Assessment of mental and behavioural non-motor symptoms of Parkinson’s Disease using Artificial Intelligence (AI): a systematic review

TL;DR

This paper reviews how artificial intelligence can help detect non-motor symptoms like cognitive issues and sleep problems in Parkinson’s disease, but more research is needed for depression and anxiety.

Contribution

The study systematically evaluates AI tools for mental and behavioral non-motor symptoms in Parkinson’s disease, highlighting gaps in depression and anxiety research.

Findings

Multimodal AI models show higher accuracy than single-source models for assessing non-motor symptoms.

Few studies focus on depression and anxiety, indicating a research gap.

External validation is needed before AI tools can be widely used in clinical settings.

Abstract

Parkinson’s disease is a progressive neurodegenerative disorder with both motor and non-motor symptoms. Mental and behavioural non-motor symptoms such as cognitive impairment, sleep disturbances, depression, and anxiety greatly affect quality of life but remain difficult to assess with traditional tools. Artificial intelligence has shown potential in healthcare, yet its role in evaluating these symptoms in Parkinson’s disease remains under-reviewed. This systematic review aims to evaluate the performance of artificial intelligence tools in diagnosing, assessing, and managing these symptoms.

Five databases (Medline, Embase, Scopus, Web of Science and PubMed) were searched up to June 2024 for peer-reviewed studies applying artificial intelligence to mental or behavioural symptoms in adults with Parkinson’s disease. Studies published before 2010 or lacking artificial-intelligence technologies were excluded. Study quality and risk of bias were assessed using QUADAS-2. Extracted data include study objectives, data sources, algorithms, best model, and diagnostic performance (accuracy, sensitivity, specificity). The study received no external financial support.

Here we show sixteen studies examine cognitive impairment and seven examine sleep disorders. However, only three studies focus on depression and one on anxiety, revealing a research gap. No meta-analysis was performed due to heterogeneity.

Artificial intelligence shows promise for assessing mental and behavioural symptoms in Parkinson’s disease, particularly cognitive and sleep disorders. Multimodal models demonstrate higher accuracy than single-source models, though external validation is necessary. The limited studies on depression and anxiety reflect existing diagnostic challenges and data limitations. Future research should refine diagnostic tools and expand multimodal approaches to these symptoms.

Parkinson’s disease causes both movement problems and non-movement symptoms such as difficulties with thinking, decision-making, memory, sleep, depression, and anxiety. These symptoms are common and deeply affect quality of life for people living with Parkinson’s, but they are often missed or underestimated by traditional clinical assessments. We systematically examined whether artificial intelligence could help detect and monitor these symptoms.We analysed twenty-seven studies and found that artificial intelligence shows promise for identifying thinking and sleep-related problems, but very few studies examined depression or anxiety. The findings suggest that combining different types of patient data improves accuracy, but more validation is needed before these tools can be used in clinical settings to help doctors provide better care.

Chou et al. systematically review 27 studies using artificial intelligence to assess mental and behavioural non-motor symptoms in Parkinson’s disease. Multimodal models outperform single-source approaches but require external validation, while limited research on depression and anxiety reflects a broader research gap in the field.

Figures