Time-stage driven pathfinding framework for optimized medical treatments

Coming up with the correct patient-specific treatment procedure is a non-trivial task. One of the main challenges is that the problem is combinatorial, and each treatment can be followed by numerous treatments. In real-life scenarios, one cannot experiment with all probable treatment combinations that may provide the necessary positive outcome. In addition, the task of correct drug prescription is challenging because, at different time-stages, the prescription may provide different results, a concept not widely explored in the literature on computational modeling. To address this task, we model the problem as a search problem and propose two algorithms that construct a directed acyclic graph (DAG), a directed cyclic graph, and a tree for each patient to explore various treatment combinations in an iterative and recursive manner. Each patient has three corresponding datasets, one for each stage, representing the features the patient has demonstrated during the recovery process. As a result, we provide a framework for identifying treatment options that may not have been explored previously while incorporating the concept of time-stage-based observations in the search procedure as novel contributions to the existing literature. We provide evaluation and scaling methods and identify current limitations and future research directions.