Alhia shows aggregated, population-level data about what people report on medications — calibrated against clinical trials.
Nothing here is medical advice, a diagnosis, or a recommendation for you. Treatment decisions belong with you and your clinician.
Two sources of truth about every treatment, side by side: the clinical evidence, translated out of journal-speak — and what 180,000+ real people report. We show you both, and the gap between them.
Decades of clinical research, graded for strength and rewritten in plain language — the part almost nobody reads.
Structured from 180,000+ first-hand accounts — the timelines, side effects, and drop-off trials often miss.
Contributing unlocks full demographic detail and every discontinuation driver, permanently. Consent is layered and never bundled — each choice below is separate.
How every number on Alhia is produced — and what it does and doesn't mean.
Community rates come from public forum discussions (licensed API access), user-submitted structured reports, and, at the verified tier, prescription-validated submissions. Trial benchmarks come from registration trials, published RCT pools, and FAERS.
Language-model extraction of drug, outcome, side effects, timing, demographics, and discontinuation reasons — versioned, with every statistic traceable to its source records. Human QA runs on random samples; measured extraction accuracy is published here each quarter.
Badges are computed from segment sample size (HIGH ≥ 3,000 · MED ≥ 500 · LOW below) and extraction certainty. Uncertain parses are down-weighted or excluded.
Community-reported rates are not incidence rates. People who post are systematically unrepresentative: problems are over-reported relative to silent success. That is exactly why every panel shows the trial benchmark beside the community rate. The gap measures the distance between controlled-trial conditions and lived, self-selected reporting — it is signal about experience and perception, not a correction of one number by the other.
Combined age × sex views are modelled from marginal segments under an independence assumption and labelled as such; their confidence badges reflect the smaller estimated n.
Public commitments, in plain English. These are structural, not marketing.
1. The consumer product is free and ad-free, forever.
2. No pharma-sponsored content, ever. No company can pay to alter its drug's profile.
3. Only aggregates are ever sold — never individual records — and only to named buyer categories: pharmaceutical medical-affairs teams, biotech research funds, academic researchers.
4. Commercial use of your contributed data is opt-in, unbundled from account creation, and revocable at any time.
5. Contributors are shown what research their data enabled.
6. Methodology, extraction accuracy, and known limitations are published openly and versioned.
A random sample of the records behind this panel. Every statistic traces to records like these through the versioned pipeline.
The full legal documents will live here. The commitments they encode:
Your email (account only), your structured reports, and — only if you choose the verified tier — a prescription image that is checked, hashed, and deleted. Raw prescription images are never retained.
Anonymized reports join public aggregates. Nothing identifying ever appears anywhere. If you tick the optional "help research along" box, your grouped data may also be licensed to three kinds of buyers — pharmaceutical medical-affairs teams, biotech research funds, and academic researchers. Only aggregates are ever sold, never individual records, you can revoke anytime with effect on all future datasets, and you'll be shown what research your data enabled.
Sell individual records. Sell to advertisers. Accept pharma sponsorship. Let any company alter its drug's data. Use dark patterns on consent — every choice is unbundled by design.