University of Virginia scientists have developed a brand new approach to machine learning — a type of artificial intelligence — to discover drugs that help minimize harmful scarring after a heart attack or other injuries.
The brand new machine-learning tool has already found a promising candidate to assist prevent harmful heart scarring in a way distinct from previous drugs. The UVA researchers say their cutting-edge computer model has the potential to predict and explain the results of medication for other diseases as well.
“Many common diseases similar to heart disease, metabolic disease and cancer are complex and hard to treat,” said researcher Anders R. Nelson, PhD, a computational biologist and former student within the lab of UVA’s Jeffrey J. Saucerman, PhD. “Machine learning helps us reduce this complexity, discover a very powerful aspects that contribute to disease and higher understand how drugs can modify diseased cells.”
“By itself, machine learning helps us to discover cell signatures produced by drugs,” said Saucerman, of UVA’s Department of Biomedical Engineering, a joint program of the School of Medicine and School of Engineering. “Bridging machine learning with human learning helped us not only predict drugs against fibrosis [scarring] but in addition explain how they work. This data is required to design clinical trials and discover potential unwanted effects.”
Combining Machine Learning, Human Learning
Saucerman and his team combined a pc model based on a long time of human knowledge with machine learning to higher understand how drugs affect cells called fibroblasts. These cells help repair the guts after injury by producing collagen and contract the wound. But they can even cause harmful scarring, called fibrosis, as a part of the repair process. Saucerman and his team desired to see if a choice of promising drugs would give doctors more ability to stop scarring and, ultimately, improve patient outcomes.
Previous attempts to discover drugs targeting fibroblasts have focused only on chosen features of fibroblast behavior, and the way these drugs work often stays unclear. This data gap has been a serious challenge in developing targeted treatments for heart fibrosis. So Saucerman and his colleagues developed a brand new approach called “logic-based mechanistic machine learning” that not only predicts drugs but in addition predicts how they affect fibroblast behaviors.
They began by the effect of 13 promising drugs on human fibroblasts, then used that data to coach the machine learning model to predict the drugs’ effects on the cells and the way they behave. The model was capable of predict a brand new explanation of how the drug pirfenidone, already approved by the federal Food and Drug Administration for idiopathic pulmonary fibrosis, suppresses contractile fibers contained in the fibroblast that stiffen the guts. The model also predicted how one other sort of contractile fiber could possibly be targeted by the experimental Src inhibitor WH4023, which they experimentally validated with human cardiac fibroblasts.
Additional research is required to confirm the drugs work as intended in animal models and human patients, however the UVA researchers say their research suggests mechanistic machine learning represents a robust tool for scientists in search of to find biological cause-and-effect. The brand new findings, they are saying, speak to the good potential the technology holds to advance the event of latest treatments — not only for heart injury but for a lot of diseases.
“We’re looking forward to testing whether pirfenidone and WH4023 also suppress the fibroblast contraction of scars in preclinical animal models,” Saucerman said. “We hope this provides an example of how machine learning and human learning can work together to not only discover but in addition understand how recent drugs work.”
The research was supported by the National Institutes of Health, grants HL137755, HL007284, HL160665, HL162925 and 1S10OD021723-01A1.
