gms | German Medical Science

Introduction: Medical monitoring devices in acute care typically combine multiple steps of data acquisition, conditioning, processing, and visualisation into an integrated package of hard- and software. Such fixed, proprietary packaging obstructs fast implementation of newly described derived parameters and their flexible translation from bench to bedside, with innovation cycles largely determined by hardware replacement cycles. Additionally, sufficiently long-term and readily accessible persistence for research and quality assurance is not provided by many manufactures. The AcuWave architecture [1] aims at decoupling the various steps of acquisition, storage, analysis, and visualisation of raw and derived data. Here, we describe our experiences in implementing the architectural concept and showcase its potential using Pulse Pressure Variation (PPV) analysis, a simple, provenly useful derived parameter unavailable in many current monitoring systems.

Methods: The AcuWave software consists of three main components: WaveDatabase (WaveDB), WaveServer, and WaveClient. The WaveDB implements a vendor-neutral, performance optimised SQL database for storing large quantities of high-resolution time series data. The WaveServer realises a plug-and-play enabled, modular execution concept for complex chains of time series data processing. Java was chosen as platform for its low entry barrier, proven stability in production-critical environments, and huge selection of existing code and development tools. Novel analyses are implemented as plugins that can trivially reuse existing persistence and processing modules. The WaveClient provides an easy-to-use graphical user interface in a touch-enabled, html5 web application optimised for instantaneous interactivity when handling large datasets through smart caching, downsampling, and preloading mechanisms. A proprietary, JSON-based RESTful protocol is used for all client-server interaction. The developer is provided with an unified interface for both research and clinical settings. To assess the translational potential of the infrastructure, a PPV analysis module was implemented and a cognitive walk-through based usability evaluation [2] conducted.

Results: The AcuWave software has successfully interfaced with various data sources ranging from archives such as PhysioNet MGH/MF Database [3] to patient recordings from Dräger IACS monitoring systems, where online data processing and visualisation with latencies, resulting from medical device interfaces, below 10 s are achieved. The PPV plugin was developed in 8 man-hours utilising common AcuWave interface templates and preprocessing modules such as beat detection. This plugin generated reproducible results in datasets from various sources. As part of the evaluation, ICU physicians rated daily usability with a median of 9/10 (Range 3-10, N=11) and provided valuable feedback about result presentation.

Discussion: The AcuWave project has evolved to a stage where computation and visualisation can be performed on data from multiple sources in the research setting. Furthermore, the system has been successfully employed for quality assurance in the clinical setting. The PPV implementation demonstrated rapid deployment of novel analyses with usability deemed suitable for routine use. Overall, AcuWave provides a modern toolchain to facilitate rapid translation from research in quantitative physiology to bedside application in acute care. For diagnostic or therapeutic intended use, regulatory challenges will likely require active industry participation and remain to be addressed.

Die Autoren geben an, dass kein Interessenkonflikt besteht.

Die Autoren geben an, dass ein positives Ethikvotum vorliegt.