June 13, 2025
Day 14- Time Domain Feature Extraction
What I Learned
Today at my internship, I continued advancing our project focused on ECG signal analysis for cardiovascular disease diagnosis. Building on the momentum from earlier in the week, I deepened my understanding of the time domain features of ECG signals — such as R-R intervals and signal amplitude fluctuations — which are vital for identifying arrhythmias and other abnormalities. We kicked off the day by ensuring our dataset was in top condition for feature extraction. After completing a prior cleanup, we finalized a filtered dataset of 21,388 rows, successfully eliminating all rows with NaN diagnostic labels. This was an important step to ensure data integrity before applying any signal processing techniques.
Blockers
No Blockers
Reflection
Today was one of those days that reminded me how much depth there is to data science — especially in the context of medical signals like ECG. Coming into the lab, I knew we were going to continue from where we left off, but I didn’t expect how much clarity I’d gain around something as abstract as time versus frequency domains. The day started with a sense of momentum. We had just cleaned up our dataset, and now we finally had a clean version with over 21,000 usable rows — no empty diagnostic labels, no NaNs. That alone felt like a small win, especially given how long we’ve been iterating on dataset preparation.