The role of Artificial Intelligence in Bariatric Surgery Weight Loss Outcomes

Since the naissance of artificial intelligence (AI) research in the 1950s, there has been unprecedented growth in the application of AI tools across many industries.

In the healthcare setting, AI tools can augment clinical decision-making and, among surgical specialties like metabolic and bariatric surgery (MBS), help guide patient expectations. Patients undergoing MBS procedures require multi-disciplinary treatment which can be lengthy and complex. Weight loss outcomes following MBS are variable as demonstrated by a 2018 JAMA Surgery Longitudinal Assessment of Bariatric Surgery (LABS) Study publication; Courcoulas et al. 1 identified six different weight loss trajectory patterns over a 7-year period following Roux-en-Y gastric bypass (RYGB). About 10% to 30% of MBS patients will suffer insufficient weight loss (IWL) post-operatively. Factors impacting IWL include demographics, socioeconomic status, and comorbidities. While success following MBS is frequently dominated by a single outcome like body mass index (BMI) or total body weight change, this is deterministic and rigid. As modern obesity treatment paradigms continue to evolve, momentum is shifting toward more personalized, predictive, and precise patient care which is where AI can have a significant impact.

Using AI to predict weight loss trajectory after MBS procedures is a promising addition for enhancing preoperative counseling and procedural selection. Current algorithms utilize large datasets and largely focus on machine learning and other statistical methods. For example, the American College of Surgeon’s MBSAQIP tool 2 and the SOPHIA study 3 provide valuable insights into weight loss trajectories after certain MBS procedures. The MBSAQIP tool provides predicted weight loss after adjustable gastric band (AGB), sleeve gastrectomy (SG), RYGB, or Biliopancreatic Diversion with Duodenal Switch (BPD/DS) using multiple patient characteristics, medical comorbidities and BMI. This tool was developed using generalized estimating equation modeling and predicts change in BMI well but only within the first 12 months post-operatively.

The “Stratification of Obesity Phenotypes to Optimize Future Therapy” or SOPHIA study was an extensive collaborative European project across multiple university and industry stakeholders. The authors used a robust machine learning model on 10,000 pts to predict 5-

year weight trajectories after SG, RYGB, or AGB. The model was validated across multiple global cohorts and was built into an easy-to-use, web-based tool. 4 The SOPHIA tool is the first true AI model that exists, including diabetes duration and smoking status which are lacking from other weight loss prediction tools. It offers high accuracy for certain subgroups but declines over time. Not surprisingly, weight loss patterns following SG and RYGB were similar up to 12 months post-operatively but separated soon after. While the SOPHIA tool has a strong methodologic approach there are caveats to its external validity as follows: 1) baseline mean BMI ≤ 48 kg/m 2 for all the testing and training studies, 2) predominantly Caucasian patients with percentages of different racial or ethnic groupings not reported, and 3) only 7 variables were included with lack of socioeconomic or behavioral factors.

While AI offers significant promise in the field of MBS, one must proceed with caution as “one size does not fit all.” Each AI tool or model will have its own design limitations, caveats for clinical applicability, and ethical considerations. These tools must be validated across different patient populations with the utmost level of transparency and communication which will require global research studies and extensive collaboration.

References:

1. Courcoulas AP, King WC, Belle SH. Seven-Year Weight Trajectories and Health Outcomes in the Longitudinal Assessment of Bariatric Surgery (LABS) Study. JAMA Surg. 2018;153(5):427-434. doi:10.1001/jamasurg.2017.5025

2. Grieco A, Huffman KM, Cohen ME, Hall BL, Morton JM, Ko CY. The Metabolic and Bariatric Surgery Accreditation and Quality Improvement Program Bariatric Surgical Risk/Benefit calculator: 1-year weight. Surg Obes Relat Dis. 2023 Jul;19(7):690-696. doi: 10.1016/j.soard.2022.12.028.

3. Saux P, Bauvin P, Raverdy V. et al. Development and validation of an interpretable machine learning-based calculator for predicting 5-year weight trajectories after bariatric surgery: a multinational retrospective cohort SOPHIA study. Lancet Digit Health. 2023 Oct;5(10):e692- e702. doi: 10.1016/S2589-7500(23)00135-8

4. The SOPHIA study weight trajectory prediction tool. Available at: https://bariatric-weight-

trajectory-prediction.univ-lille.fr/. Accessed February 1, 2025.

Elizabeth M. Hechenbleikner, MD, FACS, FASMBS

Emory University School of Medicine, Department of Surgery 

550 Peachtree Street Northeast

Medical Office Tower, 7th Floor

Atlanta, GA 30308 (USA)

404-778-3712

ehechen@emory.edu