Garbey Marc, Lesport Quentin, Öztosun Gülşen, Ghodasara Veda, Kaminski Henry J, Bayat Elham
Department of Surgery, George Washington University School of Medicine & Health Sciences, Washington, DC, USA; Care Constitution Corp, Houston, TX, USA; Laboratoire des Sciences de l'Ingénieur pour l'Environnement (LaSIE) UMR-CNRS 7356 University of La Rochelle, France.
Care Constitution Corp, Houston, TX, USA; Laboratoire des Sciences de l'Ingénieur pour l'Environnement (LaSIE) UMR-CNRS 7356 University of La Rochelle, France.
J Neurol Sci. 2025 Jan 15;468:123328. doi: 10.1016/j.jns.2024.123328. Epub 2024 Nov 25.
Patients with ALS often face difficulties expressing emotions due to impairments in facial expression, speech, body language, and cognitive function. This study aimed to develop non-invasive AI tools to detect and quantify emotional responsiveness in ALS patients, providing objective insights. Improved understanding of emotional responses could enhance patient-provider communication, telemedicine effectiveness, and clinical trial outcome measures.
In this preliminary exploratory study, fourteen patients with ALS had audio recordings performed during routine clinic visits while wearing a wireless pulse oximeter. Emotion-triggering questions related to symptom progression, breathing, mobility, feeding tube, and financial burden were randomly asked. The same questions were posed in separate psychiatric evaluations. Natural language processing (NLP) was used to analyze transcriptions, topic classifications, sentiment, and emotional states, combining pulse and speech data. AI-generated reports summarized the findings.
Pulse alterations consistent with emotional arousal were identified, with longer consultations and positive communication reducing pulse fluctuations. Financial concerns triggered the strongest emotional response, while discussions about breathing, mobility, and feeding tube increased anxiety. AI-generated reports prioritized patient concerns and streamlined documentation for providers.
This study introduces a novel approach to linking pulse and speech analysis to evaluate emotional responses in ALS patients. AI and affective computing provide valuable insights into emotional responses and disease progression, with potential applications for other neurological disorders. This approach could augment clinical trial outcomes by offering a more comprehensive view of patient well-being.
肌萎缩侧索硬化症（ALS）患者常因面部表情、言语、肢体语言和认知功能受损而在表达情感方面面临困难。本研究旨在开发非侵入性人工智能工具，以检测和量化ALS患者的情绪反应，提供客观见解。更好地理解情绪反应可改善医患沟通、远程医疗效果及临床试验结果测量。
在这项初步探索性研究中，14名ALS患者在常规门诊就诊时佩戴无线脉搏血氧仪进行了音频录制。随机询问与症状进展、呼吸、行动能力、喂食管和经济负担相关的情绪触发问题。在单独的精神科评估中提出相同的问题。使用自然语言处理（NLP）分析转录内容、主题分类、情感和情绪状态，结合脉搏和语音数据。人工智能生成的报告总结了研究结果。
识别出与情绪唤醒一致的脉搏变化，咨询时间越长且沟通积极，脉搏波动越小。经济担忧引发的情绪反应最强烈，而关于呼吸、行动能力和喂食管的讨论增加了焦虑。人工智能生成的报告对患者的担忧进行了排序，并简化了医护人员的记录工作。
本研究引入了一种将脉搏和语音分析相结合以评估ALS患者情绪反应的新方法。人工智能和情感计算为情绪反应和疾病进展提供了有价值的见解，对其他神经系统疾病也有潜在应用。这种方法可通过提供更全面的患者健康状况视图来增强临床试验结果。
