Machine learning for multidimensional response and survival after cardiac resynchronization therapy using features from cardiac magnetic resonance

Cardiac magnetic resonance defines mechanisms of sex-based differences in outcomes following cardiac resynchronization therapy

“…Furthermore, obesity, renal disease and kidney dysfunction are known to influence BNP levels ( 30 , 31 ), yet these parameters were not significant among sex within our cohort. We along with others have shown that lower levels of BNP are correlated with better CRT response ( 19 , 27 ), and since in our current cohort, and as the NICM females in our cohort had the lowest BNP levels, this measure may mechanistically explain the better outcomes of NICM females. Still, our findings of lower BNP levels in females with no difference in age or kidney function coupled with favorable outcomes warrant further investigation.…”

“…In our present study, females generally had lower CURE-SVD (and thus more dyssynchrony) than males while the female NICM group had the lowest CURE-SVD among all the groups. Our group has previously demonstrated the effectiveness of CURE-SVD in predicting LV functional improvement (LVESVI-FC) in CRT within prior cohorts and has shown that a lower CURE-SVD correlates with a more favorable LVESVI-FC ( 23 – 27 ); however, CURE-SVD’s association with female sex is a new finding. This association provides a possible mechanistic insight as to why females are more likely to have more beneficial outcomes after CRT – they have greater degrees of dyssynchrony and consequently experience more favorable changes in LV functional improvement from ventricular resynchronization.…”

The saga of dyssynchrony imaging: Are we getting to the point

“…Similarly, the application of a clustering approach (k-means method and Gaussian mixture model) to clinical and CMRderived features was also able to stratify CRT-candidates in three survival groups. Interestingly, the addition of CRT-volumetric response to selected pre-CRT data significantly increased the performance of the model for the prediction of prognosis (AUC 0.78 ± 0.04 vs. 0.86 ± 0.02) (80). Although it is difficult to directly compare classifier performances because different datasets and ML models were used in these studies, these results show the added value of multivariate analysis for the prediction of CRT response.…”

First-in-human noninvasive left ventricular ultrasound pacing: A potential screening tool for cardiac resynchronization therapy

“…Although the benefits of cardiac resynchronization therapy (CRT) have been demonstrated in clinical trials, 1 , 2 , 3 , 4 the response rate to CRT in real-world cohorts with guideline-based indications 5 remains suboptimal. 6 Although models have been developed to predict response to CRT, 7 , 8 , 9 , 10 the field lacks a noninvasive, real-time, accurate simulation to determine which patients would have a favorable response to CRT and the degree of response to be expected. Some approaches such as the ECG Belt (Medtronic plc., Mounds View, MN; an array of 40 electrodes around the mid torso) provide feedback about the effectiveness of resynchronization only after the left ventricular (LV) pacing lead has been implanted.…”