Research method
How SteadySkin handles evidence
The method starts with an exact claim, then asks which sources support it, what role each source can play, who was studied, and what the evidence still cannot answer. A strong-looking citation does not get to outrun its scope.
Last updated
Start with the claim, not the citation
SteadySkin stores substantive factual statements as individual claim records. Each record names its sources, confidence, limits, review date, and the pages that use it. Changing the wording changes the thing that must be reviewed.
A source is then assessed for that exact claim. A manual may establish a specification without establishing a clinical outcome. A payer policy may establish that payer's written rule without showing what a person's plan will cover.
Source tiers describe source type
Tiers are a starting point, not a quality score. The relevant question is whether a source is fit for the narrow job assigned to it.
| Tier | Label | How it is used |
|---|---|---|
| 1 | Regulatory / guideline | Rules, labels, guidance, and policy from regulators or guideline groups. |
| 2 | Clinical research | Randomized trials, systematic reviews, and other direct clinical research. |
| 3 | Supporting research | Observational, registry, pragmatic, or other supporting research. |
| 4 | Manufacturer | First-party manuals, specifications, warranties, pricing, and regulatory records. |
| 5 | Patient education | Plain-language education from institutions and patient organizations. |
| 6 | Lived experience - not evidence | Questions and lived experience that may shape what we investigate, but do not support a factual claim. |
Evidence role sets the boundary
The same source may be useful in several places, but every source–claim pairing gets one visible role and a plain-language statement of what it supports and does not support.
- clinical
- Helps answer a clinical evidence question within the population and outcome studied.
- coverage
- Describes a payer rule or coverage workflow, not whether care is clinically effective.
- specification
- Describes a product, manual, feature, warranty, or other first-party fact.
- workflow
- Describes how a process works without establishing a clinical outcome.
- marketing
- Records what an interested party says; it is labelled rather than treated as independent proof.
- lived experience
- Surfaces a question or personal perspective. It is not used as clinical evidence.
Confidence stays coarse
Confidence belongs to one exact claim. It does not rank a person's options or predict what will happen to them.
- Well established: Multiple strong, independent sources agree. This still does not predict an individual result.
- Reasonably supported: Relevant evidence supports the claim, with meaningful limits that remain visible.
- Limited evidence: Evidence is small, early, indirect, inconsistent, or otherwise narrow.
- Not established: The claim is discussed or marketed, but the evidence needed to establish it is absent or too weak.
Population and missing reporting stay visible
For outcome, safety, and body-site claims, the record keeps reported age, condition, severity or extent, sample size, geography and setting, skin tone or phototype, race, ethnicity, and body sites. These fields remain separate. One is never inferred from another.
If a source does not provide a field, the record says Not reported. That is a limitation in the evidence report, not a conclusion about who was or was not in the study. See the current population ledger.
Review state is not a truth badge
Draft, reviewed, published, stale, and retracted describe editorial state. They do not turn weak evidence into strong evidence. A page stays out of search while required clinical, owner, legal, accessibility, or publication review remains open.
The editorial policy explains authorship, corrections, and publication boundaries. The source library keeps the source date and limits visible.
Questions before relying on a claim
- What exact statement is the source being asked to support?
- Is the source being used for the job it can actually do?
- Who was studied, and which population fields were not reported?
- What limitation and review date travel with the claim?