Week 5 - Alex

Over the last week, I’ve split my time between the EP lab and doing background research on the clinical research project: Use of Adenosine to Determine the Electrophysiological Mechanism of Premature Ventricular Contractions.

My experience in the EP lab provides an interesting motivation and perspective for the type of research that we do in the Christini lab. The most interesting part of the procedures is when the clinicians are searching for the source of an arrhythmia. It’s usually the most time-varying and technically challenging part of the procedure. EPs typically populate voltage and conductance maps using a catheter and use 15-20 leads to make a decision where to ablate. 

Ablations don’t always terminate an arrhythmia and often lead to new arrhythmias. This is why ablation region prediction is one of the most studied areas in cardiac electrophysiology. If an algorithm could better predict where to ablate, electrophysiologists would spend less time searching where to ablate, ablate more precisely, and save time mapping the affected tissue.

In an attempt to get a jump start on the clinical research project, I dug into the literature on PVC diagnosis and treatment. I've been focusing my background research on three areas, and have a few questions that I will ask at my next meeting with Dr. Ip:

Adenosine effects on PVCs 

• Question: Do we know the response (% increase, decrease, no change) to adenosine administration during EP study for all of the patients who wore Holters?
• Question: Is it possible to get the 12 lead and any other ECG data from the EP study for these patients?
• Question: Are we going to have access to the coupling interval of every PVC-sinus beat for each patient? There is a paper which posits that coupling interval variability is a good estimate of the PVC's mechanism.

Automated PVC classification from ECG 

• Beyond looking at the cycle length, and distinguishing between two patient groups, I would like to use a learning algorithm, for further classifying PVCs. It would be great if we could classify patients from the PVC study.
• Question: What other predictive information would be useful from an algorithm that analyzed PVCs?

Automated ablation region recommendation 

• I know this is a saturated space, but have developed an interest in the methods electrophysiologists use to efficiently 1) find regions to ablate and 2) terminate an arrhythmia. After viewing a few ablation procedures, I need insight on the types of arrhythmias that require the most mapping.

This week, my plan is to implement a machine learning algorithm that interprets EKG signals, with the eventual application to our PVC problem.