Digital Twin Modeling and Graph Neural Networks for Predictive Analytics in Remote Healthcare Monitoring Platforms

The Healthcare Monitors is a new predictive analytics model that has Digital Twin Modeling and Graph Neural Networks (GNNs) to assist in remote healthcare monitoring. Existing methods are usually inadequate when handling time-dependent health data and patient-equipment-symptom inter-relationship, decreasing prediction accuracy and personalization. To overcome, a new hybrid which combines Digital Twin profiles with Informer-based Transformers and Graph Attention Networks to model strong in terms of time and relational. Healthcare Iot Dataset from Kaggle (heart rate and oxygen levels) is used to train and test. The methodology in Python uses a combination of graph and time-series to predict patient degradation and risk of alertness in real-time with an accuracy of 93%. Prediction of adverse events has been very precise and this is evidence that the system is reliable. The patient, clinician and healthcare practitioner are the beneficiaries of the methodology in terms of timely interventions and improvement in decision support.

Перевод пока недоступен