The role of artificial intelligence in neurodegenerative disorders detection and diagnosis: a systematic review

Dementia represents a healthcare challenge, as pathological changes often emerge years before overt symptoms. Early manifestations such as Mild Cognitive Impairment (MCI) and Subjective Cognitive Decline (SCD) are critical transitional stages between normal aging and dementia. Thus, distinguishing these conditions (i.e. MCI, SCD) and whether they could evolve to dementia remains crucial. However, to date, the clinical tools available, chiefly among them neuroimaging and the neuropsychological assessment remain not always clearly interpretable and resource intensive as well. In recent years, Artificial Intelligence (AI), including Machine Learning (ML) and Deep Learning (DL), has shown promising capabilities in enhancing early detection, predicting progression, and supporting differential diagnosis across neurodegenerative disorders. This systematic review aims at synthesizing the current evidence on the extent to which AI-based approaches are being applied to improve diagnostic accuracy and prognostic assessment in dementia. A comprehensive literature search of studies published between 2015 and 2025 was carried out across major scientific databases following PRISMA 2020 guidelines, and studies were evaluated with respect to data modality, methodological rigor, performance metrics, and clinical applicability. The results highlight that AI models, particularly DL architectures applied to neuroimaging, electrophysiological signals, speech and language features, biomarker profiles, and digital behavioral data, consistently achieve high diagnostic accuracy for distinguishing MCI, Alzheimer’s disease, and healthy aging. Moreover, predictive models demonstrate potential in predicting conversion from MCI to dementia and in monitoring real-world cognitive and functional trajectories using wearable or smart-home technologies. However, the evidence also reveals substantial heterogeneity, limited external validation, and methodological inconsistencies that constrain clinical translation. In conclusion, this review highlights that AI is a rapidly evolving and promising tool for the early detection and monitoring of neurodegenerative disorders. Binding together the studies reported in this review, underscore the need for standardized pipelines, larger multicenter datasets, and explainable AI frameworks to support a real clinical implementation of this technology.