How a human ranks the ECG diagnostic parameters: The pursuit of experts’ preferences based on a hidden poll

The paper presents the methodology and results of the research of human expert preferences for particular diagnostic results correlated with the actual patient disease. We applied a custom-build software recording the expert interaction with the computer while he is gathering together all necessary diagnostic statements into a final report. The module works behind the standard ECG interpretation software and recorded 1730 human-verified examinations in 11 cardiologist offices. The statistical processing of these records yield justified parameter hierarchies for 12 most common heart diseases and approximate preferences for other 17. The hierarchy of diagnostic outcome shows his utility in prioritized non-uniform transmission in wireless monitoring networks and in disease-oriented assessment of quality of manual and automated ECG interpretation.