生理学与机器学习相结合用于预测阻塞性睡眠呼吸暂停中的持续气道正压通气压力和残余呼吸暂停低通气指数

The combination of physiology and machine learning for prediction of CPAP pressure and residual AHI in OSA.

作者信息

Lo Jui-En, Schmickl Christopher N, Vaida Florin, Nemati Shamim, Singh Karandeep, Sands Scott A, Owens Robert L, Malhotra Atul, Orr Jeremy E

机构信息

Division of Pulmonary, Critical Care, and Sleep Medicine, University of California San Diego, San Diego, California.

School of Public Health, University of California San Diego, San Diego, California.

出版信息

J Clin Sleep Med. 2025 May 1;21(5):775-782. doi: 10.5664/jcsm.11498.

DOI:10.5664/jcsm.11498

PMID:39745438

原文链接:https://pmc.ncbi.nlm.nih.gov/articles/PMC12048318/

Abstract

STUDY OBJECTIVES

Continuous positive airway pressure (CPAP) is the treatment of choice for obstructive sleep apnea; however, some people have residual respiratory events or require significantly higher CPAP pressure while on therapy. Our objective was to develop predictive models for CPAP outcomes and assess whether the inclusion of physiological traits enhances prediction.

METHODS

We constructed predictive models from baseline information for subsequent residual apnea-hypopnea index and optimal CPAP pressure. We compared models utilizing clinical variables with those incorporating both clinical and physiological factors. Furthermore, we assessed the performance of regression vs machine learning. All performances, including root mean square error, R-squared, accuracy, and area under the curve, were evaluated using a 5-fold cross validation with 10 repeats.

RESULTS

For predicting residual apnea-hypopnea index, random forest models outperformed regression models, and models that incorporated both clinical and physiological variables also outperformed models using only clinical variables across all performance metrics. Random forest using both clinical features and physiological traits achieved the best performance. In both regression and random forest models, central apnea index is found to be the most important feature in predicting residual apnea-hypopnea index. For predicting CPAP pressure, there was no additional predictive value of physiological traits or random forest modeling.

CONCLUSIONS

Our findings demonstrated that the combined use of clinical and physiological variables yields the most robust predictive models for residual apnea-hypopnea index, with random forest models performing best. These findings support the notion that prediction of obstructive sleep apnea therapy outcomes may be improved by more flexible models using machine learning, potentially in combination with physiology-based models.

CITATION

Lo J-E, Schmickl CN, Vaida F, et al. The combination of physiology and machine learning for prediction of CPAP pressure and residual AHI in OSA. . 2025;21(5):775-782.

摘要

研究目的

持续气道正压通气（CPAP）是阻塞性睡眠呼吸暂停的首选治疗方法；然而，有些人在治疗期间仍有残余呼吸事件或需要显著更高的CPAP压力。我们的目标是开发CPAP治疗结果的预测模型，并评估纳入生理特征是否能增强预测效果。

方法

我们根据基线信息构建了预测模型，用于预测后续的残余呼吸暂停低通气指数和最佳CPAP压力。我们将使用临床变量的模型与同时纳入临床和生理因素的模型进行了比较。此外，我们评估了回归模型与机器学习模型的性能。所有性能指标，包括均方根误差、决定系数、准确率和曲线下面积，均采用5折交叉验证并重复10次进行评估。

结果

在预测残余呼吸暂停低通气指数方面，随机森林模型优于回归模型，并且在所有性能指标上，同时纳入临床和生理变量的模型也优于仅使用临床变量的模型。结合临床特征和生理特征的随机森林模型表现最佳。在回归模型和随机森林模型中，中枢性呼吸暂停指数被发现是预测残余呼吸暂停低通气指数最重要的特征。在预测CPAP压力方面，生理特征或随机森林建模没有额外的预测价值。