Alzheimer’s disease is characterised by the accumulation of extracellular amyloid-β plaques and intracellular Tau neurofibrillary tangles, leading to progressive memory impairment and cognitive dysfunction. Parkinson’s disease, by contrast, involves the degeneration of dopaminergic neurons in the substantia nigra and the aggregation of α-synuclein, producing motor impairments such as tremors, rigidity, and bradykinesia. Both conditions involve progressive neuronal loss, misfolded protein accumulation, and irreversible decline in cognitive or motor function. Yet the review suggests that neither disease is solely brain-centric. Alterations in peripheral systems, especially the gastrointestinal tract, play a major role in disease pathogenesis.

Gut dysbiosis in both Alzheimer’s and Parkinson’s disease commonly involves reduced microbial diversity, altered taxonomic composition, depletion of beneficial short-chain fatty acid-producing bacteria, and enrichment of pro-inflammatory taxa. These changes compromise the integrity of the intestinal barrier, leading to increased intestinal permeability—often referred to as leaky gut. Microbial products such as lipopolysaccharides translocate into the systemic circulation, triggering immune activation and chronic inflammation. This, in turn, disrupts the blood-brain barrier, activates microglia, and promotes neuroinflammatory cascades that contribute to neuronal damage and disease progression.

Short-chain fatty acids, primarily acetate, propionate, and butyrate, are produced when gut bacteria ferment dietary fibres. These metabolites maintain intestinal integrity, regulate immune responses, and modulate neuronal activity. Butyrate acts as a histone deacetylase inhibitor, influencing gene expression associated with neuroprotective and anti-inflammatory pathways. In both diseases, a reduction in SCFA-producing bacteria such as Faecalibacterium, Roseburia, and Eubacterium has been reported, diminishing production of these beneficial metabolites and contributing to increased intestinal permeability, systemic inflammation, and altered microglial activation within the brain.

The review places particular emphasis on inflammatory mediators including tumour necrosis factor alpha, interleukin 1 beta, interleukin 6, the toll-like receptor 4/nuclear factor kappa B pathway, and the NLRP3 inflammasome. Translocation of bacterial endotoxins such as lipopolysaccharide activates toll-like receptor 4, triggering NF-κB-mediated inflammatory cascades, driving microglial activation, and elevating cytokines. Gut barrier dysfunction and increased exposure to microbial products provide inflammatory priming through TLR4/NF-κB signalling, followed by NLRP3 inflammasome activation and caspase-1-mediated maturation of IL-1β and IL-18. In Alzheimer’s disease, NLRP3 activation has been associated with amyloid-β-induced microglial activation and impaired inflammatory regulation. In Parkinson’s disease, α-synuclein aggregates can activate NLRP3-dependent inflammatory responses in microglia, contributing to neuronal injury.

One of the most striking comparative findings concerns the vagus nerve. In Parkinson’s disease, pathological propagation of α-synuclein from the enteric nervous system to the central nervous system via vagal pathways has been proposed, supported by the observation that gastrointestinal symptoms such as constipation and altered motility often precede motor signs by years. Germ-free α-synuclein-overexpressing mice exhibit attenuated motor deficits that reappear after colonisation with Parkinson’s patient microbiota, implicating a causal role for the microbiota. In Alzheimer’s disease, however, current evidence does not support the generalisation that pathology originates in the gastrointestinal tract. Instead, Alzheimer’s is closely linked to metabolic dysfunction, insulin resistance, and systemic inflammatory priming, with microbial metabolites contributing to mitochondrial dysfunction, oxidative stress, and a shift in tryptophan metabolism towards the kynurenine pathway, yielding neurotoxic metabolites such as quinolinic acid.

Microbial amyloids represent another shared mechanism. The gut microbiota produces amyloid-like proteins that structurally resemble pathological proteins found in neurodegenerative diseases. These microbial amyloids can stimulate immune responses and potentially promote cross-seeding of host proteins, reinforcing amyloid-β accumulation in Alzheimer’s and α-synuclein aggregation in Parkinson’s. Certain bacteria, such as Escherichia coli, produce functional amyloids that cross-react with host proteins, promoting misfolding and aggregation. The review also highlights the role of neurotransmitters and neuromodulators, including serotonin, dopamine precursors, GABA, and acetylcholine, which the gut microbiota synthesises and which influence central neurotransmission via vagal and immune pathways.

The review then turns to recent advances in artificial intelligence and machine learning for integrating microbiome, clinical, neuroimaging, and multi-omics datasets. Traditional diagnostic approaches depend on clinical symptoms, neuropsychological testing, and imaging findings that typically reflect late-stage pathology. AI-driven models facilitate early diagnosis, disease prediction, and patient stratification by identifying subtle patterns across multimodal data, including neuroimaging, biofluid biomarkers, genetic profiles, and digital health metrics. Convolutional neural networks can detect subtle hippocampal atrophy and cortical degradation in early Alzheimer’s, while PET-based machine learning classifiers differentiate amyloid-positive from amyloid-negative individuals with high sensitivity and specificity. In Parkinson’s disease, machine learning algorithms applied to dopamine transporter imaging help identify nigrostriatal degeneration and differentiate idiopathic Parkinson’s from atypical Parkinsonian syndromes.

Microbiome analysis platforms such as QIIME 2 provide reproducible workflows for processing sequencing data, including quality assessment, feature generation, taxonomic classification, and microbial diversity analysis. Multi-omics integration tools including mixOmics and MOFA+ support multivariate integration and latent-factor modelling across heterogeneous omics datasets, facilitating the identification of coordinated molecular signatures associated with either disease. Machine learning algorithms such as random forests, support vector machines, and XGBoost can integrate clinical characteristics, genetic variants, inflammatory biomarkers, microbiome-derived features, and metabolomic profiles for disease classification, risk prediction, and patient stratification. XGBoost is particularly applicable to structured biomedical datasets because it can model non-linear relationships and interactions amongst multiple predictors whilst incorporating regularisation to reduce overfitting. Deep neural networks can analyse high-dimensional datasets derived from genome-wide transcriptomics, epigenomics, and metabolomics, and integrating genetic risk factors such as APOE with machine learning models might improve predictive accuracy for disease progression.

Digital biomarkers are also redefining early detection. AI-based wearable sensors and smartphone-based motion tracking systems generate continuous data on tremor amplitude, bradykinesia, gait instability, and motor fluctuation in Parkinson’s disease. Recurrent neural networks and long short-term memory models detect longitudinal motor patterns and identify subtle abnormalities early, enabling remote monitoring. In Alzheimer’s disease, natural language processing techniques analyse speech patterns, semantic errors, pause frequency, and linguistic diversity to detect early cognitive decline, providing a non-invasive, scalable screening tool with potential for community-level implementation. The review concludes that integrating gut-brain axis biology with AI-driven analytics provides promising opportunities for early diagnosis, personalised therapeutic strategies, and precision medicine. However, the authors caution that high predictive performance within a development cohort should not be interpreted as evidence of clinical validity without robust external and prospective validation, and that microbiome-based models present additional challenges because microbial abundance data are compositional and strongly influenced by diet, medication, geography, and sample-processing procedures.

Subject of Research: Health & Medicine

Article Title: The gut-brain axis in Alzheimer’s and Parkinson’s diseases: a comparative perspective

Article References: Upadhyay, R., C., C., Kar, S., Sathyamoorthy, S., Pitchaimuthu, X., & Sevanan, M. (2026). The gut-brain axis in Alzheimer’s and Parkinson’s diseases: a comparative perspective. Frontiers in Microbiomes, 5, Article 1942228. https://doi.org/10.3389/frmbi.2026.1942228

Image Credits: AI Generated

DOI: 10.3389/frmbi.2026.1942228

Keywords: gut-brain axis, Alzheimer's disease, Parkinson's disease, gut microbiota, neuroinflammation, artificial intelligence, precision medicine, neurodegenerative disease, gut-brain, axis, Alzheimer, Parkinson