Primary Submission Category: Causal Inference Applications
Comparisons of the Treatment and Side Effects of Several Bariatric Surgery Procedures: An Observational Study via Random Forest-based and Neural Network-based Approaches
Authors: Qishuo Yin, Qishuo Yin, Jiawei Zhang, Carlos Fernandez-Granda, Siyu Heng,
Presenting Author: Qishuo Yin*
Bariatric surgery is an effective treatment for obesity, as well as diabetes, high blood pressure, sleep apnea, and high cholesterol. Over the past decades, several bariatric surgery procedures based on different techniques have been widely performed in practice. However, there is a lack of rigorous causal comparisons of their treatment and side effects. In this work, we conduct a large-scale observational study to compare the effects of several widely used bariatric surgery procedures on Bariatric surgery complication risk and weight loss using the datasets from the American College of Surgeons Metabolic and Bariatric Surgery Accreditation and Quality Improvement Program (MBSAQIP). We apply several state-of-art machine learning-based causal inference approaches such as Causal Forest, Orthogonal Random Forest, and Dragonnet to better leverage the large datasets to provide accurate and powerful causal comparisons, both at population and individual levels. On the one hand, the results from our population-level causal comparisons provide rigorous statistical evidence for the appropriateness of the current guidelines for Bariatric surgery procedures provided by the American Society for Metabolic and Bariatric Surgery (ASMBS). On the other hand, our individual-level causal comparisons offer data-driven suggestions on Bariatric surgery procedure selections for individual patients, of which the corresponding user-friendly statistical software will be provided on the website soon.