About the data
How this atlas is built
Records last refreshed from ClinicalTrials.gov on July 21, 2026.
Who builds this
The atlas is built and maintained by Delfa, a company that builds patient-recruitment infrastructure for clinical trials. Mapping interventions, sponsors, disease areas, endpoints and recruiting sites is our core operational work, and the atlas applies that same tooling to the public Alzheimer’s disease and related dementias landscape. It is compiled from public records by that team, not automatically scraped and republished; the classifications and site resolution below are ours, and we correct them when they are wrong.
Scope
The atlas follows the National Institute on Aging definition of Alzheimer’s disease and related dementias: Alzheimer’s disease, frontotemporal dementias, Lewy body dementias, vascular contributions to cognitive impairment and dementia, and mixed dementias. It also tracks Alzheimer’s-linked mild cognitive impairment and emerging LATE/TDP-43 disease. General MCI is included only when the registry record contains an Alzheimer’s or ADRD link. Every candidate record receives a stored inclusion or exclusion decision with evidence, and the public view rechecks each record against the current cohort rules. The public Map, Trials and Assets views then apply a therapeutic gate: a study must link to at least one named therapeutic or investigational drug asset. Drug-linked observational and expanded-access studies remain eligible. Reviewed study-level exclusions remove records where a drug is used only as diagnostic preparation or to deliberately provoke impairment, rather than as a candidate therapy; broad registry purpose labels are not treated as reliable enough to make that decision by themselves. Reviewed therapeutic antibodies, vaccines and other biologics are retained when ClinicalTrials.gov types them as biological interventions. Imaging, diagnostic tests, devices, procedures, behavioral programs, dietary supplements, anaesthesia-only agents, challenge probes, pharmacokinetic cocktails, generic drug classes, unnamed arms and other non-therapeutic interventions are excluded from the public atlas. Dose, spelling, brand, code-name and formulation variants are consolidated into one public asset identity when the underlying drug is the same. See the NIA ADRD definition.
Sources
Trial records come from ClinicalTrials.gov v2 API, the public registry maintained by the U.S. National Library of Medicine. Publication citations and identifiers come from each registry record’s reference module; where a PMID is supplied, the atlas links it to PubMed. Intervention modality also comes from the study record. Organization and site enrichment use free authoritative sources including ROR, CMS, and NPPES, with the supporting evidence retained in the database. Nothing in the atlas comes from confidential or paywalled data.
Refresh cadence
The refresh pipeline is repeatable and versioned: it discovers candidates, preserves a new raw snapshot when a registry record changes, re-adjudicates cohort membership, then rebuilds normalized studies, results, publications, and sites. The date at the top of this page and in the footer is the last completed refresh; it does not imply that a registry sponsor updated every underlying record on that date.
How sites are placed on the map
Every dot on the map represents a canonical site entity linked to one or more reported trial-site records; records judged to describe the same site may be consolidated. The atlas uses the best supported coordinate available. A dot is marked as facility-level only when high-confidence, reproducible street-level evidence supports that location. Otherwise it is shown at the reported area, city, or locality, even when the registry supplies a facility name. Organization-registry locality coordinates are not presented as exact buildings. Each site drawer states that precision explicitly. Some sponsors mask their site names in the registry; those sites remain anonymized and approximate. We never infer a specific building from a facility name alone.
Limits
The atlas inherits the registry's limits: sponsors differ in how quickly and how completely they update their records, so enrollment numbers, statuses and site lists can trail reality. Disease-area labels are Delfa's classifications of registry text and can contain errors. Endpoint titles, types, and posted result values are shown from the registry and inherit the limitations of sponsor-reported records.
Corrections
Spotted something wrong, like a misplaced site, a misclassified intervention, or a study we're missing? Email luke@delfa.ai and we'll fix it in the next refresh. The atlas is built by Delfa, which builds patient-recruitment infrastructure for clinical trials.