Summary: HoneyNaps has published a study in Digital Health validating the effectiveness of its AI-powered diagnostic algorithm, Somnum, for detecting sleep apnea. The study, conducted with 1,000 patients across various levels of sleep-disordered breathing, demonstrated that Somnum achieved high sensitivity and specificity in interpreting polysomnography results. The algorithm showed strong predictive accuracy for diagnosing mild, moderate, and severe cases of sleep apnea, performing comparably to expert readings.
Key Takeaways:
- High Diagnostic Accuracy: Somnum achieved high sensitivity (98%) and specificity (96%) in detecting apnea and hypopnea from polysomnography results, aligning closely with expert interpretations.
- Strong Predictive Performance Across Severity Levels: The AI algorithm demonstrated consistent predictive accuracy (AUC scores above 0.94) across mild, moderate, and severe sleep apnea cases.
- Validated for Clinical Use: The study’s findings support the clinical efficacy of HoneyNaps’ AI technology, paving the way for broader adoption of AI-driven diagnostics in sleep medicine.
Medical artificial intelligence (AI) company HoneyNaps has published a paper demonstrating the clinical value and efficacy of its sleep disorder diagnostic algorithm, Somnum.
According to the validation results, Somnum exhibited high sensitivity and specificity in interpreting apnea and hypopnea from polysomnography (PSG) across all groups of sleep-disordered breathing patients, with excellent predictive performance in mild, moderate, and severe sleep apnea cases.
The clinical study, published in Digital Health, was conducted by Choi JiHo, MD, PhD, head of the Center for Sleep Medicine at SoonChunHyang University Hospital Bucheon, and Park MarnJoon, MD, of the Department of Otolaryngology at Inha University Hospital. It involved 1,000 adults diagnosed with various levels of sleep-disordered breathing through polysomnography, including simple snoring and mild, moderate, and severe sleep apnea.
Study Results
Comparing data interpreted by the AI-based Somnum with expert readings of polysomnography, the study showed high sensitivity (95% CI: 98.06–98.51) and specificity (95% CI: 95.46–97.79) for detecting apnea and hypopnea across all sleep-disordered breathing groups.
Somnum also demonstrated excellent predictive accuracy for sleep apnea across all severity levels. The AUC (area under the ROC curve) scores for disease prediction in mild, moderate, and severe groups were 0.9402, 0.9388, and 0.9442, respectively, with no significant differences among the groups.
“We are pleased that the clinical efficacy and efficiency of our sleep medical AI solutions are being validated through a reputable journal as well as through R&D and clinical trials, allowing steady adoption in medical settings,” says Sean Ha (Tae Kyoung Ha), president of HoneyNaps USA Inc, in a release. “We will continue our research and development efforts and publish diverse study outcomes in global journals to validate the clinical value of HoneyNaps’ medical AI technology.”
Photo caption: Somnum
File photo/HoneyNaps
Leave a Reply