Can Digital Twins Improve Patient Care?

The technology, which has been successfully used in other industries, is slowly emerging as an innovative way to “clone” organ systems and genomic data to personalize patient care.

By John Halamka, M.D., president, Mayo Clinic Platform, and Paul Cerrato, senior research analyst and communications specialist, Mayo Clinic Platform

If you’re a science fiction fan, the term digital twin might conjure up images of humans in incubators being cloned like in the movie Matrix. But the reality is far more intriguing. In plain English, a digital twin “is a virtual model designed to accurately reflect a physical object.” A digital twin of a wind turbine, for example, would receive data from a turbine’s sensors that measure energy output, temperature, and other parameters. The digital twin would then run simulations and suggest ways to improve the performance of the physical turbine. Several health care innovators are starting to apply this technology as well.

Cardiology and critical care medicine are at the forefront in this cutting-edge area of digital medicine. Amongst patients with arrhythmias, for instance, a digital blueprint can provide insights into the cause of mechanical discoordination within the heart and improve cardiac resynchronization therapy (CRT). In patients with prolonged QRS duration, CRT can be quite effective, but among those with intermediate ECG readings, the right course of action is less certain. Digital twin technology combines statistical methods with mechanistic modeling to bridge this gap. Similarly, investigators have found that “the combination of cardiovascular imaging and computational fluid dynamics enables non-invasive characterizations of flow fields

and the calculation of diagnostic metrics in the domains of coronary artery disease, aortic aneurysm, aortic dissection, valve prostheses, and stent design.”

Researchers are also exploring the value of digital twins to improve mechanical ventilation in the ICU. An international team from China, New Zealand, Germany, the Netherlands and Belgium have created a computer model to capture several pulmonary readings from individual patients’ ventilator data. The data are used to construct a virtual patient that is “fully automated using the hysteresis loop analysis method to identify lung elastances from clinical data.” Zhou et al explain the need for this approach by pointing out that, while positive end expiratory pressure (PEEP) helps keep patients’ alveoli open and improves oxygenation in ICU patients, finding the best PEEP settings varies from patient to patient. With that uncertainty in mind, they postulated that: “An accurate, predictive virtual patient or digital clone based on a mechanical model of patient-specific lung mechanics would augment clinical data, provide a more comprehensive picture of patient-specific state, and predict response to care.” While a detailed explanation of the technology would take us too far afield, the end result of their digital twin experiment suggests it has clinical potential. When they tested it on up to 18 patients, they found it accurately predicted patients’ lung response to changes in PEEP.

Digital twin modeling is also being used to represent individual patients’ genome, which may help clinicians link their genetic variants to specific disease risks. Leap of Faith Technologies* explains that the technology can serve as “a platform and interface for accessing genomic data relevant to the personal health record. The interface includes a map of the person’s genome and identified gene variants that may be relevant to diagnosed medical problems/conditions mapped to body systems, value sets, and clinical code sets over time.”

Digital twin technology has also been an active area of research at Mayo Clinic for many years. Ognjen Gajic, M.D.  and his colleagues have been investigating the potential role of a digital twin patient model to predict how well patients respond to treatment during the first 24 hours of sepsis. Their research: “confirmed the feasibility of development and prospective testing of causal artificial intelligence model to predict the response to treatment in early stages of critical illness. The availability of qualitative and quantitative data and a relatively short turnaround time makes the ICU an ideal environment for development and testing of digital twin patient models.”

Digital twins may not be quite ready for routine application in clinical practice, but the research strongly suggests it will eventually emerge as another useful tool in the AI/machine learning arsenal.

 *The mention of Leap of Faith Technology should not be interpreted as an endorsement by Mayo Clinic.