Optimized DenseNet Model for Heart Disease Prediction from Medical Health Records

One of the most well-known and fatal illnesses in the world is Heart Disease (HD) claims the lives of countless people each year. To save lives, this disease must be identified early. Initial identification enables rapid intervention and ongoing clinical monitoring, both of which are crucial but frequently constrained by the incapacity of medical personnel to continuously supervise patients. Specialists can lower fatality rates by detecting cardiac issues early and keeping a close eye on their patients. HD detection is not always reliable, and doctors cannot be in regular contact with their patients. Machine Learning (ML) techniques offer an intriguing method to identifying risk factors because they offer a stronger basis for predicting and making decisions based on data supplied by medical organizations. In order to make accurate predictions about HD, this research work presents a Deep Learning (DL) model that utilizes the DenseNet (DN) architectures. Moreover, this research focused on developing the prototype model of patient likelihood in diagnosing the HD with the availability of medical records using DN model. The dataset of 70,000 cases from Kaggle is used to test the suggested DN model. The model's accuracy is greatly influenced by the DN technique and the features in the dataset that were employed to train the model. The suggested DN model achieves an accuracy of 91.86% over 150 epochs using batch size 1, surpassing all current models.