Home › Research › Answers › What the library can answer
What this research library can and cannot answer
Every answer page on this site is a count over the same library of study records. This page describes that library honestly: what each record holds, which years and study types it covers, how the counts are made, and the questions it cannot answer.
- 30,980study records in the library
- 1932–2027years published
- 30,898records with an abstract or summary
- 29,938peer-reviewed articles
How this page was made: every figure below is counted from the 30,980 records in alzhub's library as of 2026-11-12, when the page was built. Nothing is typed in by hand. The same figures as JSON.
The library is a large sample of the published research on Alzheimer's disease and related dementias, gathered by searching public indexes (PubMed and Europe PMC, OpenAlex) and summarising each record in plain English. It is not an exhaustive index of the field, and a paper's absence here says nothing about whether it exists.
This page describes what published research in alzhub's library reports. It is not medical advice and does not apply to any one person: a doctor who knows the person's history is the only one who can say what a result or a treatment means for them.
What each record holds
A record is one paper. The table counts how many records carry each field, and what the field is used for. A question that depends on a field most records lack can only be answered for the records that have it.
| Field | Records | Share | What it is used for |
|---|---|---|---|
| Abstract or summary text | 30,898 | 99.7% | the only text an answer page can quote or count words in |
| Key findings list | 30,739 | 99.2% | short sentences drawn from the paper, quoted when the abstract has no figure |
| Study design | 7,672 | 24.8% | the design tables on answer pages count only records that name one |
| Participant count | 10,391 | 33.5% | the median and largest sample figures use only these |
| Journal | 30,977 | 100.0% | the journal tables |
| Year of publication | 30,976 | 100.0% | the year tables and the newest-study lists |
| DOI | 30,609 | 98.8% | a stable link to the paper |
| PubMed ID | 28,830 | 93.1% | a stable link to the PubMed record |
| PubMed Central ID | 4,998 | 16.1% | a free full-text copy exists at PubMed Central |
| Source URL | 30,980 | 100.0% | the link to the paper on each study page |
| Evidence-strength rating | 30,723 | 99.2% | alzhub's own high / moderate / low label in the summary |
| Open-access flag | 10,243 | 33.1% | whether the record says the paper is free to read |
The abstract or summary is the longest text a record holds: median 1,332 characters. 15,032 records carry a citation count from OpenAlex; 10,243 are marked open access; 967 are preprints.
What the records do not hold
Each of these was checked against the records, not assumed. The count is how many records carry the field.
- Full text of the paper: 0 of 30,980 recordsquestions answered only in a paper's methods, tables or supplement cannot be counted.
- MeSH or other controlled subject headings: 0 of 30,980 recordsmatching is by words in the title and abstract, not by indexed subject.
- Extracted effect sizes or outcome data: 0 of 30,980 recordsanswer pages quote the sentence that carries a figure; they do not pool or compare figures.
- Individual participant data: 0 of 30,980 recordsnothing here can say what applies to one person.
Years and study types covered
By decade of publication
| Decade | Records | Share |
|---|---|---|
| 1930s | 2 | |
| 1950s | 1 | |
| 1960s | 9 | |
| 1970s | 24 | |
| 1980s | 291 | |
| 1990s | 1,600 | |
| 2000s | 4,882 | |
| 2010s | 10,312 | |
| 2020s | 13,855 | |
| No year recorded | 4 |
9,119 records are from the last five years of coverage (2023–2027): the library leans recent.
By study design, as recorded
| Design | Records |
|---|---|
| Systematic review or meta-analysis | 1,370 |
| Randomised controlled trial | 514 |
| Other clinical trial | 385 |
| Cohort or longitudinal study | 3,423 |
| Case-control study | 299 |
| Cross-sectional study | 1,451 |
| Other recorded design | 230 |
| No design recorded | 23,308 |
7,672 of 30,980 records name a design. Every "by study design" table on an answer page counts only those, so it is a lower bound, never a share of the literature.
Other counts
| Peer-reviewed articles | 29,938 |
| Preprints | 967 |
| Rated "high" evidence strength in alzhub's summary | 3,138 |
| Rated "moderate" evidence strength in alzhub's summary | 5,831 |
| Rated "low" evidence strength in alzhub's summary | 21,667 |
| No evidence-strength rating | 344 |
| Records stating a participant count | 10,391 |
| Median participants (those records) | 229 |
| Distinct journals | 3,677 |
Where the records come from
| Source index | Records |
|---|---|
| Europe PMC | 20,186 |
| OpenAlex | 5,090 |
| PubMed | 4,222 |
| Europe PMC preprint | 780 |
| not recorded | 364 |
| Crossref | 203 |
| medRxiv | 135 |
| Journals with the most records | Records |
|---|---|
| Neurology | 1,279 |
| Alzheimer's & dementia : the journal of the Alzheimer's Association | 876 |
| Journal of Alzheimer's disease : JAD | 819 |
| Alzheimer s & Dementia | 565 |
| Brain : a journal of neurology | 475 |
| Neurobiology of aging | 431 |
| Alzheimer's research & therapy | 413 |
| Journal of the American Geriatrics Society | 400 |
| Annals of neurology | 343 |
| Archives of neurology | 317 |
Subjects the library is filed under
Each record is filed under one library subject. The subjects with the most records:
- Biomarkers 2,752
- Brain scans and imaging 2,284
- Genetics 2,239
- Risk factors 2,174
- Treatments 1,906
- Disease mechanisms 1,381
- Cognitive tests and assessment 1,173
- Digital health tools 1,094
- Risk genes beyond APOE 1,025
- Prevention and risk 1,006
- Drug development 873
- Inflammation and the immune system 869
- End-of-life care 855
- Tau biomarkers 849
- TREM2 and immune genetics 842
- Prevention trials 836
- Caregiving 835
- Disparities 833
- Getting a diagnosis 815
- Mild cognitive impairment 701
- Down syndrome and Alzheimer's 665
- Health systems and care delivery 654
- Clinical safety 630
- Brain tissue and pathology 612
- Lewy body dementia 567
- Vascular contributions 547
- Long-term cohort studies 542
- How common it is 522
- Hippocampus size on brain scans 492
- Costs and economics 199
- Clinical trials 162
- Laboratory science 46
Questions the library can answer
An answer page exists for every question with at least 10 matching records. 21,645 of the 30,980 records match at least one of them. The count beside each is the number of records its query matches today.
- MCI to dementia conversion rate 1,433
- Lecanemab (Leqembi) 381
- Donanemab (Kisunla) 221
- Aducanumab (Aduhelm) 268
- APOE4 and Alzheimer's risk 1,821
- Amyloid PET scan results 1,606
- Blood tests for Alzheimer's 2,046
- p-tau217 580
- Caregiver support 1,666
- Alzheimer's clinical trials 3,287
- ARIA: brain swelling and bleeds 514
- Donepezil and memantine 673
- Semaglutide and GLP-1 drugs 344
- Hearing loss and dementia 293
- Sleep and Alzheimer's 1,044
- Exercise and dementia 1,092
- Diet and Alzheimer's 1,849
- Is Alzheimer's hereditary? 2,439
- Down syndrome and Alzheimer's 840
- Lewy body dementia vs Alzheimer's 1,393
- Frontotemporal dementia 1,706
- Tau PET scans 821
- Life expectancy after diagnosis 1,743
- Agitation in dementia 824
- Diabetes and Alzheimer's 1,833
- Vascular dementia 1,222
- Early-onset Alzheimer's 583
- Rapid progression in dementia 402
- Loss of smell and Alzheimer's 185
- Depression and dementia 2,066
- Statins and dementia 141
- Metformin and dementia 114
- Hormone therapy (HRT) and dementia 90
- Alcohol and dementia 454
- Smoking and dementia 483
- Head injury and dementia 397
- Blood vs CSF biomarkers 1,071
- Delirium and dementia 475
- Blood pressure and dementia 1,151
What the library cannot answer
- Anything about one person. The records describe groups of study participants. No page here can say what a result means for a particular patient, what dose to take or whether a treatment is right for someone.
- Anything that is only in a paper's full text. Counts and quotes come from titles and abstracts. A result reported only in a table, a figure or a supplement is invisible here.
- Pooled or compared results. An answer page quotes each paper's own figure beside its design and year; it does not combine them into one estimate or decide which paper is right.
- Whether something has never been studied. The library is a sample. A question with few matching records may be under-researched, or its papers may use words the query does not, or they may simply not have been gathered.
- Subjects matched by meaning rather than by words. Matching is by words in the title and abstract. A paper that discusses a subject under a name the query does not list is missed; the query is printed on every page so this can be judged.
- Quality beyond what is recorded. The usefulness score rates design, journal, evidence strength and recency as recorded; it does not re-read the paper, and a record that names no design is scored as such.
How a count is produced
Each answer page defines one query: a case-insensitive pattern over the words in a record's title and abstract, printed on the page in words beside every count. The page's n is the number of records the pattern matches. The study lists are the matched records ordered by title match and then alzhub's usefulness score; the quoted sentences are whole sentences copied from a matched record's abstract or key findings that mention the subject and carry a figure, never cut or paraphrased. Each page exports its matched list as JSON. A query with fewer than 10 matches is not a page.
Questions
- How many studies does the library hold, and from which years?
- 30,980 study records, published between 1932 and 2027; 9,119 of them (29.4%) are from 2023 to 2027. 4 records have no year.
- Does the library hold the full text of each paper?
- No. A record holds the title, an abstract or summary (30,898 records, median 1,332 characters), the journal and year, and identifiers (a DOI for 30,609 records, a PubMed ID for 28,830). Every answer page is computed from titles and abstracts only.
- How is a count on an answer page produced?
- A case-insensitive pattern, printed on the page beside the count, is run over the title and abstract of every record; the count is the number of records it matches. The matched list is exported as JSON next to each page. Nothing is typed in by hand.
- What kinds of study are in the library?
- 7,672 records (24.8%) name a study design; the rest record none, so every design table on this site is a lower bound. 1,370 systematic review or meta-analysis, 514 randomised controlled trial, 385 other clinical trial, 3,423 cohort or longitudinal study, 299 case-control study, 1,451 cross-sectional study.
- Which questions can the library answer?
- 39 answer pages exist, each for a question with at least 10 matching records; 21,645 of the 30,980 records match at least one. A question whose terms do not appear in abstracts, or that needs data the records do not hold, cannot be answered here.
All answer pages → · How studies are scored → · The research library →