Machine Learning Techniques Bracing to Perceive Drug Impacts on the Human Brain

By CIOReview | Friday, December 22, 2017
236
376
66

Researchers from University College London are on an endeavor to integrate knowledge and models of the brain with clinical trial information, with the purpose of revealing new drug therapies which have till now remained undetected. To this end, they are capitalizing on machine learning techniques which can enhance the ability of clinical trials to find out whether treatments performed on the brain are effective. It is approximated that such techniques have the potential to solve the issue of high discrepancy among patients with an illness since they can take into account hundreds and thousands of variables and analyze how these variables are related to treatment outcomes—an unattainable feat for conventional approaches.

The team of investigators used brain damage caused by a stroke to carry out their analysis. They gathered large-scale data on tomography scans of strokes along with relevant patient information. For each patient, the team included the full complex anatomical damage outline. As a measure of how the stroke affected the brain, they used gaze direction assessed when the patients were undergoing brain scans and had their algorithms examine these images. Thereafter, they went on to experiment a cluster of various drugs at different doses on virtual patients to check what the end result would be. Once they got the results, they evaluated them using their machine learning technique as against the commonly used statistical methods, with the former being able to incorporate a lot more variables than what the latter normally used in studies of drugs intended for the brain.

The advantage was specifically evident for drugs that reduced lesion volume. Using traditional statistics, a lesion would need to shrink by 78.4 percent for tests to identify the effect. The machine learning technique, on the other hand, required only a 55 percent shrinkage to detect a treatment impact. The team envisions that machine learning usages—especially helpful for examining intricate structures like the human brain—will undoubtedly find their way into clinical trials and other medical research settings.