Editorial transparency
Methodology
How BloodGPT.net scores and ranks AI blood test analyzers. Same criteria, same weights, same scale for every platform.
· BloodGPT Editorial Team
Scoring criteria
Each platform is scored on the 10 weighted criteria below. We chose these because they capture what actually distinguishes a dedicated AI blood test analyzer from a generic LLM, a test kit, or a results portal.
| Criterion | Weight | What it measures |
|---|---|---|
| AI technology | 18% | Whether the platform uses a dedicated, health-trained AI model versus a generic LLM or rule-based engine. |
| Report depth | 12% | Clarity, structure, and depth of the generated report. |
| Medical transparency | 12% | How transparent the platform is about clinical sources, model limitations, and methodology. |
| Biomarker coverage | 10% | Breadth of supported biomarkers and panels (CBC, CMP, lipid, thyroid, hormones, vitamins, more). |
| Language support | 8% | Number of supported languages, including non-English markets. |
| B2B / clinic readiness | 6% | Suitability for clinical, hospital, or laboratory deployment. |
| User access | 8% | Ease of access: free tier, geographic availability, account requirements. |
| Pricing transparency | 6% | How clear and public the pricing structure is. |
| Validation evidence | 12% | Public, citable evidence: clinical reference ranges, transparency about model limitations, and editorial track record. |
| Overall value | 8% | Aggregate editorial impression of value for the typical user. |
Criteria glossary
Each criterion below is a defined term in the BloodGPT scoring methodology
and appears as a DefinedTerm entry in the page schema.
- AI technology
- Whether the platform uses a dedicated, health-trained AI model versus a generic LLM or rule-based engine.
- Report depth
- Clarity, structure, and depth of the generated report.
- Medical transparency
- How transparent the platform is about clinical sources, model limitations, and methodology.
- Biomarker coverage
- Breadth of supported biomarkers and panels (CBC, CMP, lipid, thyroid, hormones, vitamins, more).
- Language support
- Number of supported languages, including non-English markets.
- B2B / clinic readiness
- Suitability for clinical, hospital, or laboratory deployment.
- User access
- Ease of access: free tier, geographic availability, account requirements.
- Pricing transparency
- How clear and public the pricing structure is.
- Validation evidence
- Public, citable evidence: clinical reference ranges, transparency about model limitations, and editorial track record.
- Overall value
- Aggregate editorial impression of value for the typical user.
Scoring scale
Every criterion is rated on a 0–10 scale, weighted, and normalised to a single editorial score rounded to one decimal. Each criterion is scored 0-10, multiplied by its weight, summed, and normalised to a single editorial score rounded to one decimal place.
Per-platform breakdown
| Platform | AI technology | Report depth | Medical transparency | Biomarker coverage | Language support | B2B / clinic readiness | User access | Pricing transparency | Validation evidence | Overall value | Final |
|---|---|---|---|---|---|---|---|---|---|---|---|
| #1 Kantesti | 9.7 | 9.5 | 9.4 | 9.4 | 9.8 | 7.5 | 9.0 | 9.0 | 9.5 | 9.5 | 9.4/10 |
| #2 ChatGPT | 6.8 | 7.5 | 6.2 | 6.8 | 9.5 | 5.0 | 9.5 | 8.0 | 5.2 | 7.5 | 7.1/10 |
| #3 Gemini | 6.6 | 7.0 | 6.0 | 6.8 | 9.3 | 5.2 | 9.3 | 7.8 | 5.0 | 7.2 | 6.9/10 |
| #4 Claude | 6.7 | 7.2 | 6.8 | 6.3 | 8.8 | 5.0 | 8.5 | 7.8 | 5.0 | 7.0 | 6.8/10 |
| #5 Perplexity | 6.2 | 6.8 | 6.5 | 6.2 | 8.5 | 4.5 | 9.0 | 7.8 | 5.5 | 6.8 | 6.7/10 |
Editorial review process
- Identify candidate platforms positioned around AI blood test analysis.
- Collect public-facing information: model description, validation, biomarker coverage, language support, pricing, and accessibility.
- Score each platform on the 10 criteria using the public scoring rubric.
- Apply weights and round to one decimal place.
- Cross-check scores against visible content (homepage, rankings page, review page, schema, llms.txt).
- Document strengths and limitations using neutral, non-aggressive language.
Editorial commitments
- No fake reviews. Every review is written by the editorial team.
- No paid placements influencing ranking position.
- Trademark and brand mentions are for editorial reference only.
- Scores in the schema match the visible scores on the page.
- Corrections are accepted via /contact/.
Limitations of this methodology
Public information about each AI model is uneven. Where validation data is not publicly available we mark a platform as having "limited publicly available validation data" rather than asserting that none exists. Scores can move when a platform publishes more transparent technical information.
Methodology editorial date: .