A Neural Network Approach for Early Prediction of Heart Failure: A Machine Learning Study

Heart failure appears to be a serious medical illness which occurs due to insufficient pump of blood that the body needs. Heart failure is a common cause of death, thus it's critical to identify the risk factors and take preventative measures. This study uses a large dataset that was gathered from Kaggle to investigate the use of machine learning algorithms to predict heart failure. To create prediction models, the study uses well- known techniques. 13 major characteristics, including age, anemia, cholesterol, and Ejection fraction for 299 patients, make up the dataset. Pre- processing the dataset, addressing imbalances with methods like the Synthetic Minority Over-Sampling Technique, and validating the results using k-fold cross-validation (k=10) are all part of the methodology. The outcomes show how well each algorithm performs with regards to precision, accuracy, Recall and area under the curve. The decision tree has accuracy of 79.59%, Logistics Regression has 79.05%, Support Vector Machine has 79.58%, Random Forest has 81.76% while the accuracy of Neural Network is 86.44%. At 86.44% accuracy, the Neural Network classifier proves to be the most precise Machine learning models Neural Networks in particular show encouraging results when it comes to heart failure prediction. The study emphasizes how machine learning may be used in healthcare settings to improve patient outcomes, early diagnosis, and therapy planning.