AI或可预警帕金森病
作者：微信文章AI Tool Shows Promise for Early Detection of Parkinson’s Disease
Parkinson’s disease (PD) is a neurodegenerative disorder that gradually impairs movement, causing symptoms such as tremor, stiffness, and balance problems. Detecting PD early, before these motor symptoms appear, has long been a challenge, but advances in artificial intelligence (AI) are opening new possibilities.
Recent studies have explored AI tools that analyze voice and gait data to identify subtle changes years before traditional diagnosis. Using machine learning techniques such as convolutional neural networks, recurrent neural networks, support vector machines, and ensemble classifiers, these models can process signals from smartphone microphones and wearable sensors to detect early signs of PD. Key voice features include jitter, shimmer, and Mel-frequency cepstral coefficients, while gait analysis focuses on stride length, cadence, and asymmetry.
The results are encouraging. Accuracy in distinguishing early-stage PD from healthy controls ranges from 95 to 97 percent, with sensitivity reaching up to 100 percent and specificity between 92 and 95 percent. Some voice-based models even achieve an area under the curve(AUC) of 0.98. AI can detect changes up to five to seven years before symptom onset.
Studies supporting these findings had sample sizes ranging from 195 to 847 participants and used datasets such as the UCI Parkinson’s Speech Dataset and the Oxford Parkinson’s Disease Dataset. Robust cross-validation methods ensured reliable results.
Clinically, these AI tools could enable remote screening and allow primary care physicians to identify at-risk individuals early. Early detection may support timely interventions such as MAO-B inhibitors, levodopa, or aerobic exercise. Integration with telehealth platforms and wearables such as smartwatches could further expand access.
However, limitations remain. Most studies involve Western populations, and models need testing across diverse groups. Confounding factors, such as voice changes from stroke or vocal cord conditions, must also be considered. Long-term trials in prodromal groups, such as individuals with REM sleep behavior disorder, are needed to confirm the effectiveness of these tools.
AI for early PD detection is still emerging, but these findings highlight a future where subtle changes in voice and gait could provide important clues years before traditional diagnosis. This offers hope for earlier intervention and improved patient outcomes.
Souce:
Shen, Y., et al. (2025). Explainable artificial intelligence to diagnose early Parkinson’s disease via voice analysis. Scientific Reports, 15(96575). https://doi.org/10.1038/s41598-025-96575-6García, A. M., et al. (2025). Voice biomarkers as prognostic indicators for Parkinson’s disease using machine learning techniques. Scientific Reports, 15(96950). https://doi.org/10.1038/s41598-025-96950-3Lee, J., et al. (2025). AI-driven precision diagnosis and treatment in Parkinson’s disease: A comprehensive review and experimental analysis. PMC, PMC12336134. https://pmc.ncbi.nlm.nih.gov/articles/PMC12336134/
页:
[1]