| 摘要: | 背景：阻塞性睡眠呼吸中止症為反覆性的上呼吸道塌陷或阻塞，導致呼吸過程費力且氣流變淺，而嚴重者可能造成窒息。協助提早發現阻塞性睡眠呼吸中止症為重要的目標。
目的：分析臨床檢驗數據與阻塞性睡眠呼吸中止症相關性，並建構預測阻塞性睡眠呼吸中止症風險之機器學習模型。
方法：資料來源取自衛生福利部雙和醫院（委託臺北醫學大學興建經營）之睡眠中心與醫學檢驗科。收案日期自2016年至2021年，包含20歲至90歲成人與年長者，使用SPSS軟體進行統計資料分析，使用機器學習建構預測模型，提供給有需求的人（如：健檢客戶）一個可以預測阻塞性睡眠呼吸中止症風險的工具。
結果：針對AHI 15 without MiniO2預測模型，AUC,（95% CI）達到0.848（0.836 - 0.860）；對於AHI 15 with MiniO2預測模型，AUC,（95% CI）達到0.888（0.882 - 0.894）。在AHI 30 without MiniO2預測模型中，AUC,（95% CI）達到0.809（0.797 - 0.821）；而在AHI 30 with MiniO2預測模型中，AUC,（95% CI）達到0.857（0.851 - 0.863）。身體參數結合臨床檢驗數據以及最低血氧飽和度可有效預測阻塞性睡眠呼吸中止症之風險。
Background：Obstructive sleep apnea is recurrent upper airway collapse or obstruction, it causes breathing difficulty and reduced airflow, which may cause suffocation in severe cases. Early detection of obstructive sleep apnea is an important research aim.
Purpose：Analyze the correlation between laboratory data and obstructive sleep apnea, and construct a machine learning model to predict the risk of obstructive sleep apnea.
Methods：The source of data is taken from the Sleep Center and Department of Laboratory Medicine of Taipei Medical University Shuang Ho Hospital （New Taipei City, Taiwan）between January 2016 and December 2021, This study included patients who were aged between 20 and 90 years, Use SPSS software for statistical data analysis, Constructing a predictive model using machine learning, we have created a predictive model that provides a tool for individuals with a need, such as health check-up clients, to predict their risk of obstructive sleep apnea.
Results：In this study, for the AHI 15 without MiniO2 prediction model, the AUC（95% CI） was 0.848（0.836 - 0.860）. For the AHI 15 with MiniO2 prediction model, the AUC（95% CI） was 0.888（0.882 - 0.894）. In the AHI 30 without MiniO2 prediction model, the AUC（95% CI） was 0.809（0.797 - 0.821）. In the AHI 30 with MiniO2 prediction model, the AUC（95% CI） was 0.857（0.851 - 0.863）. Combining Body profile with laboratory data and Minimum peripheral oxygen saturation can effectively predict the risk of obstructive sleep apnea. |
