Analyzing Classification Models: Random Forest vs. Neural Networks in Health Prediction

Healthcare prediction has been a significant factor in saving human lives in recent years. In healthcare, intelligent systems are rapidly developed to analyze complicated relationships among data and transform them into accurate information for use in the prediction process. Consequently, artificial intelligence is rapidly transforming the healthcare industry. This project aims to forecast health conditions using classification models such as Random Forest in Google Colab and Neural Network in MATLAB based on gathered health parameters. The dataset was obtained from Kaggle. Constructing a Neural Network with MATLAB using ‘nprtool’ utilizing an 80:10:10 split for training, validation, and testing, while implementing a Random Forest in Google Colab with an 80:20 split for training and testing. These models were meticulously constructed and assessed to predict health conditions based on vital sign data. The achieved accuracies were 96.8% for the Random Forest and 87.8% for the Neural Network. The evaluation of model efficacy involved using performance and confusion matrix analyses to gauge the effectiveness of both Random Forest and Neural Network classifiers. The evaluation of the application’s performance is primarily focused on forecasting accuracy. Superior performance was observed with optimal layer and tree size set at 20 and 50, respectively, surpassing the effectiveness of alternative configurations.