Loading...
1 results
Search Results
Now showing 1 - 1 of 1
- Machine learning classification of consumption habits of creatine supplements in GYM goersPublication . Magalhães, Patrícia C.; Encarnação, Samuel; Schneider, Andre; Miguel Forte, Pedro; de Araújo Teixeria, José Eduardo; Monteiro, A. M.; Barbosa, Tiago M.; Pereira, Ana M.; ENCARNACAO, SAMUEL GONCALVES ALMEIDAThe aim is to identify usage patterns and the main factors that influence creatine supplementation, providing a basis for future educational interventions and recommendations for safe and effective use. The study was applied to gym goers in Bragança, where a QR code for a survey was released. 158 people participated, 65 non-consumers of creatine supplementation (37.34% men; 22.78% women) and 95 consumers (15.19% men; 24.68% women). Five machine learning algorithms were implemented to classify creatine consumption in gym goers: Logistic Regression, Gradient Boosting Classifier, Ada Boost Classifier, Xgboost Classifier. K-folds cross-validation was implemented to validate the machine learning performance. There was an increased proportion of females with considered themselves not sufficiently informed about the creatine effects/side effects (22.2%) in comparison to males (8.47%), p=0.03. The AdaBoost classifier exposed the best overall performance (86%) in classifying overuse of creatine in gym goers based on their Smoke habits (r = 0.33), grams of creatine used per day (r = 0.50) and lack information about the side effects of creatine intake (r = -0.33). The K-folds method validates the results with very good performance (86%). In conclusion, the five machine learning methods employed well characterized the overuse of creatine in gym goers based on smoke habits, grams of creatine per day, and lack information about the side effects of creatine intake.