The role of machine learning in advancing precision medicine with feedback control
Ksenia Zlobina, Mohammad Jafari, Marco Rolandi, Marcella Gomez
Abstract
The capacity of machine-learning methods to handle large and complex datasets makes them suitable for applications in precision medicine. Current methods automate data analysis and predict physiological outcomes of patients with various types of clinical data to inform treatment strategies. In this perspective, we propose ways in which machine learning can be leveraged even further to advance methods of optimizing patient treatment. Namely, machine learning can be used to expand applications of feedback control to direct the response of complex biological systems predictably and automatically. This paves the way for highly sophisticated treatments that continuously adapt to an individual patient’s response. The elements of control that can be improved using machine learning include sensor data analysis, modeling, and methods of reconfiguring the control algorithm “on the fly.” We discuss the control challenges unique to the analysis/control of complex biological systems, existing work, and areas that remain underdeveloped.