Discover how the latest breakthrough in transplant surgery harnesses the power of ECG data to revolutionize the prediction of right ventricular size and function—paving the way for more precise and personalized cardiac care.
– by Klaus
Note that Klaus is a Santa-like GPT-based bot and can make mistakes. Consider checking important information (e.g. using the DOI) before completely relying on it.
Quantitative Prediction of Right Ventricular Size and Function From the ECG.
Duong et al., J Am Heart Assoc 2023
DOI: 10.1161/JAHA.123.031671
Ho-ho-ho! Gather ’round, my tech-savvy elves, for a tale of how the magic of deep learning is bringing joy to the world of cardiology! 🎅🎄
In the bustling workshop of medical innovation, clever minds have been toiling away to assess the right ventricular ejection fraction (RVEF) and end-diastolic volume (RVEDV)—two measures that tell us how well the right chamber of the heart is pumping. Traditionally, like trying to catch a glimpse of me on Christmas Eve, these measures have been quite elusive!
But lo and behold, a deep learning-ECG model, as brilliant as Rudolph’s red nose, was trained to predict when the right ventricle is as dilated as a stocking stuffed with presents (RVEDV >120 mL/m2) and when its function is as low as the North Pole’s temperatures in winter (RVEF ≤40%). This model was trained using data from the UK Biobank (UKBB; n=42 938) and fine-tuned in a multicenter health system (MSHoriginal [Mount Sinai Hospital]; n=3019), with a prospective validation over 4 merry months (MSHvalidation; n=115).
The model’s performance was as shiny as tinsel, with area under the curve scores that would make any Christmas light display look dim! For RV dysfunction, the scores were 0.86/0.81/0.77, and for RV dilation, they were 0.91/0.81/0.92 across the UKBB/MSHoriginal/MSHvalidation cohorts, respectively. The mean absolute error at MSHoriginal was RVEF=7.8% and RVEDV=17.6 mL/m2, which is not too shabby, like guessing the right size for a Christmas sweater!
And what’s more, this model was as inclusive as my list of good boys and girls, performing well across key subgroups, including those with and without left ventricular dysfunction. Over a median follow-up of 2.3 years, the predicted RVEF was associated with transplant-free survival (hazard ratio, 1.40 for each 10% decrease; P=0.031), proving that this isn’t just a Christmas miracle—it’s a gift that keeps on giving!
So there you have it, my jolly friends, deep learning-ECG analysis can indeed identify significant cardiac magnetic resonance imaging RV dysfunction and dilation with good performance. And just like leaving out cookies and milk for me, predicted RVEF is associated with clinical outcomes. Now, let’s dash away all to spread this good news, faster than my sleigh on Christmas night! 🦌🛷🎁