Authors:
Nhung Hoang
1
and
Zilu Liang
1
;
2
Affiliations:
1
Ubiquitous and Personal Computing Lab, Kyoto University of Advanced Science (KUAS), Kyoto, Japan
;
2
Institute of Industrial Science, The University of Tokyo, Tokyo, Japan
Keyword(s):
Sleep Apnea, AHI, Contrast Set Mining, Longitudinal Data, SHHS.
Abstract:
Sleep apnea remains a key area of sleep research, with the Apnea-Hypopnea Index (AHI) widely used to assess its severity. This study evaluated whether AHI is truly the best indicator of sleep apnea and identified its limitations. Using the Sleep Heart Health Study and Wisconsin Sleep Cohort datasets, which provide large, longitudinal data, we also explored survey data on demographics, physiology, and daily behaviors—often overlooked in polysomnography-based studies. The results indicate that AHI may be a good indicator for mild or moderate sleep apnea, but not necessarily for normal or severe cases. We highlight some trends that can be seen from longitudinal data. Additionally, using contrast set mining method, we identified key risk factors for cardiovascular disease, including age, snoring, and smoking behavior. These results underscore the importance of considering AHI’s limitations and incorporating additional factors for more accurate sleep apnea diagnosis and risk assessment.