Are digital twins the future of better healthcare?

By: Beverly Bell & Rajesh Jaluka

The saying “Houston, we have a problem” long ago became part of everyday discourse. Less widely known, however, is a technique inspired by NASA’s Apollo 13 mission that stands to influence healthcare.

In April 1970, 210,000 miles away from Earth, an oxygen tank exploded on Apollo 13, threatening the lives of the astronauts on board.

Fortunately, in training for the mission, NASA used several simulators to prepare the crew.

So, when the SOS call came in, mission control turned to the simulators to match the conditions of the aircraft and find a solution to bring the astronauts home safely.

The NASA equipment might not have been digital, but technologists are finding new applications for the idea of a simulator.

Digital twins are highly accurate and integrated models that can simulate the performance (and potential failings) of the system they are based on.

Over the last few decades, the healthcare industry has made great strides toward a digital-first future. And now, with increased ease of access to digital records and advances in artificial intelligence, patients could soon have digital twins of their own.

These models could draw on our medical information (blood type, eye color, height, weight, x-rays, MRI results and more) and non-medical data (lifestyle data from fitness apps, phone and social media usage) to produce a digital representation of the entire person, their health and their behaviors.

Such a twin would be a game changer across the healthcare industry for providers, payors and patients, as well as pharmaceutical companies.

Below, we’ll explore some of the potential applications of digital twins for healthcare, and what they might mean for different actors across the sector.

Digital twins for providers

From a physician’s perspective, one of the most impactful applications of digital twins would be the ability to deliver more precise medical treatments for each person under their care.

For example, if you know that your female patient has inherited harmful BRAC genes—specifically, the variant BRCA1 (versus BRCA2)—steps can be taken to reduce the risk of cancer or aid early detection.

Moreover, combining a digital cadaver with virtual reality could create a three-dimensional digital twin for students to practice procedures on. Even seasoned surgeons could use these twins to master their techniques or try alternative strategies for complex and risky procedures.

Digital twins for payors

With the current emphasis on improving health equity across the delivery ecosystem, digital twins could help reduce treatment costs and lower insurance premiums.

Complications and mortality rates are high in the months following surgery,1 but a prognostic scoring based on body mass index, known drug intolerances, comorbidities and more could help to predict mortality—and therefore minimize risk.

Beyond prognostic scoring, digital twins for healthcare could help tailor the surgery plan and postoperative care. Insurance companies rely on hundreds of indicators to calculate risk and arrive at premiums for their plans. With digital twins, they could target high-risk areas, sponsor studies to find solutions that might lower those risks, and champion widespread adoption to drive the risk and costs even lower.

Digital twins for the pharmaceutical industry

Armed with digital twin technology, pharmaceutical companies could accelerate clinical trials to bring new drugs to market sooner.

Clinical trials typically have three phases. Phase 1 usually involves fewer than 100 patients. Phase 3 trials involve a few thousand. As a result, these phases can take years to complete.

People may be more willing to enroll their digital self in such trials. Theoretically, this increased enrollment could also raise the efficacy of clinical trial results. Trials on actual human beings could then be reserved for phase 3, provided phases 1 and 2 (potentially using digital twins only) offer sufficient evidence of safety and efficacy.

The idea of testing drugs or treatment on a few thousand individuals and then expecting them to work on millions is also a flawed concept. Since 2012, the U.S. Food and Drug Administration has recalled over 15,500 drugs²—equating to nearly four drugs a day.³ With digital twins, the rigor of the trial process could be significantly increased.

Furthermore, multiple chemical compositions of a medication could be tested to create multiple variations to support different diseases.

Digital twins for patients

In addition to improving health outcomes and reducing costs, a digital twin could significantly enhance patients’ experiences when dealing with acute and chronic healthcare conditions.

They would, for instance, be able to access expertise from across the world for second or third opinions before committing to a treatment plan. By becoming aware of their own unique situations, they could also make lifestyle adjustments that best fit their needs. A future might even be possible where a digital twin is kept up to date with data from wearables to predict medical emergencies such as low blood sugar episodes or heart attacks—and notify caregivers to take preventive actions.

For patients with cancer or other rare or potentially terminal diseases, experimental trials can be a last hope. Through a hospital’s digital front door, patients could locate clinical trials for specific types of cancer much more easily, and then request to participate in those same trials. 

Rather than submitting to a clinical trial physical exam screening procedure, the patient could give authorization for the researcher to access their digital twin. The researcher could then run the chemical compound against the digital twin and if it meets, say, a 50% threshold of possible impact, the person would be granted access to the trial.

Steps to guide your digital twin project

The following are some key considerations when pursuing a digital twin project, regardless of whether the use cases are related to a provider, payor, or pharmaceutical company.   

1. Hypothesize. Start by clearly defining the desired outcome of the project, its target audience and key stakeholders.
2. Evaluate. Data is the most essential element for any digital twin project. Compile the various data types that will be needed for the project and validate their availability.  Then, assess the feasibility, ethical, legal, privacy and security aspects of the project. 
3. Approve. To ensure the proper administrative review, leverage the innovation or IT strategic council to present the details and obtain approval and funding for the project.  If appropriate, have the project reviewed by the Internal Medicine Research Internal Review Board.
4. Analyze. Conduct a deeper analysis of the data to look for:
   • Unbalanced data, which can introduce bias (for example, insufficient data for certain races, genders and age groups)
   • Inadequate data records, which can lead to skewed extrapolation
   • Inconsistent data values, which can make the model untrustworthy
   • Missing data fields, which can affect the validity of the model
5. Design. Establish a data architecture to design the data lake, data catalog, integrations and more. A robust cybersecurity framework must be the foundation of the data architecture.
6. Wrangle. To maintain high data quality, build an automated data pipeline to validate, cleanse, transform and enrich your data.
7. Manage. Incorporate a sound data management and governance strategy and process rigorously to maintain your ethical, legal, privacy and security posture.  
8. Share. Demonstrate integrity by transparently communicating the details of the data in use, the way it is being used and the results with all stakeholders and data subjects.

Beverly Bell is the Clinical Advisor for US Healthcare and Government at Kyndryl.

Rajesh Jaluka is the Vice President and CTO for US Healthcare and Government at Kyndryl.

¹ Long-term mortality following complications after elective surgery: a secondary analysis of pooled data from two prospective cohort studies, British Journal of Anaesthesia, October 2022

² FDA Dashboards - Recalls, U.S. Food & Drug Administration, 2023

³ Drug recalls are common, Harvard Health Publishing, March 2023