Taposh Dutta Roy

    Advancing the Science

    274 citations · h-index 8 · Verified at hms.harvard.edu

    Current Projects

    Ethics & AI

    Autonomy Index

    Autonomy Index

    IRB Approved · Forthcoming, 2026

    Evaluating LLM readiness for clinical decision-making. Co-developed with Dr. Rebecca Weintraub Brendel at Harvard Medical School. 9 LLMs, 7,524 scored cases, four-domain framework.

    IRB ApprovedHarvard Medical SchoolClinical AILLMs
    BAIA Framework

    BAIA Framework

    Published, 2025

    Bioethics Artificial Intelligence Advisory — an agentic AI system for clinical ethical decision support. Peer-reviewed and published in Cureus, indexed in PubMed.

    BioethicsAgentic AIPubMed

    Digital Twin

    HospiTwin

    HospiTwin

    USPTO Patents Filed

    Hospital Digital Twin — agent-based simulation for optimizing emergency department bed turnaround. Won 1st Place at MIT Hacking Medicine 2025 for Best Use of Generative AI in Healthcare.

    Digital TwinGenAIMIT
    Developing a Digital Twin

    A deep dive into building a digital twin ecosystem — modeling real-world hospital operations with agent-based simulation.

    Digital TwinSimulationTutorial

    Validating digital twin process flows in healthcare using machine learning to understand bottlenecks and areas of improvement.

    Digital TwinMachine LearningIEEE

    USPTO 19/247,983

    Hospital Digital Twin — Agent-Based Simulation

    USPTO 63/804,393

    Optimizing Dynamic Simulations of Hospital Scenarios

    Hospital Operations

    An iPhone app developed in-house under my technical leadership for hospital operations. Deployed across 40+ hospitals and used in command centers and crisis situations — including the LA wildfires — to give leaders real-time visibility into patient flow, staffing, and capacity. Recognized by CIO.com as a flagship example of streamlining healthcare operations through analytics.

    iPhone AppHospital OperationsCommand CenterCrisis Response

    Quantum and Advanced Computing

    Silicon Valley Quantum & Advanced Computing

    A 1,430+ member community with 40+ expert talks and curated learning paths for the quantum era. Training programs for students and professionals in partnership with IEEE.

    CommunityIEEEEducation

    Investigating adversarial robustness of hybrid quantum-classical models for medical image classification.

    Quantum MLAdversarial AIAIP

    Hybrid quantum computing approaches for pneumonia detection from chest radiographs.

    Quantum MLMedical ImagingIEEE

    Enhancing gastrointestinal care through AI-enabled classification of endoscopic lesions.

    Quantum MLGI CareIEEE

    Past Projects

    Diabetic Retinopathy

    Diabetic Retinopathy

    Completed

    AI-powered detection and classification of diabetic retinopathy from retinal fundus images, enabling early screening and intervention.

    Medical ImagingDeep LearningHealthcare

    KDD Cup 2019

    Completed, 2019

    Chaired the ACM SIGKDD Data Science Competition with 2,800+ teams from 39 countries and 17,000+ submissions.

    ACM SIGKDDCompetitionData Science