The study developed and validated four models to predict airway hyperresponsiveness (AHR) in patients with suspected asthma, with model C being the best with AIC = 310.44. Model C includes the parameters FEV1/FVC%, MEF75%, PEF% and MMEF75-25%, which were identified as the most effective predictors of AHR using the LASSO method with 10-fold cross-validation. In the training cohort, the model achieved AUC = 0.790 at a cut-off of 0.354 (95% CI: 0.724–0.760) and in the validation cohort AUC = 0.756 at a cut-off of 0.404 (95% CI: 0.600–0.814). The model shows a good calibration with a curve close to a straight line with a slope of approximately 1, confirming the agreement between the predicted and the actual probability. Integrated Discrimination Improvement (IDI) and Decision Curve Analysis (DCA) show that Model C has a higher net benefit over extreme curves. This is the first study using machine learning to combine indices of small airway function, FEV1 and PEF to predict AHR. The model offers potential as a visual tool for early detection and standardized treatment in patients with suspected asthma.[1][2][5]