Predictive Health Monitoring of Multimorbidity: A Simulation Experiment Using OPNET Modeller

Abstract

Primary health care services are increasingly at the heart of integrated people-centred. They provide an entry point into the health system, ongoing care coordination and a person focused approach. Patient safety incidents involve active events, such as adverse drug events, intervention complications, infections and care failures, prescribing and delayed diagnosis. As a result, patients with multimorbidity are at higher risk of safety issues. In this study, monitoring is conducted on data streams from wearable body sensors enabled by Healthcare Internet of Things, overlapping time series plotting of body parameters, and a rules-based decision engine with artificial intelligence capabilities to detect risky patterns of variations in the body parameters. This research investigates a possible design for predictive health monitoring of multimorbidity that individual hospitals can implement at low costs and high operations effectiveness and feasibility. This research was conducted in OPNET modeller to: (a) Conduct simulation-based testing of a low-level Healthcare 4.0 design for monitoring of multimorbidity; (b) Suggest practical design, operational details, and limitations for predictive health monitoring of multimorbidity. The design details were captured from literature and the simulated design was created for monitoring 105 health sensors by a single hospital in its vicinity of 10 Kilometers radius. The design comprised of edge computing servers in the hospital’s premises connected to a fog computing system for running body area networks of body-attached sensors of 100 individuals. The connections between the edge computing servers and the fog computing system were 100 Mbps broadband Internet connections on Optical Fiber. The simulation was conducted to study effects of mobility of the citizens, effects of TCP session and transmission delays, effects of buffering in the WLAN routers, and effects of TCP congestion windows. The simulation pattern was defined by an operations algorithm designed with the help of technical literature review. The WLAN routers should have high capacity of buffer memories to temporarily store health data streams and large number of receiving channels to avoid information losses. The TCP session timeouts and congestions can be very harmful for individuals being monitored critically. Hence, bandwidth allocation to body area networks should be planned based on the frequency and amount of data needed by the hospitals. The body area networks deployed on individuals should have carefully conducted right sizing of number of parameters, their data volumes, transmission frequencies, and criticality of the individuals. The hospitals should classify patients in different classes based on criticality, data volumes, and frequencies of monitoring needed, and allocate appropriate multi-tiered body area networks for the different patient classes. A one size fit for all approach will fail.