This research introduces a comprehensive remote health monitoring platform designed for patients with limited mobility, focusing on enhancing clinical management and improving recovery outcomes. Bedridden patients and wheelchair users face significant health risks such as respiratory complications, pressure ulcers, muscle atrophy, and decreased bone density due to prolonged immobility. These conditions not only diminish quality of life but also hinder recovery after hospital discharge. Our innovative solution employs non-invasive infrared thermography to monitor respiratory parameters and ergonomic data, an approach that ensures patient dignity and confidentiality. An Ensemble Machine Learning model analyzes respiratory patterns and body temperature variation, enabling healthcare providers to detect early signs of respiratory deterioration, hyperpyrexia or hypothermia, facilitating timely interventions that can substantially reduce severe complications. Additionally, the Machine Learning model is trained to (i) assist caregivers in routine tasks such as patient repositioning, which is critical in preventing pressure ulcers and associated infections and (ii) detect falls, patient wandering and escape attempts. An online dashboard and reporting system serves as the endpoint of our pipeline, providing risk alerts, optimal repositioning schedules, and access to historical and statistical data for healthcare professionals. Data for this study was collected through experiments with a diverse group of over 50 volunteers and health professionals, offering a solid foundation for our statistical analysis and methodology validation.