Trinity Centre for Bioengineering, Trinity College Dublin, Dublin 2, Ireland.
Med Eng Phys. 2010 Nov;32(9):1074-9. doi: 10.1016/j.medengphy.2010.07.013. Epub 2010 Aug 7.
Currently, there are no established objective biomarkers for the diagnosis or monitoring of schizophrenia. It has been previously reported that there are notable qualitative differences in the speech of schizophrenics. The objective of this study was to determine whether a quantitative acoustic and temporal analysis of speech may be a potential biomarker for schizophrenia. In this study, 39 schizophrenic patients and 18 controls were digitally recorded reading aloud an emotionally neutral text passage from a children's story. Temporal, energy and vocal pitch features were automatically extracted from the recordings. A classifier based on linear discriminant analysis was employed to differentiate between controls and schizophrenic subjects. Processing the recordings with the algorithm developed demonstrated that it is possible to differentiate schizophrenic patients and controls with a classification accuracy of 79.4% (specificity=83.6%, sensitivity=75.2%) based on speech pause related parameters extracted from recordings carried out in standard office (non-studio) environments. Acoustic and temporal analysis of speech may represent a potential tool for the objective analysis in schizophrenia.
目前，尚无用于诊断或监测精神分裂症的既定客观生物标志物。先前已有报道称，精神分裂症患者的言语存在明显的定性差异。本研究的目的是确定对语音进行定量声学和时间分析是否可能成为精神分裂症的潜在生物标志物。在这项研究中，对 39 名精神分裂症患者和 18 名对照者进行了数字录音，让他们大声朗读儿童故事中的一段情感中立的文本。从录音中自动提取了时间、能量和发声音调特征。基于线性判别分析的分类器用于区分对照组和精神分裂症患者。通过使用在标准办公（非录音室）环境中进行的记录开发的算法进行处理，基于从记录中提取的与语音停顿相关的参数，该算法能够以 79.4%的分类准确率（特异性=83.6%，敏感性=75.2%）区分精神分裂症患者和对照组。语音的声学和时间分析可能是精神分裂症客观分析的潜在工具。
