This short note describes a microcomputer-based system that can quickly detect certain heart conditions from the output of an ECG monitor. Its major virtues are its small size (about 16K bytes) and quick analysis (on the order of milliseconds). Its weaknesses are that it only checks for a small set of conditions, and that these checks are dependent on various threshold values that must be determined for each individual.
The system functions by picking out eight features, some coded by one bit (for example, Q wave normal or abnormal) and some by two bits (for example, if the number of P peaks is equal to, greater than, or less than the number of R peaks). These features are coded as a set of 14 bits, and these bits are then used to indicate one of 19 conditions. For example, bit 3 is set to 0 if there is an abnormal Q wave, and this is taken as an indication of myocardial infarction. As another example, if bits 11 to 14 are set to 1010 this is taken as an indication of ischemic ventricular hypertrophy.
The authors have tested their system on simulated ECG signals and they state that it correctly determined each of the 19 conditions. One wonders how well the system would work in a real clinical setting.