Optimization of Resource Management based on Long Short-Term Memory and Inverse Reinforcement Learning Model for Patient Health Tracking

From past few years, the cloud computing has enabled the healthcare sector to optimize resource allocation and utilization by integrating modern technology into digital applications. However, traditional healthcare monitoring environments often struggle with effective resource management, having a limited impact on activity observation and patient health tracking on a regular basis. To overcome these challenges, this research proposes Long Short-Term Memory and Inverse Reinforcement Learning (LSTM-IRL) model to optimize resource allocation, improves activity observation and enhance patient health tracking. Initially, data is collected from Healthcare-Internet of Things (H-IoT) dataset which simulates sensor data from wearable devices. This data preprocessed with Wavelet Denoising technique by removing noise from signals through wavelet transforms, improves signal quality. Next, Short-Time Fourier Transform (STFT) is used to extract time-frequency representation of signal, allowing for capture both spectral and temporal features. After that, LSTM networks are employed to predict resource allocation, while IRL algorithm is utilized to learn reward function and optimal policy for patient health tracking. This integrated approach considers both clinical outcomes and resource Management for optimized patient healthcare. The proposed LSTM-IRL model obtained better results when compared with exiting Deep Reinforcement Learning (DRL) method in terms of accuracy (92.5%) and F1-score (94.7%).