Effective disease diagnosis is a significant unmet need on a global scale. The development of a model for early diagnosis and efficient treatment faces enormous hurdles due to the complexity of the many disease mechanisms and underlying symptoms of the patient population. In recent times, various ensemble-based ML models are assisting doctors with early diagnosis. But one of the biggest challenges for these models is the learning selection of diversity-based classifiers to enhance the model’s performance. Many researchers are attempted to improve classification accuracy with ensemble learning approaches but they failed. To improve the disease diagnosis performance in this study, a novel diversity-based evolutionary ensemble framework with a genetic algorithm is proposed. To improve the predictive performance we have used four diversity-based classifiers such as K-nearest neighbour, support vector machine, logistic regression, and decision tree using the bootstrapped approach to generate 20 diverse base learners with five bootstrapped bags using fivefold cross-validation. And also to improve the predictive performance, we have proposed a novel fitness function. To test the robustness, the model was run 20 times, and the average performance and average ensemble complexity of the proposed model were computed. Finally, our proposed model tested with four bench-mark disease datasets such as Pima Indian Diabetes (PID), Chronic Kidney Disease (CKD), Statlog Heart Data (SHD), and Wisconsin Breast Cancer (WBC). These results are compared with state-of-the-art ensemble models and non-ensemble models. Results demonstrated that in comparison with ensemble models, the proposed model performance is superior in terms of accuracy, AUC, and specificity on PID, accuracy, sensitivity, and specificity on CKD, accuracy, AUC, specificity on SHD, and accuracy, AUC, and specificity on WBC. In comparison with non-ensemble models, the proposed model performance is best in terms of accuracy, precision on PID, accuracy, AUC, precision and F1-score on CKD, accuracy, AUC, sensitivity, precision, and F1-score on SHD and accuracy, AUC on WBC.