What is Drug repurposing?
Drug repurposing involves investigating existing drugs for new therapeutic purposes. It is a useful approach to distinguish new uses for existing drugs, providing a quicker transition from bench to bedside. For example, earlier Botox injections were accepted to treat crossed eyes. However, they are now used to treat migraines and are a top cosmetic strategy to reduce wrinkles’ appearance.
Real-world data that contains observational data about millions of patients is captured by electronic health records or insurance claims and prescription data. This data provides information on large cohorts of users for many drugs. But getting to those new uses involves time-consuming and costly randomized clinical trials to ensure that a drug considered effective for one disorder will be useful as a treatment for something else. Drug repurposing could lower the risk associated with the safety testing of new medications and reduce the possible time to introduce a drug into the market for clinical use.
A new deep learning framework
Researchers at Ohio State University have proposed a useful and efficiently customized framework that can generate and test multiple patients for drug repurposing using a retrospective analysis of real-world data. The proposed deep-learning method emulates randomized clinical trials for drugs present in a large-scale medical claims database.
The motivation is to find drugs for diseases without any current treatment. Though the study focuses on repurposing drugs to prevent heart failure and stroke in patients with coronary artery disease, the framework is flexible and could be applied to any condition if the disease outcome is defined.
Real-world evidence consists of many factors or confounders (such as age, gender, race to disease severity, and the existence of other diseases) that make it challenging for humans to work on. By employing machine learning algorithms, we can now handle hundreds or thousands of these parameters efficiently.
The team has used insurance claims data on about 1.2 million heart-disease patients, giving information on their prescribed treatment, illness outcomes, and different potential confounders’ values. The deep learning algorithm also considers the passage of time in each patient’s experience for every appointment, prescription, and diagnostic examination.
The researchers categorized the active drug and placebo patient groups by applying causal inference theory. With this theory, they didn’t solve whether drug X or drug Y works for a specified disease or not but comprehended which treatment will have better performance. This addresses the issue of having multiple treatments. The model tracked patients for two years and compared their disease status to whether they took medications, the drugs they received, and when they started the medical treatment.
The analysis yielded nine drugs considered likely to provide the therapeutic benefits (out of which three are currently in use) and identified six candidates for drug repurposing. The study also suggested that a diabetes medication (metformin and escitalopram) used to treat depression and anxiety could lower heart failure and stroke risk in the model patient population. At present, both of the drugs are being examined for their effectiveness against heart disease.
The researchers state that this work reveals how AI can test a drug on a patient, speed up hypothesis generation, and clinical trial. However, they also claim that physicians will always make drug decisions, which will never be replaced.
Tanushree Shenwai is a consulting intern at MarktechPost. She is currently pursuing her B.Tech from the Indian Institute of Technology(IIT), Bhubaneswar. She is a Data Science enthusiast and has a keen interest in the scope of application of artificial intelligence in various fields. She is passionate about exploring the new advancements in technologies and their real-life application.