The problem
Type 1 Diabetes (T1D) is a chronic disease in which the body can no longer produce insulin. Managing it is hard: patients must control many behavioural factors that affect their glycemic outcomes.
The approach
The paper explores T1D patient behaviours with a learning approach based on Signal Temporal Logic (STL). STL formulas learned from real patient data characterise behaviour patterns associated with different levels of glycemic control.
These logical characterisations are readable, so they can give clinicians and patients feedback on behavioural changes that could improve T1D control.
Results
Both individual-level and population-level behaviour patterns are learned from a clinical dataset of 21 T1D patients.