Background
Hypertrophic cardiomyopathy (HCM) is commonly concomitant with sleep-disordered respiration (SDB), which may trigger antagonistic cardiovascular occasions. Though an applicable method to SDB prevents cardiac remodelling, detection of concomitant SDB in sufferers with HCM stays suboptimal. Thus, we aimed to develop a machine learning-based discriminant mannequin for SDB in HCM.
Strategies
Within the current multicentre examine, we consecutively registered sufferers with HCM and carried out nocturnal oximetry. The end result was a excessive Oxygen Desaturation Index (ODI), outlined as 3% ODI >10, which considerably correlated with the presence of reasonable or extreme SDB. We randomly divided the entire contributors right into a coaching set (80%) and a check set (20%). With information from the coaching set, we developed a random forest discriminant mannequin for prime ODI primarily based on scientific parameters. We examined the flexibility of the discriminant mannequin on the check set and in contrast it with a earlier logistic regression mannequin for distinguishing SDB in sufferers with HCM.
Outcomes
Amongst 369 sufferers with HCM, 228 (61.8%) had excessive ODI. Within the check set, the realm beneath the receiver working attribute curve of the discriminant mannequin was 0.86 (95% CI 0.77 to 0.94). The sensitivity was 0.91 (95% CI 0.79 to 0.98) and specificity was 0.68 (95% CI 0.48 to 0.84). When the check set was divided into low-probability and high-probability teams, the high-probability group had a better prevalence of excessive ODI than the low-probability group (82.4% vs 17.4%, OR 20.9 (95% CI 5.3 to 105.8), Fisher’s precise check p<0.001). The discriminant mannequin considerably outperformed the earlier logistic regression mannequin (DeLong check p=0.03).
Conclusions
Our examine serves as the primary to develop a machine learning-based discriminant mannequin for the concomitance of SDB in sufferers with HCM. The discriminant mannequin could facilitate cost-effective screening assessments and coverings for SDB within the inhabitants with HCM.