This methods page documents how Healthy Aging Atlas generated the Supplement Evidence Density Map 2026, including source data, cleanup rules, included fields, and limitations.
This content is for educational purposes only and is not medical advice. These statements have not been evaluated by the FDA. This product is not intended to diagnose, treat, cure, or prevent any disease.
The report uses HAA's DEC-068 v2 supplement-goal evidence matrix: Full 282-supplement matrix re-score. Supersedes supplement-matrix-scored.json (184-supplement, 1629 cells).
The source file reports 2,221 valid scored cells before publication cleanup, 2,442 raw pairings, and formula weights of 35% demand, 25% evidence, 20% commercial intent, 15% corpus-gap lift, and -5% YMYL penalty.
| Step | Count |
|---|---|
| Source matrix cells | 2221 |
| Editorial and category rows removed | 34 |
| Duplicate supplement-goal identities removed | 31 |
| Total rows removed during publication cleanup | 65 |
| Cleaned report denominator | 2156 |
| Field | Meaning |
|---|---|
| supplementName | Normalized supplement or ingredient label. |
| conditionSlug | Mapped marketed goal or claim target. |
| rctCount | Number of PubMed-indexed randomized controlled trials this pipeline matched to the pairing in a capped per-supplement sample. It is not a PubMed census; a zero can be a matching gap. |
| metaAnalysisCount | Number of mapped systematic review or meta-analysis signals in the source matrix. |
| monthlyDemand | Search-demand proxy from DataForSEO or evidence proxy where fresh keyword data was unavailable. |
| evidenceStrength | Matrix-level evidence-strength label used for prioritization, not a clinical recommendation. |
| ymylHazardLevel | Editorial risk label used to prioritize medical-review caution. |
The lead stat is calculated as no-match rows divided by all cleaned rows: 475 / 2,156 = 22%. It reproduces this pipeline's matching result and is an upper bound on the true evidence gap, not a count of pairings without published trials.
The 3+ RCT stat is 820 / 2,156 = 38%. The 10+ RCT stat is 79 / 2,156 = 3.7%. The at-least-one-meta-analysis stat is 1,128 / 2,156 = 52.3%.
Cluster names come from the source matrix and are used to show evidence-density asymmetry across the supplement landscape.
| Cluster | Pairings | Zero R C T | 3+ R C T | 10+ R C T | Meta-analysis |
|---|---|---|---|---|---|
| General Wellness | 463 | 29.6% | 30% | 2.2% | 55.1% |
| Cardiovascular | 254 | 16.5% | 46.9% | 9.4% | 55.5% |
| Metabolic Health | 240 | 20% | 34.2% | 1.7% | 59.2% |
| Immune Support | 230 | 12.6% | 55.2% | 5.7% | 43% |
| Cognitive Health | 171 | 31% | 33.3% | 1.8% | 60.2% |
| Stress & Mood | 150 | 23.3% | 38.7% | 4% | 54% |
| Liver & Detox | 93 | 18.3% | 36.6% | 2.2% | 55.9% |
| Inflammation | 84 | 11.9% | 32.1% | 1.2% | 44% |
| Athletic Performance | 80 | 17.5% | 37.5% | 1.3% | 40% |
| Gut Health | 63 | 19% | 42.9% | 4.8% | 42.9% |
| Energy & Fatigue | 60 | 11.7% | 45% | 1.7% | 28.3% |
| Beauty & Hair | 51 | 7.8% | 54.9% | 5.9% | 35.3% |
| Joint Health | 47 | 14.9% | 36.2% | 6.4% | 59.6% |
| Hormonal Health | 44 | 25% | 29.5% | 4.5% | 61.4% |
| Longevity | 28 | 46.4% | 10.7% | 0% | 71.4% |
The cleaned JSON summary, generated CSV, and QA report are emitted by the generator script so future HAA agents can regenerate the report and compare denominators.
| Resource | U R L |
|---|---|
| Permanent dataset DOI | https://doi.org/10.5281/zenodo.21412552 |
| Zenodo archive | https://zenodo.org/records/21412552 |
| Public report | https://healthyagingatlas.com/research/evidence-density-map-2026/ |
| CSV download | https://healthyagingatlas.com/downloads/evidence-density-map-2026.csv |
| Generator script | |
| QA report | |
| Cleaned JSON |
In the source matrix, probiotic appeared as a condition keyword for several probiotic strains. That describes a product category, not a health outcome, so those rows were removed from the published denominator.
Antioxidant is broad, but HAA has an antioxidant-support claim family. The report keeps it as a marketed claim category and discloses that broad claim categories are not the same as disease outcomes.
Rows marked as editorial rejects were excluded so the public denominator reflects supplement-goal pairings HAA is willing to discuss as consumer-facing supplement topics.
No. RCT count is a density signal only. It does not score risk of bias, sample size, intervention dose, duration, comparator, outcome relevance, or funding source.
Yes. The generator emits the cleaned CSV, JSON summary, and QA report. The displayed percentage is no-match rows divided by all cleaned rows. It reproduces this pipeline result, not the true prevalence of pairings without published trials.
The free guide summarizes supplement evidence, quality signals, and safety questions in a practical checklist format.
Draft research page pending medical/legal review · Editorial policy · Affiliate disclosure
