Toward accessible fitness assessment: estimating VO ₂ max in youth using intelligent hybrid optimization non-exercise modeling

max, is essential for youth health, but conventional testing is labor-intensive. This study proposes a hybrid Neuro-Fuzzy Inference System (ANFIS), optimized using Rider Optimization Algorithm (ROA), Electromagnetic Field Optimization (EFO), and Sharpness-Aware Minimization (SAM), for scalable prediction. The procedure involved preprocessing demographic and activity data, splitting the dataset into 80% training and 20% testing sets, and optimizing ANFIS hyperparameters using different optimizers. The models showed strong predictive accuracy (R² = 0.9789-0.9860), with EFO-ANFIS achieving the best overall performance and lowest prediction bias. The study presents an interpretable and scalable tool for predicting youth cardiorespiratory fitness.