Data acquisition systems make it very easy to collect huge amounts of data from many sources at once. The real challenge is in data reduction: obtaining useful information from the mass of numbers. As an example, imagine digitizing a blood pressure trace from an arterial catheter at 50 Hz. The tracing would look something like Figure 5.1 and at a heart rate of 88 beats/min there are thirty-four data points per beat. What we really want is the mean, systolic and diastolic pressure for each beat, plus the instantaneous heart rate: four values per beat.

There are several ways to tackle such problems. At the ‘grass roots’ level you can write your own programs from scratch in a language such as C, Pascal or BASIC. The next step up is to use libraries of routines that are called by your program. There are libraries in the public domain but these are often older algorithms written in FORTRAN. LabWindows, from National Instruments, takes care of the user interface, graphics, file handling, and so forth whilst still giving you the flexibility of a low-level programming language. Higher-level ‘languages’ like Matlab, Gauss, Igor, LabView, and HiQ perform all the housekeeping behind the scenes and allow you to concentrate on the data analysis whilst still retaining some programming structure. Higher still are complete acquisition and analysis packages (Acqknowledge, SuperScope) that are designed for people with very little programming skill. They each have their own environment for data analysis and provide a fairly comprehensive library of functions.