For years, manufacturing and engineering companies have been using digital models of products and systems for enhanced design, simulations, and maintenance. Unlike computer simulations, they incorporate Internet of Things (IoT) data to predict how a real-world system will respond to changes. Medical technologists are now applying the concept to the most complex system of all: the human body.
Revolutionary Potential
The concept of a digital model that could be used for development and troubleshooting goes back at least to the NASA space programs of the 1960s, but gained greater currency for health applications during the coronavirus pandemic. An accurate digital twin would, in theory at least, monitor a patient’s health and serve as a testbed for many different variations of therapies when needed.
Rapid growth in artificial intelligence (AI), big data, and data science will power digital twins, giving the technology the ability to revolutionize the entire healthcare system, according to a 2024 review paper in the journal npj Digital Medicine. That paper, “Digital Twins for Health: A Scoping Review,” defines a digital twin as “more than just a digital replica or a virtual model of a physical system, it can be a sophisticated representation designed to faithfully mirror the real-world system in real time, analyze its behavior, and provide predictive insights using advanced simulation, machine learning, and reasoning to help decision making.”
Another recent study that examined interest in digital twins in the field of digital health found that while they are still poorly explored in primary healthcare, they could have the potential to “improve diagnoses, treatments, care coordination, and remote monitoring.”
Commercial interest in the technology is not surprising. Last year, China announced that so-called bio digital twins would be one of its “future industries.” The global market for digital twins in healthcare could grow to $21 billion by 2028, according to MarketsandMarkets.
Basic digital twins are being trialed by people attempting to improve their health. A California startup called Twin Health is using its AI-powered Whole Body Digital Twin to replicate users’ metabolisms and generate advice to help them lose weight. The twins are generated from sensor data such as glucose monitors and pressure cuffs, lab results, and data on diet and activity. A related app can provide daily recommendations on meals, exercise, sleep, stress, and other factors affecting health. A 2023 study of 319 people with Type 2 diabetes published in Endocrine Practice found a digital twin helped to significantly improve a patient’s hyperglycemia and markers for fatty liver disease. Another study, published in Scientific Reports last year, followed 1,853 people with Type 2 diabetes and found that those using digital twins had better glycemic control and less need for diabetic medication, concluding that digital twins could be a “transformative tool” in diabetes management.
Precisely Modeling the Heart
As the leading cause of death for most people, heart disease is the focus of several digital twin projects. U.K.-based Adsilico, a spinoff from the University of Leeds, harvests enormous amounts of data to create AI-generated digital hearts that can represent specific patient metrics such as blood pressure, weight, and MRI and CT data, as well as diseases and ethnicity. Digital twins of heart implants are then inserted into an AI simulation to check for adverse effects. The company says that compared to conventional clinical trials, this approach can save time and money while capturing more of the diversity in human physiology, providing for more thorough testing and safer medical devices.
Some of the other possibilities of cardiac twins are being explored in a partnership between Japan’s National Cerebral and Cardiovascular Center and NTT Research that brings together scientists in the U.S. and Japan. It’s an attempt to turn what’s known about the physiology of the human cardiovascular system into a model that will simulate various drug therapies and other treatments to see their effects on a virtual patient. The goal is not only individualized bio-twins, but a system that could autonomously diagnose and treat patients based on how their twin responds.
“We’re trying to create a precise simulation of someone’s heart,” said Kazuhiro Gomi, head of NTT Research, part of Japan’s largest telecom company. “With that, we could automate procedures to restore a patient after acute failure or heart attack. With IT and connectivity, you could provide this top-notch treatment anywhere in the world and lower medical costs.”
The project’s concept Autonomous Closed-Loop Intervention System (ACIS) would consist of a drug-delivery device connected to a patient and controlling the timing and dosage of drugs, as well as data on their digital twin cardiovascular system and data from the general population. The device would monitor the patient, updating the digital twin and changing dosages as needed. Doctors would input the goal of treatment and supervise.
For now, it’s a science-fiction scenario, but researchers are developing the basis for ACIS: a non-AI, mechanistic model with analogs of electrical circuits representing heart contractions and blood flow in the circulatory system, and also incorporating data representing the lungs, kidneys, and hormone activity. One challenge, however, is getting the parameters of the system to match the values seen in medical literature, a difficult task since only some physiological functions can be measured. The researchers implemented the Bayesian inference statistical technique to derive parameter probabilities based on limited clinical data. After 26 rounds, the results were within 0.1 standard deviations from clinical values for the model’s 20 outputs.
“There will be a spread of these parameters that work as digital twins for any patient,” said Jon Peterson, a scientist at NTT’s Medical & Health Informatics Laboratories. He said it could be five to 10 years before the system could be ready for a clinical trial, but it’s one more step in realizing the dream of a digital double that serves as a warning system and testbed. In the meantime, he said, “We’re going to precompute a digital population, so everyone will already have at least the basics of their digital twin computed.”
Further Reading
Katsoulakis, E., Wang, Q., Wu, H. et al.
Digital twins for health: a scoping review. npj Digitial Medicine 7, 77 (2024). https://doi.org/10.1038/s41746-024-01073-0
El-Warrak, L.D., and de Farias, C.M.
Could digital twins be the next revolution in healthcare?, European Journal of Public Health, 2024;, ckae191, https://doi.org/10.1093/eurpub/ckae191
Joshi, S., Shamanna, P., Dharmalingam, et al.
Digital Twin-Enabled Personalized Nutrition Improves Metabolic Dysfunction-Associated Fatty Liver Disease in Type 2 Diabetes: Results of a 1-Year Randomized Controlled Study,
Endocrine Practice 29, 12 (2023). https://doi.org/10.1016/j.eprac.2023.08.016
Shamanna, P., Erukulapati, R.S., Shukla, A. et al.
One-year outcomes of a digital twin intervention for type 2 diabetes: a retrospective real-world study. Scientific Reports 14, 25478 (2024). https://doi.org/10.1038/s41598-024-76584-7
Tim Hornyak is a Canadian journalist based in Tokyo, Japan, who writes extensively about technology, science, culture, and business in Japan.
Join the Discussion (0)
Become a Member or Sign In to Post a Comment