Smart Healthcare - IoT and Artificial Neural Network(ANN)

With the use of the Internet of Medical Things (IoMT), the growing accessibility of wearable devices, and the continual reduction in the cost of sensor-based technologies and green computing, internet-connected sensors allow users to track, analyze, and share welfare information employing a cloud-based platform. Artificial Neural Networks (ANN) find extensive use across various applications, including healthcare systems, drawing inspiration from the biological concept of neurons. This study explores the prediction of heart attack risks using a combination of Internet of Things (IoT) sensors and artificial neural networks (ANN). Through the integration of IoT sensors such as the pulse sensor and LM35 temperature sensor, essential physiological data related to heart rate and body temperature was collected. An ANN model was developed, focusing on predictive capabilities using the sigmoid function, feedforward and backpropagation techniques, and random weight initialization. The investigation further utilized Keras with TensorFlow and optimized model parameters using Adam Optimization and Later with One Hot Encoding plus a Genetic Algorithm (GA). Additionally, the study involved training neural networks on higher-quality UCI datasets to evaluate performance. Scatterplots were generated to visualize relationships between variables. Results demonstrated promising performance improvements, with the model achieving close to ${9 0 \%}$ accuracy on the testing dataset. Encoding techniques were employed to handle categorical variables effectively, contributing to the model’s robustness and generalization. Overall, this research offers valuable insights into leveraging IoT sensors and ANN for heart attack risk prediction.