A Design-Based Comprehensive Digital Twin Framework for Intelligent Detection

Chikungunya (CGN) is a rapidly expanding mosquito-borne viral disease which causes a significant burden of disease in global health and socio-economics. The issues of delayed detection, symptomatic overlap with other arboviruses and lack of capability to integrate real time patient information with macro-environments are significant challenges for traditional reactive public health systems. The present paper proposes a novel proactive, IoT-enabled Digital Twin (DTW) intelligent healthcare monitoring framework to enhance the continuous monitoring and early outbreak forecasting. It combines the telemetry of real-time vital sign monitoring from Internet of Medical Things (IoMT) with the ecological monitoring of vector habitats by sensors. Unique temporal granulation techniques are applied to the data to optimize processing times: Within Attribute Granulation (WAG), Between Attribute Granulation (BAG), and Cross Dataset Granulation (CDG). A pipeline for unsupervised classification of patient cohorts into functional infectious or non-infectious classes is based on the similarity of clinical features and implemented using a k-Means Clustering (k-MC) procedure. The pipeline for unsupervised classification of patient cohorts into functional class of infectious or non-infectious is based on the similarity of clinical features and implemented through a k-Means Clustering (k-MC) procedure. Prognostic modeling relies on Patient Severity State (PSS) vector that is evaluated by combining a Bayesian Belief Model (BBM) for uncertainty management with an Artificial Neural Network (ANN) powered by back propagation. The proposed dual-paradigm framework obtained an outstanding average classification Precision (93.97%) and a Sensitivity (93.57%) in the simulated test set and a 48% mean decrease in overall localized prognostic prediction errors compared to the traditional baseline systems simulating the rural agricultural region of India (N = 62,325).