Back
on
by

Digital Twins in Healthcare: Examples, Uses, and Why Biosensors Are the Missing Link

Digital twins in healthcare
Digital twins in healthcare explained: the definition, the main types, real examples from cardiac devices to diabetes care and clinical trials, benefits and challenges, and why continuous biosensor data keeps a twin accurate.

A digital twin in healthcare is a virtual model of a patient, organ, or care process that continuously updates with data from its real-world counterpart and uses it to predict outcomes and guide decisions. Unlike a one-time simulation, a digital twin keeps learning: new data, such as glucose readings from a continuous glucose monitor (CGM), flows into the model, and the model’s predictions feed back into care. Examples already in clinical research include virtual hearts used to test cardiac devices, patient-specific twins that tune automated insulin delivery, and digital twins of control patients in clinical trials.

The hard part is usually not the model. It is the data. Most healthcare digital twins run on episodic records, such as lab results and clinic visits, which cannot show how a body changes day to day. Biosensors provide continuous, real-world data that keeps a twin up to date. This guide explains what digital twins in healthcare are, the main types, real examples, the benefits and challenges, and why biosensor data infrastructure decides whether a twin works.

What is a digital twin in healthcare?

The National Academies of Sciences, Engineering, and Medicine defines a digital twin as “a set of virtual information constructs that mimics the structure, context, and behavior” of a system, which “is dynamically updated with data from its physical twin, has a predictive capability, and informs decisions that realize value.” The National Academies report (released December 2023, published 2024) adds that “the bidirectional interaction between the virtual and the physical is central to the digital twin.”

FDA uses the same idea. In its discussion paper on AI in drug development, FDA describes a fully developed digital twin as “a physical component … a virtual component, and automated data communications between the two” (FDA).

Applied to a patient, a 2024 scoping review in npj Digital Medicine describes a digital twin as “a virtual representation of a person which allows dynamic simulation of potential treatment strategy” (Katsoulakis et al.).

How digital twins differ from traditional models

Traditional model
Digital twin
Data
Historical, often population averages
Individual data, updated continuously
Updates
Built once, revised occasionally
Updated as new data arrives
Direction
One-way: data in, prediction out
Two-way: predictions inform care, and outcomes update the model
Output
General risk estimates
Individual predictions and “what if” scenarios

Types of digital twins in healthcare

Type
What it models
Example uses
Organ twin
A single organ, such as the heart
Testing cardiac devices, planning procedures
Patient twin
One person’s physiology across systems
Personalizing treatment, tuning insulin delivery
Process or system twin
A care process, hospital unit, or supply chain
Planning capacity, staffing, and patient flow
Population or trial twin
Groups of patients or trial participants
Predicting control outcomes in clinical trials, modeling disease spread

Examples of digital twins in healthcare

Virtual hearts for medical device testing

Dassault Systèmes launched the Living Heart Project in 2014, bringing together medical experts and engineers to build a realistic simulated human heart. Later that year it signed a five-year research agreement with FDA to use the model for testing pacemaker leads and other cardiovascular devices. A 2019 extension included “an in silico clinical trial aimed to reduce animal testing or the number of patients required,” and in 2024 the company published the ENRICHMENT Playbook, a guide to using virtual twins in clinical trials (Dassault Systèmes).

Siemens Healthineers has also built research cardiac twins, including a pilot with Heidelberg University Hospital to predict which patients benefit from cardiac resynchronization therapy. The company has said that clinical trials “must precede any routine treatment” (Siemens Healthineers).

Patient twins for type 1 diabetes

In a randomized trial published in npj Digital Medicine in 2025, University of Virginia researchers mapped each participant with type 1 diabetes to a cloud-based digital twin built from their automated insulin delivery data. The twin re-tuned each person’s artificial pancreas about every two weeks. Among 72 participants who completed the 6-month trial, time in range rose from 72% to 77%, and A1C fell from 6.8% to 6.6% (Kovatchev et al.; UVA).

Whole-body twins for type 2 diabetes

Twin Health builds a “Whole Body Digital Twin” for type 2 diabetes from CGM data, activity sensors, smart scales, blood pressure readings, and meal logs. In a 12-month randomized trial at Cleveland Clinic published in NEJM Catalyst, 71% of participants using the program reached an A1C below 6.5% without glucose-lowering drugs other than metformin, compared with 2.4% in usual care (Cleveland Clinic).

Digital twins in clinical trials

Unlearn uses machine learning on historical data to create digital twins of trial participants that predict how each person would fare on the control treatment. In September 2022, the European Medicines Agency issued a qualification opinion for Unlearn’s PROCOVA method, which uses these predictions to make trials more efficient (Unlearn).

Benefits of digital twins in healthcare

  • Personalized treatment. Twins can test treatment options on a virtual patient before trying them on a real one.
  • Earlier warning. Continuously updated twins can flag changes before symptoms appear.
  • Safer, faster device development. Virtual organs can reduce animal testing and the number of patients needed for some studies.
  • More efficient clinical trials. Trial twins can reduce the number of participants needed in control groups.
  • Better operations. Process twins help hospitals plan capacity and patient flow.

The market is growing fast from a small base: MarketsandMarkets estimated digital twins in healthcare at $1.6 billion in 2023 and projected $21.1 billion by 2028 (MarketsandMarkets).

Challenges and disadvantages of digital twins in healthcare

Most of the hard problems come from data, not algorithms.

  • Static data. Lab results, imaging, and clinic notes are snapshots. A twin built only from them cannot follow day-to-day changes in a patient.
  • Population averages. Models trained on averaged or synthetic data hide individual variation, which is the point of a twin.
  • No feedback loop. Many so-called twins are one-way models. Without a stream of new data, they drift away from the real patient.
  • Validation. Twins need ongoing validation, traceability, and performance monitoring, which most projects lack.
  • Regulation. A twin that recommends treatment is likely to be regulated as medical device software, with the evidence that requires.
  • Privacy and security. Twins combine sensitive data from many sources, which raises privacy and cybersecurity requirements.
  • Cost and complexity. Building and maintaining the data pipelines, models, and validation is a long-term investment.

Why biosensors are the missing link

A digital twin is only as current as its data. Biosensors close the gap between the virtual model and the real patient.

  • Continuous data. Biosensors such as CGMs measure every few minutes, around the clock, so a twin can follow real physiology instead of guessing between clinic visits. See what biosensors can measure.
  • Real-world context. Wearables capture how patients respond to food, activity, sleep, and medication in daily life.
  • The feedback loop. Sensor data shows whether a recommended change worked, which is how a twin learns.

Both diabetes examples above depend on continuous sensor data: the UVA twins on automated insulin delivery data, and Twin Health on CGM and other connected devices.

Infrastructure matters more than the sensor

Raw sensor data is not ready for a model. Teams need infrastructure that:

  • Ingests data from many devices and normalizes formats, units, and timestamps
  • Detects and flags gaps, noise, and sensor errors
  • Keeps a traceable record of where each data point came from and how it was processed
  • Scales from a pilot to thousands of patients without rebuilding
  • Delivers data and outputs to apps, clinicians, and health records securely

See our guides to biosensor data integration and CGM data integration.

How digital twins are regulated

The FDA has no single rule for digital twins. How a twin is regulated depends on what it is used for:

The future of digital twins in healthcare

  • From monitoring to prediction. Twins will increasingly forecast changes before they become clinically apparent.
  • From prediction to prescription. Twins will simulate treatment options and help personalize dosing and care plans, as the diabetes trials already show.
  • Beyond individual patients. Aggregated twins will model programs, devices, and care pathways.
  • More sensor data. As over-the-counter and long-wear biosensors spread, more people will generate the continuous data that twins need.

The organizations that succeed will invest in validated, scalable data infrastructure first.

Frequently asked questions

What is a digital twin in healthcare?

A digital twin in healthcare is a virtual model of a patient, organ, or care process that is continuously updated with data from its real-world counterpart. It predicts outcomes and informs decisions, and the results feed back into the model. The two-way link between the virtual and the physical is what separates a twin from a static model.

What are examples of digital twins in healthcare?

Examples include Dassault Systèmes’ Living Heart model, used with FDA to test cardiovascular devices; patient twins that tune automated insulin delivery in type 1 diabetes; Twin Health’s whole-body twin for type 2 diabetes, built from CGM and other sensor data; and Unlearn’s digital twins of clinical trial participants.

What types of digital twins are used in healthcare?

The main types are organ twins (e.g., virtual hearts); patient twins, which model one person’s physiology; process or system twins, which model hospital operations and care processes; and population or trial twins, which model groups of patients or clinical trial participants.

What are the disadvantages of digital twins in healthcare?

The main challenges are static or averaged data, missing feedback loops, limited validation, regulatory requirements for twins that recommend treatment, privacy and security risks, and the cost of building and maintaining data pipelines and models.

How do biosensors support digital twins?

Biosensors such as CGMs and wearables provide continuous, real-world data that keeps a digital twin up to date between clinic visits. They also show whether a recommended change worked, which is the feedback a twin needs to learn and stay accurate.

Are digital twins regulated by the FDA?

It depends on their use. FDA has guidance on using computational models as evidence in device submissions, and a twin that analyzes patient data to recommend treatment is likely to be regulated as medical device software. The FDA has no single rule for digital twins.

Build the data foundation for your digital twin

Digital twins are only as good as the data behind them. Sequenex builds the biosensor data infrastructure that twins depend on: device integration, secure cloud platforms, data pipelines, and clinician and patient apps, on its NEX platform. Contact us or see our biosensor and CGM software development services.

Want to schedule a demo of NEX?

Contact us
SaMD and Connected Devices Software Experts
© 2026 Sequenex. All rights reserved.