Randomized‐Grid Search for Time‐Efficient Hyperparameter Tuning in Cardiovascular Disease Classification

Abstract

Cardiovascular diseases (CVDs) are serious illnesses of the heart, which require accurate diagnosis to prevent fatal consequences. Hyperparameter tuning plays a critical role in optimizing machine learning (ML) model performance by selecting suitable parameter configurations for improved accuracy and generalization. Grid search systematically evaluates predefined hyperparameter combinations, whereas random search selects configurations randomly, enabling broader exploration at reduced computational cost. An efficient tuning strategy is therefore essential when developing classification models where time is as important as predictive capability. In this work, we propose a new hyperparameter tuning approach to tune the hyperparameters of ML models for CVD classification. It combines the global exploration of random search with the focused, exhaustive refinement of grid search in the most promising regions, balancing exploration, and exploitation to yield a robust and time‐efficient model. We evaluate the method on two datasets, the UCI Heart Disease and Framingham Heart Study datasets, across eight classical, and ensemble classifiers. Experimental results show that Randomized‐Grid Search outperforms traditional tuning methods on both datasets while substantially reducing training time for most models. These findings emphasize the predictive performance and computational efficiency of the proposed method, underlining its potential to advance ML applications for timely and accurate CVD diagnosis.