Digital twins in healthcare

What if clinicians could test a treatment on a virtual patient before treating the real one? Digital twins offer the promise of making this a reality across healthcare, allowing clinicians to simulate, compare and optimise interventions using patient-specific virtual models.

What is a healthcare digital twin?

A digital twin is a virtual representation of a physical object, system or process. In healthcare, the physical counterpart may be the patient, their organ or anatomical structure, or a physiological process for that patient.  

A healthcare digital twin may combine anatomical information from medical imaging, physiological measurements from sensors and diagnostic equipment, patient characteristics, medical history and lifestyle data, mathematical models, machine-learning predictions, and data generated as the patient responds to treatment.

The twin is patient-specific, dynamic and capable of being updated as new observations become available. As well as representing the patient’s present condition, it can simulate possible future states and compare alternative interventions¹.

Examples of digital twins in healthcare:

1. Ophthalmology: virtually testing treatment before surgery

The eye is highly measurable and its optical behaviour is governed, at least in part, by well-understood physical principles. Instruments can measure the parameters of a patient’s eye to construct a patient-specific optical model.

In cataract and refractive-lens surgery, such a model could compare intraocular lenses before one is implanted. A surgeon might simulate a monofocal, toric, extended-depth-of-focus or trifocal lens and examine predicted refraction, depth of focus, retinal image quality, contrast performance or optical disturbances. In laser refractive surgery, a model of the cornea and optical system can be used to assess candidate ablation profiles and predict changes in corneal shape and optical aberrations.  

In principle, a virtual optical model could allow comparison between laser correction, refractive-lens exchange and different IOL designs. Simulated vision (for example showing the patient what driving at night, reading or using a computer might look like after treatment) may improve patient decision making in selecting treatment and help manage patient expectations.  

In commercial practice Alcon's wavelight® plus platform for laser vision correction uses the SightMap diagnostic platform to collect detailed patient-specific measurements, including corneal tomography, ocular wavefront data and axial-length biometry, to create a virtual "Digital Eye Twin" of the individual's eye. The system then applies ray-tracing algorithms to simulate how light propagates through that specific eye and virtually refines the treatment profile before surgery. The digital twin is further used to account for expected biomechanical changes and epithelial remodelling following treatment. By enabling patient-specific simulation prior to intervention, wavelight® plus demonstrates how digital twins can move beyond diagnosis and prediction to directly influence therapeutic decision-making and clinical outcomes.

2. Diabetes: from physiological model to personalised intervention

In diabetes, the twin is not principally an anatomical model. It can instead be a dynamic representation of the patient’s metabolism, integrating physiological and behavioural data to understand how an individual responds to factors such as food, activity, sleep and stress.

In commercial practice Twin Health's AI Digital Twin™ creates a personalised, real-time model of an individual's metabolism using data from sensors and other smart devices. The model learns how factors such as food, activity, sleep and stress affect that individual's metabolic health and uses those insights to provide personalised recommendations. As new data are collected, the digital representation can adapt to changes in the individual, illustrating a key feature of digital twin technology: a continuing connection between the physical patient and their evolving virtual counterpart.

3. Cardiovascular medicine: a virtual heart for planning treatment

Cardiovascular medicine is among the more mature fields for patient-specific computational modelling. Cardiac twins may combine CT- or MRI-derived anatomy with ECG data, electrical-conduction models, tissue characteristics, pressure measurements and blood-flow simulations.  

For example, in structural-heart procedures, models can simulate valve implantation, device positioning and resulting changes in blood flow. In electrophysiology, a virtual heart may help identify tissue responsible for an arrhythmia and simulate candidate ablation strategies..  

In commercial practice Medtronic’s partnership with DASI Simulations is a digital-twin example in structural heart care: the companies are integrating DASI’s patient-specific predictive modelling into the TAVR workflow so that clinicians can simulate alternative transcatheter heart-valve deployment scenarios using patient CT anatomy. In electrophysiology, inHEART markets an AI-enabled digital twin of the heart that converts CT or MR images into interactive three-dimensional cardiac models for ablation planning and integration with electroanatomical mapping systems.

Protecting healthcare digital twins with European patents – showing a technical contribution

Digital-twin inventions will usually be treated as computer-implemented inventions containing a mixture of technical and non-technical features. Under the European Patent Office’s established approach, mathematical or algorithmic features can support inventive step where they contribute to the technical character of the invention.

The value of a digital-twin lies in how it enables clinicians and medical devices to do things differently. For European patent protection, the same point is crucial. For patentability, it is not the abstract digital-twin which is interesting, but the use of that digital twin to solve a technical problem that produces a technically meaningful result.

References

1. Vallée A. “Digital twin for healthcare systems.” Frontiers in Digital Health. 2023;5:1253050. doi:10.3389/fdgth.2023.1253050.