Emerging cloud-based predictive maintenance for hemodialysis reverse osmosis purified water

Abstract

A hemodialysis machine using filtered water from a reverse osmosis (RO) system purifies the blood of patients with chronic kidney disease (CKD) in order to balance critical minerals for a normal life. The Biomedical Engineering Maintenance Services (BEMS) at the Ministry of Health (MoH) contract hospital ensure the integrity of the RO system through a systematic procedure and key performance indicators (KPIs) that are supervised by Biomedical Engineers' strict and timely monitoring. Nonetheless, the existing maintenance strategy is insufficient to predict actual degradation behaviours and does not focus on benchmarking a single KPI to ensure the overall system's fitness level. In addition, the derived historical trend of water pressures, water flow rates, and water conductivity level, all of which define the health or effectiveness of the RO system, are monitored using analogue gauges, which have proven ineffective in terms of cost and personnel. This paper highlighted the simulated model of an automated fault diagnosis prediction model based on machine learning (ML) and sensors to detect anomalies in the RO system before a breakdown occurs. The goal of this model is to increase the performance of important operations so that they can run smoothly. Seeing as the failure of the equipment may be predicted, the right time for maintenance work can be scheduled ahead of time before a real breakdown occurs. This is significant as the findings of this paper allow for a transparent assessment of the RO system's and other prospective medical devices' health in order to improve the uptime guarantee. In a nutshell, this paper's fundamental contribution is its capacity to reduce superfluous maintenance routines and optimise the cost of RO system maintenance.