Si, Lei, Xue, Hao, Tucker, Joseph D., Sylvia, Sean (2024)
PLOS Medicine
Tuberculosis Detection and the Challenges of Integrated Care in Rural China: A Cross-Sectional Standardized Patient Study
Sylvia, Sean, Xue, Hao, Zhou, Chengchao, Shi, Yaojiang, Yi, Hongmei, Zhou, Huan, Rozelle, Scott, Pai, Madhukar, Das, Jishnu (2017)
PLOS Medicine
DOI: 10.1371/journal.pmed.1002405Artificial Intelligence in Health Care
Sylvia, Sean, Oliva, Junier (2024)
North Carolina Medical Journal
DOI: 10.18043/001c.120561Digital Approaches to Enhancing Community Engagement in Clinical Trials
Tan, Rayner K. J., Tang, Weiming, Tucker, Joseph D., Sylvia, Sean (2022)
Contemporary Clinical Trials
DOI: 10.1016/j.cct.2022.106758Addressing 6 Challenges in Generative AI for Digital Health: A Scoping Review
Templin, Tara, Perez, Monika W., Sylvia, Sean, Leek, Jeff, Sinnott-Armstrong, Nasa (2024)
PLOS Digital Health
DOI: 10.1371/journal.pdig.0000503Tools & Outputs
Collaborate With Us
We're always looking for research partners, health system collaborators, and clinical sites to advance our work in clinical AI evaluation.
Get in Touch