Patrick Saux, MSc, Pierre Bauvin, PhD, Violeta Raverdy, MD, Julien Teigny, BEng, Hélène Verkindt, MD, Tomy Soumphonphakdy, MSc, Maxence Debert, MSc, Anne Jacobs, MSc, Daan Jacobs, BSc, Valerie Monpellier, PhD, Phong Ching Lee, MBChB, Chin Hong Lim, FRCS, Johanna C Andersson-Assarsson, PhD, Lena Carlsson, ProfMD, Per-Arne Svensson, ProfPhD, Florence Galtier, MD, Guelareh Dezfoulian, MD, Mihaela Moldovanu, MD, Severine Andrieux, MD, Julien Couster, MD, Marie Lepage, MD, Erminia Lembo, MD, Ornella Verrastro, PhD, Maud Robert, ProfMD PhD, Paulina Salminen, ProfMD PhD, Geltrude Mingrone, ProfMD PhD, Ralph Peterli, MD, Ricardo V Cohen, MD PhD, Carlos Zerrweck, ProfMD, David Nocca, ProfMD PhD, Carel W Le Roux, ProfMD PhD, Robert Caiazzo, ProfMD PhD, Philippe Preux, ProfPhD, and François Pattou, ProfMD PhD
Summary: Background: Weight loss trajectories after bariatric surgery vary widely between individuals, and predicting weight loss before the operation remains challenging. We aimed to develop a model using machine learning to provide individual preoperative prediction of 5-year weight loss trajectories after surgery. Methods: In this multinational retrospective observational study we enrolled adult participants (aged ≥18 years) from ten prospective cohorts (including ABOS [NCT01129297], BAREVAL [NCT02310178], the Swedish Obese Subjects study, and a large cohort from the Dutch Obesity Clinic [Nederlandse Obesitas Kliniek]) and two randomised trials (SleevePass [NCT00793143] and SM-BOSS [NCT00356213]) in Europe, the Americas, and Asia, with a 5 year follow-up after Roux-en-Y gastric bypass, sleeve gastrectomy, or gastric band. Patients with a previous history of bariatric surgery or large delays between scheduled and actual visits were excluded. The training cohort comprised patients from two centres in France (ABOS and BAREVAL). The primary outcome was BMI at 5 years. A model was developed using least absolute shrinkage and selection operator to select variables and the classification and regression trees algorithm to build interpretable regression trees. The performances of the model were assessed through the median absolute deviation (MAD) and root mean squared error (RMSE) of BMI. Findings: 10 231 patients from 12 centres in ten countries were included in the analysis, corresponding to 30 602 patient-years. Among participants in all 12 cohorts, 7701 (75·3%) were female, 2530 (24·7%) were male. Among 434 baseline attributes available in the training cohort, seven variables were selected: height, weight, intervention type, age, diabetes status, diabetes duration, and smoking status. At 5 years, across external testing cohorts the overall mean MAD BMI was 2·8 kg/m2 (95% CI 2·6–3·0) and mean RMSE BMI was 4·7 kg/m2 (4·4–5·0), and the mean difference between predicted and observed BMI was –0·3 kg/m2 (SD 4·7). This model is incorporated in an easy to use and interpretable web-based prediction tool to help inform clinical decision before surgery. Interpretation: We developed a machine learning-based model, which is internationally validated, for predicting individual 5-year weight loss trajectories after three common bariatric interventions. Funding: SOPHIA Innovative Medicines Initiative 2 Joint Undertaking, supported by the EU's Horizon 2020 research and innovation programme, the European Federation of Pharmaceutical Industries and Associations, Type 1 Diabetes Exchange, and the Juvenile Diabetes Research Foundation and Obesity Action Coalition; Métropole Européenne de Lille; Agence Nationale de la Recherche; Institut national de recherche en sciences et technologies du numérique through the Artificial Intelligence chair Apprenf; Université de Lille Nord Europe's I-SITE EXPAND as part of the Bandits For Health project; Laboratoire d’excellence European Genomic Institute for Diabetes; Soutien aux Travaux Interdisciplinaires, Multi-établissements et Exploratoires programme by Conseil Régional Hauts-de-France (volet partenarial phase 2, project PERSO-SURG).