Scientists at the University of Virginia have harnessed the power of machine learning to identify a drug that could significantly reduce harmful scarring after heart attacks and other injuries.
This cutting-edge computer model not only promises breakthroughs in cardiac treatment, but also has the potential to revolutionize drug discovery for a variety of complex diseases.
research The research, led by computational biologists Anders R. Nelson, Ph.D., and Jeffrey J. Sauserman, Ph.D., from UVA’s School of Biomedical Engineering, combines decades of human knowledge with a new approach called “logic-based machine learning.”
How do drugs affect fibroblasts?
The multidisciplinary team aimed to better understand how drugs affect fibroblasts. Fibroblasts are essential cells for heart repair, but they are also known to cause harmful scarring known as fibrosis.
“Many common diseases such as heart disease, metabolic diseases, and cancer are complex and difficult to treat,” explains Dr. Nelson. “Machine learning can help reduce this complexity, identify the most important factors contributing to disease, and better understand how drugs change diseased cells.”
Unlike previous efforts that focused on specific aspects of fibroblast behavior, UVA researchers used an innovative machine learning model to identify 13 promising drugs in human fibroblast cells. We predicted the impact on
This model identified a potential candidate to prevent scarring and explained how it works. This dual feature is important for designing effective clinical trials and understanding potential side effects.
idiopathic pulmonary fibrosis
One of the findings of this study is the potential of pirfenidone, a drug already approved by the FDA for idiopathic pulmonary fibrosis. This model revealed a new explanation for how pirfenidone inhibits contractile fibers within fibroblasts and contributes to cardiac stiffness.
This model predicts the effects of the experimental Src inhibitor WH4023 on different types of contractile fibers, and this finding was experimentally validated using human cardiac fibroblasts.
Although future studies are needed to test the effectiveness of these drugs in animal models and human patients, the UVA team is optimistic about the transformative potential of machine learning.
“We look forward to testing whether pirfenidone and WH4023 also inhibit fibroblast contraction in scars in preclinical animal models,” emphasizes Dr. Sausserman. We hope this is an example of how machine learning and human learning can work together to not only discover new drugs, but also understand how they work. ”