A study undertaken by the Oregon State University College of Engineering reveals that ML techniques can offer powerful new tools for progress of personalized medicine, healthcare that optimizes outcomes for individual patients. This is based on unique aspects of biology and disease features.
The research with ML addresses long due unsolvable questions in biological systems at the cellular level. These systems tend to have high complexity, firstly, due to the large number of individual cells and, secondly, due to the highly nonlinear manner in which cells behave.
Meanwhile, nonlinear systems pose a challenge to upscale methods, which is the primary method by which researchers can model biological systems accurately at larger scales that are mostly relevant.
A linear system in mathematics or science implies any change in the input of the system results in a proportional change to the output. For example, it might describe a slope that increases by 2 feet vertically for each foot of horizontal distance covered.
Meanwhile, nonlinear systems don’t work in this way, and many of the systems of the world, including biological ones are nonlinear.
Funded partly by the U.S. Department of Energy, the new research is one of the first examples that uses ML to address challenges with modeling nonlinear systems and understanding complex processes that might be present in human tissues.
Importantly, the advent of ML has given a new tool in our repository to solve problems that could not be solved before. While are tools necessarily are not new, the applications are very different. The application of ML in a more constrained manner is beginning, and this allows to solve physical problems that had no solution earlier.