Scientists at the University of Virginia have made an exciting breakthrough in the use of artificial intelligence (AI) for medical research.
They’ve developed a new machine learning approach to find drugs that could prevent the harmful scarring that often occurs after a heart attack or other types of injuries.
This innovative tool has already identified a promising drug candidate that works differently from existing treatments to prevent heart scarring.
The team, led by computational biologist Anders R. Nelson, PhD, and Jeffrey J. Saucerman, PhD, believes their computer model can also be applied to predict and explain how drugs might affect a range of other diseases.
Diseases like heart disease, cancer, and metabolic disorders are complex and challenging to treat because of their intricate nature.
Machine learning simplifies this complexity by identifying key factors in disease development and understanding how potential drugs could alter diseased cells.
What sets the UVA team’s work apart is their combination of machine learning with decades of accumulated human knowledge.
This blend not only predicts effective drugs against conditions such as fibrosis (scarring) but also sheds light on how these drugs work at a cellular level. Such insights are crucial for designing clinical trials and anticipating possible side effects.
Fibroblasts are cells that play a critical role in repairing the heart after an injury. They produce collagen to heal wounds but can also cause fibrosis, leading to complications.
Previous research efforts have struggled to find targeted treatments for heart fibrosis, partly because the complete effects of drugs on fibroblast behavior were not fully understood.
To address this, the UVA team introduced “logic-based mechanistic machine learning.” This new method not only predicts the effects of drugs but also provides explanations for how they influence fibroblast behavior.
For instance, the model revealed a novel mechanism by which the drug pirfenidone, already approved for treating lung fibrosis, can suppress fibroblast activities that lead to heart stiffness.
Moreover, the model pointed to an experimental drug, WH4023, showing how it targets different fibroblast components to prevent scarring.
These findings, verified through experiments with human cardiac fibroblasts, highlight the potential of this machine learning approach to accelerate the discovery and understanding of new treatments.
Although further research, including animal studies and clinical trials, is necessary to confirm the effectiveness of these drugs, the UVA team’s work represents a significant step forward.
By combining machine learning with deep biological understanding, they are paving the way for developing new therapies for heart injuries and a wide array of other diseases.
If you care about heart health, please read studies that vitamin K helps cut heart disease risk by a third, and a year of exercise reversed worrisome heart failure.
For more information about heart health, please see recent studies about supplements that could help prevent heart disease, stroke, and results showing this food ingredient may strongly increase heart disease death risk.
The research findings can be found in PNAS.
Copyright © 2024 Knowridge Science Report. All rights reserved.
