Overall, AI research in neurodegenerative diseases suffers from significant limitations in reproducibility, data inclusivity, and clinical translatability, and a set of recommendations can be adopted to address these issues and improve reliability and downstream clinical utility.
Abstract
Background
The rising global burden of neurodegenerative diseases underscores an urgent need for advanced research in diagnosis, prognosis, and treatment. Artificial Intelligence (AI) methods, particularly when applied to multimodal data, offer a powerful tool to address these challenges. However, a comprehensive overview and critique of the current landscape of AI methods is lacking.
Methods
4,685 records of peer-reviewed, primary research articles were screened and 1,956 articles reviewed in full text, yielding 1,186 included studies. For each included study, clinical objectives, disease focus, data modalities, modelling approach, evaluation strategy, and reporting practices were extracted.
Results
Fewer than 5% of studies integrated pharmacological treatments into their predictive models, limiting the extent to which models can directly inform clinical decision-making. Neuroimaging was the predominant input modality, while integration of other clinically relevant data types was relatively rare. Reproducibility rates remain critically low at 35%, and external validation practices fail to use geographically and demographically diverse datasets.
Conclusions
Overall, AI research in neurodegenerative diseases suffers from significant limitations in reproducibility, data inclusivity, and clinical translatability. We provide a set of recommendations that can be adopted to address these issues and improve reliability and downstream clinical utility.
The progressive nature and overlapping clinical features of neurodegenerative diseases like Alzheimer's disease, Parkinson's disease, Frontotemporal Dementia, Huntington's disease, Dementia with Lewy Bodies and Amyotrophic lateral sclerosis make them tough to diagnose. While AI has shown great accuracy in diagnosis, it has the drawback of being hard for users to interpret, making it difficult to be broadly adopted in clinical practice. The current review aims at analyzing the state-of-the-art on the use of Explainable Artificial Intelligence (XAI) for the diagnosis of neurodegenerative diseases, focusing on AI models, explainability methods, datasets, validation methods, clinical applications, and implementation challenges. The review was conducted according to PRISMA 2020 guidelines and included the PubMed, IEEE Xplore and ScienceDirect databases of research articles published from 2018 to 2026. A total of 25 studies were included for qualitative synthesis from 454 retrieved studies. The results showed that convolutional neural networks and hybrid deep learning models are more widely studied than other methods, and that Grad-CAM, SHAP, LIME, and attention mechanisms are the most common explainability methods. The dataset of the Alzheimer's Disease Neuroimaging Initiative (ADNI) is widely used in existing studies, which is why Alzheimer's disease is still the major focus of study. While diagnostic capability is excellent, the translation to clinical practice is hampered by limited external validation, retrospective data sets, small sample sizes and regulatory issues. Future studies should focus on multimodal learning, prospective multi-center and various datasets along with trustworthy XAI frameworks to enable the routine clinical implementation.
Jasmine Chifurumnanya Nnabue, V. Nwaocha, Emmanuel Ejembi· Nature Journal of Emerging S...· 0 citations
INTRODUCTION
Artificial intelligence (AI) is increasingly recognized as a transformative paradigm within transplantation medicine, offering advanced computational approaches capable of integrating heterogeneous clinical, biological, imaging, and molecular datasets to improve predictive accuracy and decision-making. Liver transplantation represents a uniquely complex clinical domain characterized by high-dimensional data, nonlinear interactions among risk factors, and critical time-dependent decision processes, thereby providing an ideal context for AI-enabled analytics.
METHODS
The objective of this systematic review was to critically synthesize current evidence regarding AI applications in liver transplantation, with emphasis on data modalities, algorithmic methodologies, targeted clinical outcomes, validation strategies, and reported performance metrics. A comprehensive search of MEDLINE, Scopus, and the Cochrane Library identified 1045 records following duplicate removal and automated filtering.
RESULTS
After screening and eligibility assessment, 65 studies met the inclusion criteria. Laboratory data represented the most frequently utilized input (n = 35), followed by clinical (n = 28), demographic (n = 19), imaging (n = 13), and genetic or molecular data (n = 5), with several studies employing multimodal integration. Deep-learning architectures and neural network-based approaches predominated, with additional contributions from ensemble learning methods and conventional machine-learning algorithms. Across multiple clinical domains-including diagnostic classification, prognostic modeling, graft survival prediction, and treatment optimization-AI systems demonstrated high predictive performance, frequently surpassing traditional risk stratification tools such as model for end-stage liver disease and Survival Outcomes Following Liver Transplantation scores. Imaging-based models achieved particularly strong segmentation accuracy, whereas genomic and molecular approaches demonstrated excellent discriminative capability in oncologic and graft-related outcomes.
CONCLUSIONS
Despite these promising findings, significant methodological limitations persist, including data heterogeneity, insufficient external validation, risk of bias, and challenges related to interpretability, fairness, and ethical deployment. Overall, AI represents a highly promising adjunct to clinical decision-making in liver transplantation; however, robust prospective validation, standardized reporting frameworks, and clinically interpretable implementations remain necessary prior to widespread adoption.
Panagiotis Boutos, James L. Rogers, Efthymia Kouvela et al.· Journal of Surgical Research· 0 citations
Neurodegenerative diseases, including Alzheimer’s disease (AD), Parkinson’s disease (PD), frontotemporal dementia (FTD), and amyotrophic lateral sclerosis (ALS), pose a growing global health burden with limited early diagnostic tools. Radiomics, which extracts high-dimensional quantitative features from medical images [1, 2], combined with artificial intelligence (AI) methods, has emerged as a promising approach to enhance diagnostic accuracy and prognostic prediction in neuroimaging. However, no prior systematic review has comprehensively evaluated the methodological quality, reproducibility challenges, and diagnostic performance of AI-driven radiomics studies across multiple neurodegenerative diseases using established quality assessment frameworks. This systematic review was conducted following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 guidelines [3]. A comprehensive literature search was performed across PubMed/MEDLINE, Scopus, Web of Science Core Collection, and Embase databases from January 2017 to March 2026. Grey literature sources including conference proceedings from RSNA, ISMRM, and OHBM were systematically searched but excluded from the final synthesis. Two independent reviewers screened titles, abstracts, and full texts with substantial inter-reviewer agreement (Cohen’s kappa = 0.87). Methodological quality was assessed using the Quality Assessment of Diagnostic Accuracy Studies version 2 (QUADAS-2) tool and the Prediction model Risk Of Bias ASsessment Tool (PROBAST). Disagreements were resolved through consensus discussion, with a third reviewer consulted when necessary. Data extraction included study design, imaging modality, radiomic feature extraction methodology, AI/ML algorithm, sample size, performance metrics, validation strategy, and external validation status. From 1452 screened records, 60 studies met inclusion criteria and were included in the qualitative synthesis. The majority focused on AD and mild cognitive impairment (MCI) (n = 35, 58%), followed by PD and movement disorders (n = 15, 25%), FTD (n = 6, 10%), and other neurodegenerative conditions (n = 4, 7%). Structural MRI was the most commonly used modality (n = 38, 63%), followed by PET (n = 14, 23%) and SPECT (n = 8, 13%). Support vector machines (n = 22), convolutional neural networks (n = 18), and random forests (n = 12) were the most frequently employed AI methods. Reported area under the receiver operating characteristic curve (AUC) values ranged from 0.75 to 0.98 for AD diagnosis and 0.78 to 0.95 for PD classification. However, quality assessment revealed that only 12 studies (20%) performed external validation, and 28 studies (47%) were rated as having high risk of bias, primarily due to small sample sizes, lack of independent test sets, absence of prospective validation, and inadequate reporting of feature extraction parameters. Stratified analysis revealed that studies employing deep learning methods reported significantly higher AUC values (median 0.91) compared to classical machine learning approaches (median 0.85), though deep learning studies also exhibited higher risk of bias due to greater model complexity relative to sample sizes. Meta-analysis was not feasible due to substantial heterogeneity in imaging protocols, feature extraction pipelines, and outcome definitions. AI-driven radiomics demonstrates potential for improving neuroimaging-based diagnosis and prognosis of neurodegenerative diseases. However, the field remains substantially limited by methodological heterogeneity, insufficient external validation (only 20% of studies), high risk of bias (47% of studies), and critical reproducibility challenges including scanner variability, feature instability, and data leakage. The predominantly retrospective, single-center nature of existing evidence limits clinical generalizability. Future research should prioritize multi-center prospective validation with pre-registered protocols, standardized radiomics workflows adhering to Image Biomarker Standardisation Initiative (IBSI) guidelines, rigorous assessment of feature reproducibility across scanners and sites, and integration with multiomics data to facilitate responsible clinical translation.
Shih-Shuan Fang, Sheng-Han Chen· The Egyptian Journal of Radi...· 0 citations
These technologies show promise in reducing human error and enhancing mental health care delivery; however, persistent challenges include data privacy, ethical considerations, and the need for diverse, large-scale datasets.
Juster Donal Sinaga· Journal of Society Counselin...· 0 citations
Parkinson’s disease (PD) is a neurodegenerative brain condition that significantly impairs behavior, movement and speech. The Hoehn and Tahr staging scale is used to assess the extent of the condition. These evaluations can be costly, unpredictable, and time-consuming for individuals. Hence, the need to explore innovative approaches for diagnosis to enhance clinical results is highlighted by the lack of a definitive therapy. By using massive databases of organized data to improve diagnosis accuracy, artificial intelligence (AI) offers an opportunity to completely reinvent PD identification. Therefore, the purpose of this paper is to perform a systematic review of the literature with an emphasis on the significant advancements being made in this domain of research. The accessible existing research works published from 2022 to 2026 are gathered and examined in order to accomplish the analysis’s goal. Additionally, this review offers a thorough analysis of current research with an emphasis on various PD markers, adapted strategies and outcome measures. The considered articles are compared on the basis of their objective, databases, data type, utilized AI approaches and result. The results show an extensive variety of research conducted globally and a substantial advancement of PD detection employing AI techniques in recent years. Moreover, this research reveals a great deal of opportunity in employing fresh indicators and AI techniques in healthcare decisions, which could result in a more methodical and accurate identification of PD.
V.R. Anugu, Jimmy Singla· 2026 4th International Confe...· 0 citations
(1) Background: Pancreatic ductal adenocarcinoma remains one of the most lethal malignancies due to late diagnosis, aggressive tumor biology, and limited therapeutic options. Artificial intelligence has emerged as a promising tool to improve detection, risk stratification, and treatment planning. This study aims to review the current clinical applications of artificial intelligence in the management of pancreatic cancer and evaluate its translational potential. (2) Methods: A comprehensive literature search was conducted across major databases (PubMed/MEDLINE, Scopus, Web of science and Cochrane) for studies published between 2015 and 2026. Eligible studies included clinical investigations and systematic reviews reporting quantifiable outcomes related to diagnosis, staging, prognostication, and treatment response using artificial intelligence methods. (3) Results: A total of 24 studies were included, most of which were retrospective and utilized imaging, histopathology, and clinical datasets. Artificial intelligence demonstrated high diagnostic performance, particularly in imaging-based detection and lesion characterization, with several models achieving excellent accuracy. Applications in staging, surgical planning, and prognostication also showed promising results, although external validation and prospective data were limited. (4) Conclusions: Artificial intelligence has significant potential to enhance the management of pancreatic cancer, particularly as a decision-support tool. However, further prospective validation and integration into clinical workflows are required before widespread adoption.
A. Fotiadou, I. Margaris, K. Evangelou et al.· Onco· 0 citations