Core software for on-line data acquisition and post-hoc analysis: report of clinical testing and evaluation
© Current Science Ltd 1998
Published: 1 March 1998
The continuing progress of standard medical care results in an overload of patient's data, which are difficult to handle by traditional techniques. A software package is perspectively an appealing choice to deal with such a challenge. Aim of this study was the development and testing of an automated real-time data acquisition and analysis system for intensive care survey. The system included two PCs, the first for data acquisition (PC1: 166 MHz Pentium CPU, 32 MB RAM, 2 GB HDD, A/DC National Instruments AT-MIO 16E-10), and the second for data analysis (PC2: 200 MHz Pentium CPU, 32 MB RAM, 2 GB HDD). PCs were both networked, with shared HDDs, keyboard, mouse, video output, and printout subsystem. Complete hemodynamic, respiratory and neurological parameters were monitored by HP Component Monitoring System 66S/68S, which included a VueLink interface to coupled ventilators and Licox GMS, Serial Output (RS-232, 38400 baud per s) and up to 8 waveform analog outputs. Original software for data acquisition, storage and analysis was developed by object-oriented LabVIEW 4.0. Data acquisition was performed by PC1. At start, the user is allowed to select up to 51 numerical parameters (serial input RS 232, 4 data strings per min) and up to 16 waveforms (analog input, sampling rate: 128–256–512 points per s). Flag points (free-text or codified) could be added. As an example, it was possible to display the waveforms, up to 5 trend graphs or 6 histograms of selected digital parameters. Up to 3 graphic trends from selected digital parameters were automatically printed on demand (every 6–8–12–24 h). The complete sets of numerical parameters were stored as following: a) read-only spreadsheet files, whose length was limited to 2 h each one because of technical reasons; b) daily spreadsheet files (1st data string of minute); c) binary files (waveform storage) with circular buffering in 48 h cycles (max. length). Post-hoc analysis was performed by PC2, regardless of eventual simultaneous data acquisition by PC1. When system starts, the user chooses the optional on-line or off-line work (from already stored data). The task can involve up to 8 digital parameters (trends) and up to 4 waveforms. Different software tools like cursors and pointers were available to retrieve significant events, to select time intervals and to zoom the graphs, allowing to perform different procedures. Analysis capabilities on numerical parameters (trends) included: frequency tables of selected parameters, numerical regressions, vector calculus. Input data from daily spreadsheet files could be edited. Waveform analysis included basic statistics, integral and derivative calculus. Different reports, graphs and tables were printed at request. The assessment of our system included a preliminary laboratory test and clinical tests. The system proved an easy interface to different equipment. The user interface and response times were favorably accepted. Data storage and display were reliable and flexible. The equipment can be located at patient bedside as conveniently as at a ward desk. The use of 2 CPUs linked in a local network allowed effective simultaneous data acquirement and analysis. Other features included highly configurable data acquisition, to suit the case at hand, easy and fast retrieval of information, automatic generation of complete and accurate reports to illustrate trends or to correlate different parameters. More extensive field tests should be carried out to assess the impact of the system to ICU and operating rooms and its effectiveness.