Cardiovascular Bio Digital Twin Promises Faster Medical Diagnoses and Better Outcomes

Imagine if a doctor could almost instantly rule out two common drugs as being ineffective for a given patient, and quickly settle on using a third, thus saving weeks of trial and error. Or if that same doctor could essentially predict a given patient would develop hardened arteries in a few years, and counsel the patient on how to avoid that fate.

That is the promise of the Cardiovascular Bio Digital Twin under development at the NTT Research Medical and Health Informatics (MEI) Lab.

At the recent NTT Research Upgrade 2023 event, Iris Shelly, a Research Scientist in the MEI Lab, said her group chose the cardiovascular system as the first target for the bio digital twin technology for a few reasons. For one, in looking at the top 10 causes of death worldwide, ischemic heart disease and stroke together make up almost half of them, she said.

The cardiovascular system is also a good target because it essentially consists of a pump, the heart, that circulates blood through a series of valves and pipes. “Thanks to this basic basis in fluid mechanics, we can use some relatively simple equations to model this basic blood flow, or hemodynamics,” she said.

In the cardiovascular bio digital twin model, numerous other body systems are layered on top of that basic framework. They include the autonomic nervous system to orchestrate blood pressure and heart rate, kidneys for renal system interaction with blood and fluid volumes, and lungs to provide oxygenation of the blood.

This generic model is then tuned to individual patients using a software platform from Embody Bio that enables it to be combined with an individual patient’s clinical data as well as data gathered by running the model through a range of parameter values on a large digital population.

“These internal parameter values are our hypotheses,” Shelly said. “They’re basically our best guess at the range of physiological values we could see in the real population.”

With this digital twin model at their disposal, the idea is that doctors can far more easily determine effective treatments on patients as well as make predictions.

Say a patient comes to the doctor with hypertension. The doctor needs to find a way to lower the patient’s blood pressure. Normally that would mean prescribing a drug, sending the patient home, and waiting a couple of weeks or more to see whether it works. The bio-digital twin enables the doctor to run tests for the drug in the office and immediately see how effective it would be on that particular patient. Maybe the doctor finds the third drug is the one that performs best for this patient.

“Without the digital twin, these two unsuccessful treatment options could have cost weeks or months in the patient’s journey to better health, potentially adding uncomfortable side effects,” Shelly said. “Not to mention the financial cost of the drugs themselves and just the negative health impacts of persistent high blood pressure.”

In a similar fashion, a doctor could also use the digital twin to peer into the future and predict that a patient who is healthy today is nonetheless at risk of developing high blood pressure a few years down the road. At that point, the doctor can recommend lifestyle changes that could alter that outcome, whether physical activity or changes to diet.

Iris Shelly is a scientist in the Medical and Health Informatics laboratories. After graduating with a degree in biomedical engineering from Washington University in St. Louis, she began working in software verification of implantable cardiac devices. At the same time, she completed a master’s degree in electrical engineering, with a focus on biological signal processing. Most recently, she worked as an applied research engineer, designing and developing cardiac signal and arrhythmia detection algorithms for implantable cardiac monitors and pacemakers. Her work in the MEI lab focuses on development of the Cardiovascular Bio Digital Twin, starting with supporting the model architecture and interfaces.