They say everybody has a double, and that could be very good news for some heart patients. A Carle Illinois College of Medicine researcher and his team are developing tools to create ‘digital twins’ for heart failure patients to help pinpoint the cause and track the development of heart disease. The team’s work may eventually identify the best treatments for an individual patient’s disease profile and offer new hope for patients whose disease doesn’t respond to existing treatments.
CI MED MD/PhD candidate Pranav Dorbala is leveraging his background in computer science and the latest medical research to create individualized digital twins or virtual representations of a patient’s heart. “Through this ‘digital twin’ approach, we can leverage the prior knowledge of science and the data-fitting abilities of machine learning to create a much more powerful model of the human heart,” Dorbala explained.
Working with a multidisciplinary team from the Coordinated Science Lab’s DEPEND group (under the leadership of Electrical and Computer Engineering Professor Ravishankar Iyer), Dorbala is using data on genetics, proteins, and other factors known to play a part in heart failure to create a personalized digital replica. “If we model a heart in the computer, we can tailor the digital model to fit the patient’s physical heart,” he said. And unlike a physical model, digital twins are flexible under changing conditions and over time. “Through this digital model, we can leverage machine learning, deep learning, and reinforcement learning (ML) to simulate the effects of various individual patient-related factors, heart structure and function, and treatments to identify the risk of developing heart failure and novel treatment options.”
“Heart failure is largely driven by changes in the heart that occur to compensate for a loss of heart function (either due to specific insult like a heart attack or due to the aging process including the effects of high blood pressure, diabetes, and other factors),” Dorbala explained. “When the heart over-corrects for this loss of function, heart failure occurs.” This process of heart ‘remodeling’ is one of the main pathways in the development of heart failure, making the digital double’s ability to change even more valuable in identifying causes and potential treatments.
Dorbala’s digital twin research builds on his earlier work focused on understanding the various mechanistic pathways the heart uses to remodel itself. The work looked at protein systems and pathways in the development of a broadly defined type of heart disease known as Heart Failure with Preserved Ejection Fraction (HFpEF) which has no standard effective treatment. “If we identify these pathways, we can group patients with similar pathway-based changes to better predict who is at risk for heart failure and who may benefit from specific drug therapies,” Dorbala said.
The team’s model will be tested on large groups of clinical data from real patients, tracking which patients develop heart failure over time. “Our next steps are to identify methods to output translational clinical measures from our digital twin to maximize the benefit of this innovation in the clinical setting,” Dorbala said. “We want to integrate the proteomic pathway data such that we understand how enrichment of specific biological pathways results in specific remodeling patterns in the heart.”
Dorbala says the ultimate step in the digital twin research project will integrate clinical trial data to identify therapies, including drug treatments, that result in the greatest benefit and best outcomes in patients with heart failure that hasn’t responded to existing therapies.
Dorbala’s mentors on the digital twin project include project lead, Electrical and Computer Engineering and CI MED Professor Ravishankar Iyer, and collaborator Dr. Amil Shah of the University of Texas Southwestern Medical Center. The team also works closely with collaborators at the Brigham and Women’s Hospital (Harvard Medical School) in Boston.