Predicting Medical Device Life Expectancy and Estimating Remaining Useful Life Using a Data-Driven Multimodal Framework

Accurately predicting the life expectancy of medical devices is crucial in optimizing healthcare operations, managing costs, and ensuring patient safety. Medical devices in clinical environments must be maintained, replaced, or refurbished on time to prevent malfunctions that could compromise patient care. The prediction of the Remaining Useful Life (RUL) of these devices enables improved resource utilization and reduced downtime. This study develops a predictive model to predict the medical device lifespan while estimating its RUL to prevent unexpected failures that could disrupt essential healthcare services. A data-driven multimodal framework is proposed to estimate the RUL, particularly for devices classified as Beyond Economic Repair (BER). This innovative framework integrates unstructured information from maintenance reports with structured device data, such as usage history, manufacturer specifications, and operational conditions, to create a comprehensive dataset that enhances predictive accuracy. Findings indicate that an ensemble learning technique demonstrates exceptional performance, achieving a Root Mean Square Error (RMSE) of 0.1563 and an R-squared (R²) value of 0.9749. Thus, by implementing data-driven maintenance and replacement strategies derived from these findings, healthcare organizations can improve operational efficiency, reduce costs, and ensure patient safety. This proactive approach allows healthcare institutions to allocate resources more effectively, prioritize maintenance for devices at higher risk of failure, and ultimately enhance the reliability of medical equipment in clinical settings.