Support

Personalizing treatment plans for IBD with machine learning

Inflammatory bowel diseases (IBD), including Crohn’s disease and ulcerative colitis, are debilitating diseases of the intestine that afflict over a million Americans. Unfortunately, treatment options are limited, expensive and frequently ineffective. Even the class of drugs considered the gold standard for treating IBD, so called biologics that target the inflammatory protein TNF alpha, are very expensive and only effective in about half of all patients.

This remains a critical problem in IBD treatment, since exposure to anti-TNF therapy can make patients less responsive to other biologics after anti-TNF treatment has failed. Plus, it can take months to determine whether a patient is improving. Thus, it would be very beneficial if we could predict whether a patient will respond to anti-TNF therapy.

To address this problem, I have developed a machine learning program that can distinguish between patients who will respond to anti-TNF (responders) and those who will not (non-responders). Machine learning is a type of artificial intelligence that can learn from data and progressively improve its accuracy in recognizing patterns without being actively programmed. In our case, I use patient gene expression data to determine whether patients will respond to anti-TNF therapy. By inputting gene expression data of known responders and non-responders, the program is trained to recognize gene expression patterns that predict whether patients will benefit from anti-TNF therapy.

So far, I have identified two genes that predict with approximately 90 percent accuracy whether a patient will respond to anti-TNF before treatment is administered. Receiving a SPARK grant would allow me to considerably increase the patient data set used to train the program. This will improve prediction accuracy and help uncover additional genes whose expression predicts anti-TNF responsiveness. Ultimately, this project would allow us to determine whether anti-TNF is the most promising first-line treatment eliminating months or even years of ineffective treatments for patients.