[Multidimensional clinical phenotypes of obstructive sleep apnea hypopnea syndrome based on unsupervised clustering algorithm].

Objective

To advance the precision diagnosis and treatment of obstructive sleep apnea-hypopnea syndrome（OSAHS）, this study aimed to identify phenotypes with distinct pathophysiological and clinical features based on multidimensional clinical data using unsupervised clustering methods.

Methods

Polysomnography indicators （apnea-hypopnea index[AHI], lowest oxygen saturation, oxygen desaturation index, arousal index） and clinical symptom scale data（Epworth Sleepiness Scale[ESS], Insomnia Severity Index[ISI], Self-Rating Anxiety Scale[SAS], Self-Rating Depression Scale[SDS10]） were collected from 6 633 OSAHS patients who attended the Departments of Otorhinolaryngology-Head and Neck Surgery at the Second Affiliated Hospital of Xi'an Jiaotong University, Xi'an Central Hospital, and the Stomatological Hospital of Xi'an Jiaotong University. After Z-score normalization, K-means clustering was applied. The optimal number of clusters was determined using the elbow method and silhouette coefficient, and the results were visualized using box plots and radar charts.

Results

Following data cleaning, 1 331 OSAHS patients were included in the final analysis and classified into four clinical phenotypes: ①Classic sleepy type（n=227）: characterized by concurrent objective sleep disturbance and subjective sleepiness; ②Asymptomatic hypoxic type（n=383）: characterized by severe physiological abnormalities but mild subjective symptoms; ③High symptom-mild physiological type（n=413）: characterized by high subjective symptom burden but relatively mild objective abnormalities; ④Symptom perception-dominant type（n=308）: characterized by prominent emotional and insomnia symptoms with the mildest physiological abnormalities. Significant differences were observed among phenotypes in sex distribution（P<0.05）, whereas no statistically significant differences were found in age, body mass index, or AHI. Significant differences were observed among phenotypes in BMI（P=0.023） and SDS（P=0.016）, whereas no statistically significant differences were found in age or AHI.

Conclusion

OSAHS patients exhibit multidimensional clinical phenotypes. Classification based on the integration of objective AHI with multidimensional subjective and objective information should be adopted to formulate the individualized treatment strategies.