Objective: This study reviews AI’s role in early PD detection, comparing AI-driven diagnostic methods with conventional clinical approaches in accuracy, reliability, and applicability.

Background: Parkinson’s Disease (PD) is a progressive neurodegenerative disorder marked by motor and non-motor symptoms. Early detection is crucial for improving outcomes, but traditional diagnostic tools, such as the Unified Parkinson’s Disease Rating Scale (UPDRS), dopamine transporter imaging (DATscan), and single-photon emission computed tomography (SPECT), have limitations in identifying PD at preclinical stages. Advances in Machine Learning (ML) and Deep Learning (DL) enable Artificial Intelligence (AI) to analyze neuroimaging, speech signals, motor function, and biomarkers with greater precision. Studies suggest AI models outperform traditional methods, providing higher sensitivity and specificity in early PD detection.

Method: A narrative review analyzed 40 studies published over 10 years, selected for their relevance to AI applications in PD diagnosis on PubMed

Results: AI models demonstrate high accuracy in detecting early PD across multiple modalities. In neuroimaging, Deep Learning models, such as ResNet and U-Net CNN architectures, achieved up to 92% accuracy, surpassing DATscan and SPECT in sensitivity. AI-based speech analysis using BERT and LSTM models identified early PD with 85-90% sensitivity, outperforming UPDRS speech subscales. Motion data from accelerometers and gyroscopes, analyzed by AI, detected PD-related gait abnormalities with 90% accuracy, enabling remote monitoring. Hybrid AI models combining neuroimaging, speech, and sensor data improved early-stage classification, reducing false positives and achieving more reliable differentiation from other movement disorders.

Conclusion: AI-driven diagnostic approaches offer superior sensitivity and specificity compared to traditional methods, suggesting a paradigm shift in early PD detection. However, clinical translation faces challenges, including data standardization, algorithmic interpretability, and large-scale validation. Ethical concerns, such as algorithmic bias and false positives, must be addressed to ensure AI’s safe and equitable application in clinical practice. Future research should prioritize multimodal datasets, clinical trials, and integration into neurological workflows to enhance AI’s impact on PD diagnosis and care.

« Back to 2025 International Congress

MDS Abstracts - https://www.mdsabstracts.org/abstract/artificial-intelligence-in-the-early-detection-of-parkinsons-disease-advances-challenges-and-future-perspectives/