Toward Designing Mobile Software to Predict Hypoglycemia for Patients with Diabetes

Abstract

To alert diabetes patients of incipient hypoglycemia, we developed a hypoglycemia prediction algorithm and elicited design inspiration for new glucose management software. To identify the predictive factors, we conducted surveys, interviews, and diary studies, and developed a prediction model that uses self-monitored blood glucose. We tested the accuracy of prediction algorithms achieved by different machine learning methods, and found that the proposed algorithms have potential to predict hypoglycemia. Based on the proposed algorithm, we designed a new mobile application concept to support patients’ self-care, especially to avert hypoglycemia.