Intelligent processing of ambulatory monitoring data in gestational diabetes

This paper describes a system to analyze self-monitoring data of gestational diabetic patients with the final goal of obtaining a daily assessment of their metabolic control. Our approach is based on a causal probabilistic network that represents qualitative medical knowledge expressed as relations between causes and effects and their probabilities. The system is able to manage incomplete data and to make reasoning under uncertainty, the two most important constraints when analyzing ambulatory monitoring data. The prototype works with blood glucose and ketonuria data, as well as timing and quantity deviations of insulin and food intakes. The outcomes provided by the system are information on patient transgressions of the prescribed treatment, the need of treatment adjustments and alterations on the patient metabolism.