Mobile application-based oximetry: a potential toolfor appropriate referral of patients with respiratory symptoms examined via telemedicine

ABSTRACT Objective To calculate the positive likelihood ratio to determine whether telemedicine is able to optimize referral to the emergency department. Methods Unicenter study with 182 consecutive patients admitted to Hospital Israelita Albert Einstein due to respiratory symptoms. All patients were submitted to oxygen saturation measurement using the standard method Welch Allyn finger device vital sign monitor and a 2-minute evaluation (Binah.ai mobile application). The reproducibility of oxygen saturation measurements made with both methods was investigated using interclass correlation coefficients and analysis of dispersion. Bland-Altman plots were constructed and kappa concordance coefficients used to examine data normality. Accuracy was also estimated. Results Oxygen saturation measurement differences between methods were ≤2% in more than 85% of cases. The mean difference (bias) between methods was near zero (0.835; Bland-Altman analysis). Oxygen saturation measurements made using the Binah.ai mobile application had an average ability to detect patients with altered oxygen saturation levels compared to the conventional method (ROC analysis). The positive likelihood ratio of the mobile application was 6.23. Conclusion Mobile applications for oxygen saturation measurement are accessible user-friendly tools with moderate impact on clinical telemedicine evaluation of patients with respiratory symptoms, and may optimize referral to the emergency department.

【摘要】研究目的：计算阳性似然比，以明确远程医疗能否优化急诊科转诊流程。研究方法：本研究为单中心研究，纳入182例因呼吸道症状就诊于以色列阿尔伯特·爱因斯坦医院的连续入组患者。所有患者均接受采用标准设备——伟伦（Welch Allyn）指式生命体征监护仪的血氧饱和度测量，同时使用Binah.ai移动应用程序完成2分钟评估。采用组内相关系数与离散分析评估两种方法血氧饱和度测量结果的可重复性；构建布兰德-奥特曼（Bland-Altman）图，并使用Kappa一致性系数检验数据正态性，同时估算诊断准确度。研究结果：超过85%的病例中，两种方法测得的血氧饱和度差值≤2%；两种方法的平均差值（偏倚）接近0（0.835；布兰德-奥特曼分析）。经受试者工作特征曲线（ROC）分析，相较于传统检测方法，使用Binah.ai移动应用程序的血氧饱和度测量结果对血氧饱和度异常患者具有平均水平的检出能力。该移动应用的阳性似然比为6.23。研究结论：血氧饱和度测量类移动应用属于便捷易用的工具，对呼吸道症状患者的临床远程医疗评估具有中等应用价值，可优化急诊科转诊流程。