Efficient IoT-machine learning assisted heart failure prediction using adaptive fuzzy-based LSTM-RNN algorithm

Abstract

Chronic diseases like diabetes, Heart Failure (HF), malignancy, and severe respiratory sickness are the leading cause of mortality around the globe. Dissimilar indications or traits are extremely difficult to identify in HF patients. IoT solutions are becoming increasingly commonplace as smart wearable gadgets become more popular. Sudden heart attacks have a short life expectancy, which is terrible. As a result, a patient monitoring of heart patients based on IoT-centered Machine Learning (ML) is presented to help with HF prediction, and treatment is administered as necessary. Verification, Encryption, and Categorization are the three phases that make up this developed model. Initially, the datasets from the IoT sensor gadget are gathered by authenticating with a specific hospital through encryption. The patient’s integrated IoT sensor module then transfers sensing information to the cloud. The Improved Blowfish Encryption (IBE) approach is used to protect the sensor data transfer to the cloud. Then the encrypted data is decrypted, and the classification is performed using the Adaptive Fuzzy-Based Long Short-Term Memory with Recurrent Neural Network (AF-LSTM-RNN) algorithm. The results are classed as malignant or benign. It assesses the patient’s cardiac state and sends an alert text to the doctor for treatment. The AF-LSTM-RNN-based HF prediction outperforms the existing techniques. Accuracy, sensitivity, specificity, precision, F-measure and Matthews Correlation Coefficient (MCC) are compared to existing procedures to ensure the planned research is genuine. Using the Origin tool, these metrics are shown as research findings.