IEEE Trans Neural Syst Rehabil Eng. 2018 Aug;26(8):1495-1507. doi: 10.1109/TNSRE.2018.2851787. Epub 2018 Jun 29.
Although smartphone technology provides new opportunities for the recording of speech samples in everyday life, its ability to capture prodromal speech impairment in persons with a high risk of developing Parkinson's disease (PD) has never been investigated. Speech data were acquired through a smartphone as well as a professional microphone with a linear frequency response from 50 participants with a rapid eye movement sleep behavior disorder that are at a high risk of developing PD and related neurodegenerative disorders. Additionally, recordings of 30 newly diagnosed, untreated PD patients and 30 healthy participants were evaluated. Acoustic assessment of 11 speech dimensions representing the key aspects of hypokinetic dysarthria in the early stages of PD was performed. Smartphone allowed the detection of speech abnormalities in participants with a high risk of developing PD. Acoustic measurements related to fundamental frequency variability, duration of pause intervals, and rate of speech timing extracted from spontaneous speech were sufficiently sensitive to significantly separate groups (area under curve of 0.85 between PD and controls) and showed very strong correlation and reliability between the professional microphone and the smartphone. Speech-based biomarkers collected through smartphones may have the potential to revolutionize the diagnostic process in neurodegenerative diseases and improve stratification for future neuroprotective therapy in PD.
虽然智能手机技术为日常生活中的语音样本记录提供了新的机会，但它在捕捉有发展为帕金森病（PD）风险的人出现前驱期言语障碍方面的能力尚未得到研究。通过智能手机和具有线性频率响应的专业麦克风，从 50 名快速眼动睡眠行为障碍患者（他们有发展为 PD 和相关神经退行性疾病的高风险）中获取了语音数据。此外，还评估了 30 名新诊断、未经治疗的 PD 患者和 30 名健康参与者的录音。对代表 PD 早期运动减少性构音障碍关键方面的 11 个语音维度进行了声学评估。智能手机能够检测出有发展为 PD 高风险的参与者的语音异常。从自发语音中提取的与基频变化、停顿间隔时长和语速有关的声学测量值在区分 PD 组和对照组时具有足够的敏感性（曲线下面积为 0.85），并且在专业麦克风和智能手机之间显示出非常强的相关性和可靠性。通过智能手机收集的基于语音的生物标志物可能有潜力改变神经退行性疾病的诊断过程，并改善 PD 中未来神经保护治疗的分层。
