Neurodegenerative diseases like Alzheimer's disease (AD) and Parkinson's disease (PD) are irreversible disorders of a progressive nature. Early diagnosis, which helps early intervention, leads to a longer, healthier life. Conventional modalities of diagnosis are usually insensitive at an early phase. Artificial Intelligence (AI) is a trend that promises to identify such subtle pathological patterns in data of various modalities. The purpose of this systematic review is to assess the ability of AI-based tools to detect neurodegenerative disorders in the early stages, identify their synergy with clinical practice, and reveal current gaps. The literature search was based on the use of several databases and other resources, resulting in 328 unique records. The screening, followed by an assessment of eligibility based on PRISMA, revealed 12 primary studies. R was applied to perform a meta-analysis to estimate a pooled AUC and determine the heterogeneity. Publication bias was analysed by funnel plots and formal tests. The respective studies leverage techniques of AI, which include support vector machines and deep neural networks, and evaluate data types that include MRI, blood biomarkers, speech, and wearable sensors. The AUC pooled was 0.90 (95% CI: 0.88-0.91), implying great diagnostic precision. Substantial heterogeneity was experienced (I² = 30.3%). Research combining multimodal data and hybrid AI strategies produced the greatest results. There was only a little publication bias, as detailed in the funnel plot symmetry and statistical tests. Tools realised through AI illustrate strong diagnostic capabilities of neurodegenerative illnesses at an early stage. Nonetheless, additional external confirmation, long-term findings, and interpretability are required to be used clinically. Artificial intelligence (AI) can aid conventional diagnostics, leading to earlier and more exact interventions.
Raña et al. (Fri,) studied this question.