27,380 Alzheimer's and dementia studies in plain English, each scored out of 100.
Previous research has shown patients with mental health issues receive worse quality of diabetes and cardiac care and have poorer outcomes compared with those without mental health issues. 5 key findings →
Evidence: low
Full summary, score breakdown and related studies → Original source
Alzheimer’s disease (AD) target evidence is distributed across biological, chemical, structural, regulatory, and trial-registry sources, complicating reproducible target evaluation and modeling. 1 key finding →
Evidence: low
Full summary, score breakdown and related studies → Original source
To distinguish these cases, we trained three types of supervised 3D transformers (ViT3D, MINiT, NIT) and, as a non-collapsing alternative, a frozen self-supervised encoder (DINO), to classify people with dementia or mild cognitive impairment (MCI) versus cognitively normal (CN) individuals from 3D brain MRI, using data from the Alzheimer’s Disease Neuroimaging Initiative (ADNI). 2 key findings →
Evidence: low
Full summary, score breakdown and related studies → Original source
A bstract Deep learning models for electroencephalography (EEG) pathology detection achieve competitive accuracy but offer limited interpretability. 5 key findings →
Evidence: low
Full summary, score breakdown and related studies → Original source
Alzheimer disease (AD) is a disorder of the nervous system that mostly impacts the brain cells and causes the slow destruction of the brain. 1 key finding →
Evidence: low
Full summary, score breakdown and related studies → Original source
Unlike traditional eager decision trees (eagerDT), which lacked sufficient training data for 17%-33% of patient profiles, lazyDT-MCCPath successfully modeled all profiles and achieved a 0.17 improvement in AUROC (0.66 vs. 0.83). lazyDT-MCCPath effectively translates pre-existing MCC profiles into personalized dementia risk probabilities for emerging conditions, which may help clinicians prioritize MCCs to monitor more closely. 1 key finding →
Evidence: low
Full summary, score breakdown and related studies → Original source
Patients with peripheral artery disease (PAD), chronic limb-threatening ischemia (CLTI), and/or diabetic foot ulcer (DFU) are at increased risk for limb loss and death. 3 key findings →
Participants: 178,995 Evidence: moderate
Full summary, score breakdown and related studies → Original source
Machine learning (ML) methods have shown promise in modeling clinical trajectories associated with neurodegenerative disorders such as Alzheimer’s disease (AD). 3 key findings →
Design: longitudinal study Participants: 15,727 Evidence: moderate
Full summary, score breakdown and related studies → Original source
Neuroimaging studies that pool data across clinical sites often suffer from site effects — variability in imaging measures that arises from technical heterogeneity across sites as opposed to true biological signal. 5 key findings →
Design: longitudinal study Evidence: moderate
Full summary, score breakdown and related studies → Original source
Open-source deep learning models promise scalable quantification of MRI-visible perivascular spaces (PVS), an emerging marker of glymphatic dysfunction in Alzheimer’s disease (AD). 4 key findings →
Evidence: moderate
Full summary, score breakdown and related studies → Original source
The Alzheimer’s Disease Neuroimaging Initiative (ADNI) is widely used to train machine learning models for Alzheimer’s disease, yet whether it provides consistent ground truth for predictive modeling has not been systematically tested. 3 key findings →
Design: longitudinal study Evidence: moderate
Full summary, score breakdown and related studies → Original source
Immunofluorescence staining of FFPE tissue following Xenium processing also preserves cellular morphology, allowing the morphological complexity of distinct cell types to be captured and quantitated, especially for glial cells with highly ramified processes. 5 key findings →
Evidence: low
Full summary, score breakdown and related studies → Original source
Subcortical brain alterations are a key feature of dementia disease progression. 1 key finding →
Evidence: low
Full summary, score breakdown and related studies → Original source
Artificial-intelligence (AI) systems for Alzheimer’s disease (AD) increasingly use neuroimaging or spontaneous speech, but reported performance is difficult to interpret across studies because cohorts, endpoints, languages, preprocessing, and validation protocols differ. 5 key findings →
Evidence: low
Full summary, score breakdown and related studies → Original source
Deep learning classifiers trained on structural magnetic resonance imaging (MRI) are increasingly proposed for early detection of Alzheimer’s disease (AD) and mild cognitive impairment (MCI), yet the pipelines that produce these models depend on shared datasets, pre-trained weights, and federated updates that an adversary can poison. 1 key finding →
Evidence: low
Full summary, score breakdown and related studies → Original source
Digital medication-management tools for Alzheimer's disease (AD) typically record only whether a dose was taken, leaving the temporal and behavioral structure of adherence largely unused as a predictive signal. 2 key findings →
Evidence: moderate
Full summary, score breakdown and related studies → Original source
The reproducibility crisis in biomedical research calls for re-evaluation of how common analytic approaches affect results. 3 key findings →
Evidence: moderate
Full summary, score breakdown and related studies → Original source
Of 22,025,599 individuals screened, 2,051,275 were eligible; matched cohorts ranged from 10,113 to 857,773, with median follow-up of 4.7 to 7.2 years. 5 key findings →
Design: cohort study Evidence: high
Full summary, score breakdown and related studies → Original source
Model-inferred and reference diagnoses showed exact agreement rates from 18.8% to 56.2% with poor to slight chance-corrected agreement (Cohen’s κ = −.178 to .196). 5 key findings →
Evidence: low
Full summary, score breakdown and related studies → Original source
Natural language processing has been examined as a possible biomarker for cognitive decline and dementia, but it has also been associated with components of well-being and psychological distress. 3 key findings →
Evidence: low
Full summary, score breakdown and related studies → Original source
Validated on synthetic ground truth, the audit separates a generalizable anchor (PTGDS) from a dataset-specific signal and a global-expression artefact, providing a reusable control for trajectory-based molecular staging. 5 key findings →
Evidence: moderate
Full summary, score breakdown and related studies → Original source
Among 9,386 propensity score-matched pairs of ND patients and controls, covariates were well balanced. 5 key findings →
Design: retrospective cohort Evidence: moderate
Full summary, score breakdown and related studies → Original source
We examined whether natural language processing and acoustic analysis of spoken biographical recall could identify lexical and temporal speech features associated with language and memory performance in TLE. 5 key findings →
Design: cross-sectional Evidence: moderate
Full summary, score breakdown and related studies → Original source
Alzheimer’s disease (AD) detection from structural magnetic resonance imaging (MRI) using deep learning depends on large, labelled datasets. 1 key finding →
Participants: 295 Evidence: moderate
Full summary, score breakdown and related studies → Original source
Activities of daily living (ADLs) provide important indicators of functional decline in people living with dementia, motivating the need for continuous in-home monitoring. 1 key finding →
Evidence: low
Full summary, score breakdown and related studies → Original source
Free full text links and the Full text may need a subscription label come from Unpaywall, Europe PMC and OpenAlex records of free copies, checked 10 Oct 2026. A study with neither is unknown, not paywalled. Show only studies that are free to read
This is alzhub's own score. It is not an official metric, not an impact factor, and no journal, publisher or ranking body endorses it. We built it so that a reader who is not a scientist can tell a landmark trial from a conference abstract without having to already know the difference. Every point comes from one of 20 components, all of them printed on the study itself, so you can disagree with our arithmetic rather than having to trust the total.
Journals are placed in ordered tiers, never given a numeric impact factor — we could not verify current impact factors from a source we control, and an invented one would be worse than none. Tier 1 is the top-tier general journals (NEJM, The Lancet, JAMA, Nature, Science, Nature Medicine); tier 2 is the leading specialty journals; tier 3 is every other peer-reviewed journal. A separate part scores how much the journal's own papers are cited, which tells the tier-3 journals apart.
Two rules we don't bend. First, we never fill in a missing value: if a record does not state a sample size or a design, that component scores zero, and if we simply have no data for a part, such as a citation count, that part is left out and the study is scored on the parts we could check. The study card says which. Second, anything that is not a peer-reviewed article is capped at 16/100 and carries a warning chip — so a conference abstract or a company press release can never outrank a peer-reviewed paper, no matter how good it looks.
We rank a rigorous null result — a good trial showing that something does not work — above an early positive finding. Someone deciding whether to buy a supplement is served by that, and the rigour is already counted by the design and journal components.
The score is computed from the published data alone and is reproducible: the same file always yields the same numbers.
See the whole rubric drawn out → — four charts covering all 27,380 studies: how the scores are spread and where the cap bites, which of the 20 components the library is strong and weak on, which subjects rest on stronger evidence, and what kinds of study each subject is actually built from.
This is a large sample of the Alzheimer's literature, not an exhaustive index of it. It holds 27,380 records, built mostly from automated pulls of Europe PMC (60.6% of the library) and OpenAlex (18.6%), plus smaller pulls from PubMed, Crossref and medRxiv. If a study, a drug, or a finding isn't here, that is not evidence it doesn't exist in the literature — it may simply not have been pulled in yet.
1,032 records (3.8% of the library) are not peer-reviewed — 962 of those are preprints posted before any journal review, and the rest are conference abstracts, regulatory notices and reports. Every study card on this page states its status plainly, and the score above is capped for anything that isn't a peer-reviewed article.
See the full, machine-readable breakdown of this library's coverage →