The paper proposes a diagnostic system using random search algorithm and random forest model for heart failure prediction. The authors assert that their approach is efficient and shows better performance than the existing methods. The paper is well-written and easy to follow. However, the following changes should be carried out:

Suggested Major and Minor Changes

• The Introduction Section is too long. For this reason, it should be divided into two parts, namely “Introduction” and “Related Works”.

• Only a very old dataset was used in this study. The proposed approach should be tested on different datasets like “the Z-Alizadeh Sani dataset (2017)” to prove its efficiency.

• The classification accuracy comparison was only performed with the existing methods which have lower accuracy rates than the proposed method. In literature, there are many related studies that have higher accuracy values. In Table 5, there are just four studies from the last 2 years. Recent studies about heart disease prediction should be added.

• The performance comparison with the existing methods was realized for only classification accuracy metric. The results of the other evaluation metrics such as sensitivity, specificity and MCC should be added.

• The Conclusion Section should be expanded.

• Future works should be added.