Researchers at University College London are attempting to combine knowledge and models about the brain with clinical trial data, with the ultimate goal of discovering new drug therapies that would otherwise be undetected. The researchers hope to use machine learning techniques that can process lots of data to find correlations that can be spotted in similar datasets compiled later.
The investigators compiled the largest ever collection of tomography scans of strokes along with relevant patient data, including treatment, and had their algorithms study these images. The team then virtually tried a bunch of different drugs at different doses on these virtual patients to see what would be the expected results. They then assessed the results using their machine learning system versus commonly used statistics, with the former being able to take into account a lot more than the few variables typically used in the latter in studies of drugs targeting the brain.
“Stroke trials tend to use relatively few, crude variables, such as the size of the lesion, ignoring whether the lesion is centred on a critical area or at the edge of it. Our algorithm learned the entire pattern of damage across the brain instead, employing thousands of variables at high anatomical resolution. By illuminating the complex relationship between anatomy and clinical outcome, it enabled us to detect therapeutic effects with far greater sensitivity than conventional techniques,” said Tianbo Xu, the first author of the study appearing in journal Brain.
The researchers showed a probable high level improvement in how well the effects of trialed drugs can be noticed that would otherwise be missed using typical statistical models.
Study in journal Brain: High-dimensional therapeutic inference in the focally damaged human brain…