Prediction of successful weight reduction after laparoscopic adjustable gastric banding.

BACKGROUND/AIM: Compared with conventional pharmacological therapies, bariatric surgery has been shown to cause greater and- sustained weight loss. It was aimed to evaluate weight loss in obese patients after laparoscopic adjustable gastric banding surgery using information typically available during the initial evaluation studied before bariatric surgery and genes. METHODOLOGY: 74 patients undergoing laparoscopic adjustable gastric banding (LAGB) were enrolled. Artificial Neural Network technology was used to predict weight loss. RESULTS: We studied 74 patients consisting of 22 men and 52 women 2 years after operation. Mean age was 31.7 +/- 9.1 years. 27 (36.5%) patients had successful weight reduction (excess weight loss >50%) while 47 (63.5%) did not. ANN provided predicted factors on gender, insulin, albumin and two genes: re4684846_r, rs660339_r which were associated with success. CONCLUSION: Artificial neural network is a better modeling technique and the predictive accuracy is higher on the basis of multiple variables related to laboratory tests. Our finding gave demonstrated result that obese patients of successful weight reduction after laparoscopic adjustable gastric banding surgery were women, having little lower insulin and albumin, and carrying GG genotype on rs4684846 and with at least one T allele on rs660339. In these cases, weight loss will give better results.

Obesity and the decision tree: predictors of sustained weight loss after bariatric surgery.

BACKGROUND/AIMS: Bariatric surgery is the only long-lasting effective treatment to reduce body weight in morbid obesity. Previous literature in using data mining techniques to predict weight loss in obese patients who have undergone bariatric surgery is limited. This study used initial evaluations before bariatric surgery and data mining techniques to predict weight outcomes in morbidly obese patients seeking surgical treatment. METHODOLOGY: 251 morbidly obese patients undergoing laparoscopic mini-gastric bypass (LMGB) or adjustable gastric banding (LAGB) with complete clinical data at baseline and at two years were enrolled for analysis. Decision Tree, Logistic Regression and Discriminant analysis technologies were used to predict weight loss. Overall classification capability of the designed diagnostic models was evaluated by the misclassification costs. RESULTS: Two hundred fifty-one patients consisting of 68 men and 183 women was studied; with mean age 33 years. Mean +/- SD weight loss at 2 year was 74.5 +/- 16.4 kg. During two years of follow up, two-hundred and five (81.7%) patients had successful weight reduction while 46 (18.3%) were failed to reduce body weight. Operation methods, alanine transaminase (ALT), aspartate transaminase (AST), white blood cell counts (WBC), insulin and hemoglobin A1c (HbA1c) levels were the predictive factors for successful weight reduction. CONCLUSION: Decision tree model was a better classification models than traditional logistic regression and discriminant analysis in view of predictive accuracies.