Objective: Intentional insulin omission is a unique inappropriate compensatory behavior that occurs in patients with type 1 diabetes mellitus, mostly in females, who omit or restrict their required insulin doses in order to lose weight. Diagnosis of this underlying disorder is difficult. We aimed to use clinical and laboratory criteria to create an algorithm to assist in the detection of intentional insulin omission.
Method: The distribution of HbA1c levels from 287 (181 females) patients with type 1 diabetes were used as reference. Data from 26 patients with type 1 diabetes and intentional insulin omission were analysed. The Weka (Waikato Environment for Knowledge Analysis) machine learning software, decision tree classifier with 10-fold cross validation was used to developed prediction models. Model performance was assessed by cross-validation in a further 43 patients.
Results: Adolescents with intentional insulin omission were discriminated by: female sex, HbA1c>9.2%, more than 20% of HbA1c measurements above the 90th percentile, the mean of 3 highest delta HbA1c z-scores>1.28, current age and age at diagnosis. The models developed showed good discrimination (sensitivity and specificity 0.88 and 0.74, respectively). The external test dataset revealed good performance of the model with a sensitivity and specificity of 1.00 and 0.97, respectively.
Discussion: Using data mining methods we developed a clinical prediction model to determine an individual's probability of intentionally omitting insulin. This model provides a decision support system for the detection of intentional insulin omission for weight loss in adolescent females with type 1 diabetes mellitus.
Keywords: Hemoglobin A1c; Insulin omission; data mining; decision tree; type 1 diabetes.
Copyright © 2013 Wiley Periodicals, Inc.