Blockchain-Enhanced Heart Rate Prediction: Comparing Machine Learning and Time Series Approaches in MIoT

This research investigates the preservation of heart rate data in Medical IoT (MIoT) systems using blockchain technology. The study evaluates several machine learning and time series models, including Decision Tree Regressor, Neural Network, LSTM, and ARIMA. The study compares these models’ prediction performance and ability to maintain data integrity in a blockchain-secured environment. The evaluation metric performance indicates towards Decision Tree Regressor (DTR) and Neural Network (NN) models which outperform in forcasting heart rate data with minimum MSE, MAE and more viable R2 and Explained Variance scores. These outcomes suggest great accuracy and reliability.In contrast, the LSTM and ARIMA models perform inadequately, displaying substantial error metrics and negative scores, highlighting that they are unsuited for this particular prediction job. Overall, the Decision Tree Regressor and Neural Network models are the most successful in forecasting heart rate in this study.