An intelligent intervention strategy for patients to prevent chronic complications based on reinforcement learning

Chronic complications refer to defects that initially have a limited impact on chronic patients’ normal lives, but that may accumulate into a major danger if they are not treated in a timely manner. A natural problem is how to take appropriate intervention strategies to prevent chronic patients from developing severe chronic complications. We develop a personalized preventive intervention model (abbreviated as PPIM) to intelligently administer chronic patients’ oral medications across their lifespans based on the reinforcement learning framework. In the developed model, the states include the patient’s medication state and chronic complication status, the actions include medication choice and dosage adjustment, and the reward depends on the patient’s health status after the medical treatment. A state transition matrix is established to formulate the change in the patient’s states and to estimate the reward. The action that maximizes the reward will be chosen. Theoretical analysis proves that PPIM can effectively prevent or delay the progression of chronic complications. Simulations on a dataset of 1429 diabetic patients show that compared with the “one size fits all” intervention strategy, PPIM reduces the five-year incidence rate of Diabetic Nephropathy (DN) by 17.78 % and delays the time to develop DN by 7.38 years on average.