Predicting the dropout risk after treatment of obesity: Logistic regression analysis and deep neural network analysis

Backgrounds Severely obese patients must follow strict regimens of diet, exercise, and medical therapy. However, such comprehensive weight-loss programs have high dropout rates. In this study, we developed a machine learning prediction model to aid in the early detection of high-risk-dropout patients.Methods 102 severely obese patients were monitored for 3 years to assess their risk of dropout from a comprehensive weight-loss program. The program targeted a 5% weight loss. It consisted of three main components, which include behavioral modification (goal setting and charting weight four times daily), diet, and exercise. A machine learning model was developed to predict dropout risk based on a 1-year dropout event. To extend the prediction ability past 1 year, we plotted a 3-year Kaplan-Meier survival curve using a deep learning (DL) algorithm and logistic regression (LR) classifications.Results Their mean age was 49±14 years, 43% were male, BMI was 42 kg/m2, and hemoglobin A1c was 7.6%. Additionally, 76% had diabetes, 21% had impaired glucose tolerance. After 1 year, the dropout rate was 19%. Using oral hypoglycemic agents had a lower risk of 3-year dropout [Odds ratio 0.26 (95% CI: 0.08 to 0.83, p = 0.023)]. The area under the curve (AUC) was better with DL than LR methods for predicting dropout at 3 years (0.97 vs. 0.77, p<0.001). The AUC for DL was also better than LR using binary classifications (0.86 vs. 0.68, p=0.001).Conclusions We demonstrated a higher precision with machine learning than with the standard logistic regression, based on limited sample size and information available during hospital admission. It is vital to note that machine learning was more accurate than standard analysis. This may have clinical significance because machine learning could be used to identify high-risk groups and allow for early intervention.