Digital Twins in Health Care: Clinical, Ethical and Legal Perspectives

Digital Twins have advanced various fields in engineering for decades and are now entering clinical care, posing several critical challenges.

A digital twin, built of data from a patient and the environment, aims to mimic the structure and behavior of the physical patient, is dynamically updated with new data, and seeks to make predictions about the patient and inform decisions.  But how should such digital entities be regulated, and how do and will they affect clinicians, clinical decisions, and patients?

This multi-disciplinary workshop brings together leading experts – data scientists, clinicians, ethicists, legal scholars and others – to explore these complex and growing cutting edge clinical, ethical and legal questions.

This event is part of the Frontiers in Data Science and AI initiative at the Data Science Institute, Columbia University, and is co-sponsored by the Columbia University Masters of Bioethics Program and the Northeast Big Data Innovation Hub.

---

Event Recording

The recording will be published soon!

---

Event Details & Agenda

Monday, April 27, 2026 (9:00 AM – 5:00 PM ET)
In-Person & Zoom Option

Location: Columbia School of Social Work – Room C05
Address: 1255 Amsterdam Ave, New York, NY 10027 – Map

· · ─ · ─ · ·

9:00 AM – 9:10 AM: Welcome Remarks (10 min)

• Robert Klitzman, Professor of Psychiatry (in Sociomedical Sciences), Columbia University Irving Medical Center; Program Director, Bioethics Program, School of Professional Studies, Columbia University

9:10 AM – 9:30 AM: Introduction: What are digital twins and why are they important? (20 min)

• Florence Hudson, Executive Director, Northeast Big Data Innovation Hub

9:30 AM – 10:45 AM: Current Clinical Uses I: How are DTs now being used in clinical care, and what clinical, provider, patient, regulatory, ethical or legal challenges arise? (75 min)

• Moderator: Florence Hudson, Executive Director, Northeast Big Data Innovation Hub (in-person)
• Dan Isaacs, GM & CTO of the Digital Twin Consortium, Chief Strategy Officer, Object Management Group (in-person)
• Eric Stahlberg, Executive Administrative Director for the Institute for Data Science in Oncology at MD Anderson Cancer Center (in-person)
• Grace Peng, Director of Mathematical Modeling, Simulation and Analysis, National Institute of Biomedical Imaging and Bioengineering (NIBIB) within the National Institutes of Health (virtual)
• Chakra Chennubhotla, Associate Professor of Computational and Systems Biology, University of Pittsburgh, Chief Executive Officer, PredxBio, Inc. (virtual)

10:45 AM – 11:00 AM:  Break

11:00 AM – 12:30 PM: Clinical Uses II and Clinician Perspectives (90 min)

• Moderator: Charles E. Binkley, Director of AI Ethics and Quality, Hackensack Meridian Health; Associate Professor of Surgery, Hackensack Meridian School of Medicine; and Lecturer in Bioethics, School of Professional Studies, Columbia University (in-person)
• Nancy Lee, MD, Radiation Oncologist and Early Drug Development Specialist, Memorial Sloan Kettering Cancer Center (virtual)
• Kristin Myers, Associate Professor of Mechanical Engineering, Columbia Engineering (in-person)
• Martin Genet, Professor, Mechanics Department & Solid Mechanics Laboratory, École Polytechnique (in-person)
• Florence Doo, Assistant Professor, Diagnostic Radiology & Nuclear Medicine, University of Maryland School of Medicine; Director of Innovation, University of Maryland Medical Intelligent Imaging (UM2ii) Center; and Faculty, University of Maryland-Institute for Health Computing (UM-IHC) (virtual)
• Caroline Chung, Vice President and Chief Data & Analytics Officer and Co-Director of the Institute for Data Science in Oncology at MD Anderson Cancer Center (virtual)

12:30 PM – 1:30 PM: Lunch and Discussion (1 hour)

1:30 PM – 3:30 PM: Other Stakeholder Perspectives (2 hours)

• Moderator: Robert Klitzman, Professor of Psychiatry (in Sociomedical Sciences), Columbia University Irving Medical Center; Program Director, Bioethics Program, School of Professional Studies, Columbia University (in-person)
• Orlando Lopez, Director of the NIDCR Dental, Oral, and Craniofacial (DOC) Biomaterials, Devices, and Clinical Technologies Program, Division of Extramural Research (virtual)
• Matthias Braun, Director of the Department of Social Ethics, University of Bonn (virtual)
• Nancy Berlinger, Senior Research Scholar, The Hastings Center for Bioethics (in-person)
• Glenn Cohen, James A. Atwood and Leslie Williams Professor of Law; Deputy Dean of Harvard Law School; Faculty Director of the Petrie-Flom Center for Health Law Policy, Biotechnology & Bioethics at Harvard (in-person)
• Jodyn Platt, Associate Professor of Learning Health Sciences, Medical School Associate Professor of Health Management and Policy, School of Public Health, University of Michigan Medical School (in-person)
• Sean Whiteley, Founder, Axomem Private Limited (virtual)

3:30 PM – 3:45 PM:  Break

3:45 PM – 5:00 PM: Open Discussion, Future Directions, and Next Steps (1 hour, 15 min)

5:00 PM: Event Ends

---

Speaker Details

Listed in order of program:

Talk Title: Building Communities for Medical Digital Twins

Abstract: Medical digital twins hold tremendous potential to realize the promise of precision medicine and precision health. The successful adoption of medical digital twin approaches is a community effort, spanning a full range of stakeholders with the individual at the center. presentation will share insights and perspectives in efforts building informed medical digital twin communities, identifying critical issues across these communities, and discuss potential paths forward. 

Talk Title: Biomedical Digital Twins – the Ethics of Twinning

Abstract: Worldwide, there is a growing realization that the future of personalized health will depend on digital twins – a real-time pairing of physical and virtual systems that continuously adapt (with bi-directional feedback) to inform human decisions (see NASEM definition). The National Institutes of Health (NIH) is encouraging the community to develop digital twin technologies, using the NASEM definition, to ensure governance in NIH research processes and to enhance public trust in scientific discoveries and clinical solutions. This presentation will discuss the challenges associated with designing, developing and using biomedical digital twins (BDTs). The presenter will briefly share tools developed by the Interagency Modeling and Analysis Group (IMAG), Multiscale Modeling Consortium and the NIH Bridge to Artificial Intelligence (Bridge2AI) program for the dynamic assessment of ethics associated with legal and regulatory considerations. The idea of “twinning” will be discussed as the critical mindset required to iteratively embed values into our BDT designs and contextual use cases for health.

Talk Title: Towards the Use of Digital Twin in the Treatment of Cancer Using HPV+ Oropharyngeal Carcinoma as a Model

Abstract: Standard treatment for HPV-associated oropharyngeal carcinoma (HPV+ OPC) carries significant long-term toxicity despite excellent cure rates, driving the need for personalized approaches. Over the past decade, we have developed a hypoxia-guided radiation de-escalation strategy using intratreatment FMISO PET, demonstrating that 75% of patients can safely receive 30Gy — markedly reducing toxicity while preserving 97% five-year overall survival across 430 patients in prospective phase II trials. However, this approach remains binary, assigning patients to either 30Gy or 70Gy without capturing the full spectrum of biologic heterogeneity. This lecture will also explore the next step: the use of Digital Twin technology to further individualize treatment in HPV+ OPC. By integrating multi-scale tumor modeling, real-time imaging, multiomics data, and patient-reported outcomes into a dynamic simulation framework, Digital Twins hold promise for continuously adaptive, patient-specific treatment optimization. While we are not yet there, HPV+ OPC with its favorable biology and rich prospective datasets represents an ideal model to develop and validate this approach, which ultimately is applicable to all cancers.

Talk Title: Toward Pulmonary Mechanical Digital Twins

Abstract: The development of “digital twins” in healthcare offers significant opportunities to understand and treat complex conditions like Idiopathic Pulmonary Fibrosis (IPF), a progressive disease that severely alters lung structure and mechanics. This presentation outlines our framework for building pulmonary mechanical digital twins by bridging routine clinical imaging, such as 3D CT scans, with physics-based biomechanical modeling. Through a dedicated personalization pipeline, we tailor these virtual models to reflect an individual patient’s unique lung, mapping both shape, healthy and diseased regions, motion, etc., to non-invasively estimate potential biomarkers—like regional tissue compliance and localized physical stresses—that cannot be directly measured in the clinic. By simulating this complex mechanical environment, we aim to investigate whether physical stresses actively drive IPF progression and, ultimately, provide clinicians with powerful new virtual tools for earlier diagnosis, accurate prognoses, and highly personalized respiratory care.

Talk Title: Should You Believe What You See? Medical Imaging and the Reality of Digital Twins.

Abstract: A CT scan today is increasingly two things at once: a diagnostic image for the patient in front of us, and a data substrate for AI (and digital twins). Many of today’s “digital twins” are built from medical imaging, updated by medical imaging, or evaluated against it. Which means that a conversation about digital twins in medicine, should also discuss this interface with medical imaging. We will discuss where imaging-based digital twins actually sit on the spectrum from research prototype to regulated clinical tool, and what the field means when it calls a model a “twin” in the first place. Is a “twin” any computational model that simulates, predicts, or stands in for patient biology – or is it something stricter, a continuously updated, bidirectional model of an individual patient as the consensus definition proposes?  The talk will walk through cases to look at the ethical, legal, and governance questions that radiologists are seeing from the reading room, for open discussion on what the future looks like for digital twins.

Talk Title: Building Human-Centered Digital Twins: Integrating Prediction, Interaction and VVUQ for Precision Oncology

Abstract: Digital twins are redefining healthcare from reactive delivery to presenting observations to informed decisions under quantified uncertainty with continuously updated virtual simulations of individual tumors, patients or even populations. The National Academies of Sciences, Engineering, and Medicine describe digital twins as dynamic virtual constructs that mirror and evolve with their physical counterparts to support data‑driven decisions. Clinical research pursuits of this vision are advancing through patient‑specific tumor twins, adaptive radiation therapy twins, and personalized systemic therapy twins that leverage multimodal data and mechanistic and AI models to simulate outcomes to guide treatment.

A central challenge is ensuring effective human–digital twin interaction, enabling clinicians to interrogate predictions, understand model behavior, and integrate insights into real‑world workflows. Achieving this requires robust research and development in timely multimodal data integration, intuitive interfaces, and rigorous verification, validation, and uncertainty quantification (VVUQ). Embedding VVUQ strengthens trust, clarifies model limits, and supports safer, more transparent decision‑making. Together, these efforts position digital twins as powerful tools for advancing precision oncology through proactive, personalized, and clinically interpretable insights.

Talk Title: Beyond the Model: What It Takes to Implement Biomedical Digital Twins Across the Health Care Ecosystem

Abstract: Biomedical digital twins are increasingly recognized for their potential to advance prediction, personalization, and decision support in health care, yet many remain limited to prototype or pilot-stage use. A major barrier is not only technical performance, but whether these systems are implementable, governable, interoperable, and sustainable in real care environments. This presentation introduces a fit-for-implementation perspective that extends beyond fit-for-purpose evaluation by addressing the practical conditions required for adoption: Decision-grade evidence, workflow fit, human factors, interoperability, governance, accountability, reimbursement alignment, and lifecycle readiness. Regulatory intelligence is emphasized as core pillar for aligning development with evidence expectations, risk management, and translational decision-making. Using the dental, oral, and craniofacial innovation ecosystem as a representative biomedical testbed, the presentation highlights how fragmented data, complex care workflows, and translational barriers can inform broader implementation strategies for trustworthy, scalable digital twins across the health care ecosystem.

Talk Title: On the Ethics of Human Digital Twins in Medicine and Health

Abstract: Processes of digitization are reshaping the concept of bodies and thus also of concepts to protect them – as bodily integrity. Against the example of digital twins this talk is addressing the increasing entanglement between human bodies and digital technologies. It argues that digital representations of the body are not merely external tools but are closely connected to, and can actively influence, the bodies they simulate. In this context, bodily integrity must be reconsidered as encompassing both risks of violation and potentials for enhancement through digital systems.

Talk Title: Who Wants a Twin?: Individual Patients, Digital Twins, and Health Disparities

Abstract: This talk will consider digital twins as a clinical tool that is also a consumer good, positioned as something healthy patients will want, and as a potential health equity tool that relies on a patient’s interest in engaging with their own health data and its predictive value.

Talk Title: Deploying Digital Twins: Impacts on Accountability and Patient Autonomy 

Abstract: Digital twins promise to leverage complex data analytics to innovate patient care well beyond current capabilities. As a dynamic technology, ethical evaluation and governance systems will also need to adapt. Applying ecological models of evaluation and emerging lessons-learned from the impact of AI can inform approaches to development and deployment of digital twins. Drawing on research with patients about the use of AI in their care, critical ethical issues will be explored. Examining the issue of autonomy provides an example of how systems will need to change to incorporate digital twins and related technologies in ways that ensure trustworthiness and benefit to people and institutions.

Talk Title: From Building to Bedside: Digital Twins, Data Fusion, and the Hard Questions

Abstract: Hospital digital twins are increasingly discussed as tools for individual patient monitoring, but infectious disease surveillance demands something fundamentally different: a twin that models both the patient and the population simultaneously. One patient’s multi-drug-resistant organism colonisation affects every other patient on the ward, every staff member, every visitor, and potentially the broader community. This dual nature — patient-specific and group-impacting — creates ethical and legal tensions that single-patient digital twin frameworks were never designed to address.

Drawing on five years of building operational digital twins for hospital infection surveillance in Singapore, this talk presents three concrete projects spanning large-scale pathogen surveillance, AI risk prediction with anonymised training data, and edge video capture for ward monitoring. Each project surfaces distinct challenges at the intersection of clinical benefit, data protection, and public health obligation. The surveillance system fuses lab results, patient movements, spatial context, and temporal patterns across entire hospital campuses, while edge video analytics using vision-language models raises the sharpest questions about the boundary between routine monitoring and evidence with entirely different legal obligations.

The talk confronts four unresolved challenges: the free-rider problem in public health data, where individuals withdraw consent but continue to benefit from collective surveillance; the tension between pandemic preparedness and purpose-limitation principles; the regulatory ambiguity of when a digital twin becomes a medical device; and the problem of inadvertent data classification escalation, where operational data unexpectedly acquires legal or ethical obligations it was never designed to carry. A central question is the “peacetime-to-wartime” transition — at what national alert level does the ethical calculus shift from protecting individual anonymity to identifying and tracing for community safety