Patients can have an elevated risk of major cardiac complications following non-cardiac surgery. Myocardial ischemia is a lack of oxygen in the heart tissue and may be a precursor to infarction and cardiac death. This ischemia can be detected in the ST segment of a patient's electrocardiogram (ECG) hours before lasting tissue damage occurs. The Perioperative Ischemia Reduction Studies (PROSE) seek to leverage early detection of ischemia to establish an intervention window to stop the ischemic

more »

... scade before it results in lasting tissue damage. Ambulatory ECG is used to monitor patients for PROSE; however due to the mobile nature of the recording, ambulatory ECG is subject to higher levels of noise contaminants. This can degrade the ST analysis used in the detection of ischemia, resulting in high false alarm rates that render ambulatory ischemia monitoring clinically impractical. This thesis investigates the possibility of modifying alarm outputs on the basis of ECG signal quality to increase positive predictive value (PPV) of alarms without detriment to existing sensitivity. First, this thesis presents methods to automatically quantify ECG signal quality in a single ECG lead via a signal quality index (SQI) and validates these methods using both non-ischemic and ischemic ECG data. This thesis then proposes and evaluates three system approaches to modifying alarms using this SQI. Resulting modified alarms reveal increased PPV from 0.41 to 0.85 while maintaining sensitivity. These results are encouraging and indicate that these methods could help provide for practical monitoring of myocardial ischemia via ambulatory ECG in the context of PROSE. iii