Heart Attack Criticality Stage Prediction: A Deep Learning-based Model for Enhanced Accuracy

This paper deals with the development of a model to predict heart attack criticality stages and its performance in terms of various metrics. The model performs very well for all stages of heart attack severity. For Criticality Stage 1, the Precision, F1-Score and Recall are all 98.21%, which is a reliable classification. The model's total accuracy, as trained using a collection of 10,680 photos, is 98.31%. Criticality Stage 2 improves slightly with a Precision, Recall, and F1-Score of 98.31%. Best performance is found in Criticality Stage 3 with values of 98.40% for all three metrics, which proves the consistency and accuracy with which the model identifies the criticality of heart attacks. Besides that, the accuracy of the model remains at 99% in all the stages, therefore, reinforcing the reliability of heart attack classification. Support values of each stage - 3360 for Stage 1, 3560 for Stage 2, and 3760 for Stage 3 indicate a very well-balanced dataset contributing to the robust performance of the model. The high support proportions at 0.31, 0.33, and 0.35 of each respective stage also emphasize a balanced representation of each class in the dataset. The overall accuracy level with future applications in heart attack diagnosis improvement and severity prediction will be achieved through this model. The approach could, therefore, be applied in a clinical setting to aid health practitioners in reaching timely and informed decisions regarding their patients' treatment.