Methodology
HearScape separates what is source-backed from what still needs review. The main database focuses on public records that are useful now, while the research backlog keeps earlier, historical, or less-confirmed leads from disappearing. Automated feeds help with recall, but company identity, current status, and evidence strength still depend on source review.
Corrections & Update Requests
Send a source-backed correction if a company, status, trial, indication, or evidence label looks wrong or out of date.
Build-Time Validation Checks
These checks run during the production build. They catch unresolved industry candidates, unknown core fields, missing review metadata, inflated confidence on trial-only records, silo classification gaps, and institution-like sponsors marked as industry companies.
Open source-level evidence library →Open trial activity view →Why devices are a separate atlas
Therapeutics
32 recordsDrug, biologic, gene therapy, regenerative, RNA, steroid/anti-inflammatory, and other pharmacologic programs.
Devices & Neurotechnology
129 recordsA broad market atlas for hearing aids, implants, tinnitus products, diagnostics, fitting tools, accessibility software, and rehabilitation.
Therapies usually leave a structured public trail through trial and regulatory systems. Many devices do not. The device atlas therefore combines regulatory records, manufacturer sources, publications, clinical registries, and manual market review. It is intentionally labeled broad rather than complete.
Current and historical marketed products, related clinical technology, and publicly documented software.
Listing does not establish FDA approval, quality, safety, availability, or suitability.
First-party reviews use six dimensions, enter moderation as pending, and publish only after review.
Coverage guide rules
Show a definite answer only when a national program or current state law provides one.
Label the result plan-specific when deductible, network, benefit cap, or authorization can change the cost.
Use ranges for market context; never present them as a member-specific quote.
Coverage sources are dated in the interface. Original Medicare and CMS implant policies come from CMS; Medicaid child coverage comes from federal EPSDT rules; state commercial mandates link to statutes; plan-specific benefits must be verified in the current Evidence of Coverage or Summary of Benefits.
Data Sources
| Source | Authority | Used For | Refresh / Review |
|---|---|---|---|
| ClinicalTrials.gov | Primary | Trial registrations, phases, enrollment, sponsors, trial IDs | Pipeline rebuild |
| openFDA / FDA labels | Primary | Approved products, labels, application identifiers, manufacturers | Pipeline rebuild |
| Manual curation | Primary | Company/product identity, modality, stage/status, source links, and review notes | Curator update |
| Company and regulatory sources | Primary | Program ownership, current stage/status, marketed-device/product evidence | Manual review |
| Publications | Corroborating | Mechanism, trial result context, device/therapy scientific basis | Manual review |
| AI enrichment | Candidate only | Recall-oriented extraction and triage queues; never final authority by itself | Reviewed before promotion |
Current Source Mix
Live Data Check Limits
Source Hierarchy
- 1Regulatory sourcesFDA, openFDA, device/product labeling, and official regulatory records are highest authority for approval and marketed status.
- 2Clinical registriesClinicalTrials.gov is authoritative for registered study metadata, but not enough by itself to prove current product identity or commercial status.
- 3Company disclosuresPress releases, pipeline pages, annual reports, and investor materials are accepted for ownership, development stage, and program continuity when current and specific.
- 4Peer-reviewed publicationsPublications corroborate mechanisms, trial results, and scientific rationale, especially for devices or completed studies.
- 5AI/generated candidatesGenerated records stay provisional until source-backed fields are normalized into the schema and promoted by manual review.
Entity Resolution Rules
Program identity: records are grouped by canonical program/product name, aliases, trial IDs, sponsor/company disclosures, target gene, modality, and indication. Trial-arm artifacts such as placebo, saline, usual care, and procedure-only controls are excluded or flagged.
Organization normalization: subsidiaries, acquired programs, historic owners, sponsors, and collaborators are preserved in allCompanies, while the current best-known owner or lead sponsor is used as the primary organization when supported.
Company vs. institution: universities, academic hospitals, government institutes, clinical centers, labs, and foundations are not allowed to count as industry sponsors unless a separate commercial sponsor or company entity is supported. The UI labels non-industry records as sponsors/institutions, and the build gate fails if an institution-like primary sponsor is marked industry-sponsored.
Stage and status: clinical phase, marketed status, approval status, active/completed/discontinued state, and unknown state are kept separate so completed trials are not misread as active products.
Promotion threshold: a record should not be treated as strong evidence until core identity, sponsor, modality, stage/status, indication, and at least one strong public source are aligned.
Confidence Scoring
Confidence is a triage score, not a substitute for source review. Verified manual overrides can set confidence directly, while generated records use completeness signals such as company identity, modality, indication, clinical stage, source support, and review flags.
The main database highlights records with stronger public support and practical relevance. The research backlog keeps earlier, historical, lower-confidence, or unresolved records available without mixing them into the default view.