Interpretability analysis and external validation study on the efficacy inference of biologics in Ulcerative colitis (UC) patients

Ulcerative colitis (UC) is a long-term condition that causes ongoing inflammation in the colon, or large intestine. Inflammation is the body’s natural response to injury or infection, but in UC, the immune system mistakenly attacks healthy tissue in the colon. This leads to swelling, pain, and sores in the lining of the colon, causing symptoms like frequent diarrhea, stomach cramps, and fatigue. UC affects millions of people worldwide and can greatly interfere with daily life.

To help manage UC, many patients are treated with medicines called biologic therapies. Biologics are medications that come from living organisms, like proteins and genes. They can reduce inflammation and control symptoms in people with UC. However, people respond differently to these treatments, and doctors often cannot predict which biologic will work best for a given patient.

We will conduct research to create a tool that helps doctors choose the most effective treatment for each individual with UC. This tool will use machine learning, a type of computer-based method that can detect patterns in large amounts of information.

To build this tool, we will analyze data from previously conducted clinical trials of people with UC who have been treated with biologics. This data will include information such as age, symptoms, lab test results, and other health details. We will divide the data into two groups: one to train the tool, and one to test how well it works. We will assess the tool’s accuracy and examine which patient characteristics are most useful for predicting treatment response. We will also compare the new tool to a more traditional method called logistic regression, a commonly used method in medical research to predict the chances of a certain outcome, such as whether a patient will respond to treatment based on different personal characteristics.

Our aim is to develop a reliable and easy-to-use prediction tool that supports personalized treatment decisions for people with UC. By helping doctors select the most suitable biologic therapy from the start, this research could improve patient outcomes, reduce trial-and-error in treatment, and lead to better quality of life.