A Self-Adaptively Evolutionary Screening Approach for Sepsis Patient

Today, sepsis syndrome is one of the leading cause of death globally, and is of great clinical importance. In this paper, we present a self-adaptively evolutionary sepsis screening system to shorten the time of syndrome detection and improve the positive effect of treatment, with the screening frequency and content can be automatically adjusted according to the current status of the patient. First, we propose a novel graphical computation model named AdapDBN with a clearly defined syntax for the medical knowledge presentation, especially for the presentation of the pathophysiology model of the disease. Then, the semantics of AdapDBN is formally defined for the evolutionary inference of syndrome onset probability. Finally, we demonstrate how to initialize AdapDBN with sepsis-related epidemiologic statics, published clinical research and physician's knowledge and how to incorporate it into existing sepsis screening and decision support flow. We evaluate its effectiveness and superiority with comparisons to existing computation techniques.