AI Tongue & Constitution Explorer
The public page can explain the experience before the model is live, but it must not fake an upload flow, accuracy number or diagnosis.
What This Tool Is
The Explorer is an educational wellness tool that can organize supported visible tongue-image features into traditional constitution-oriented context. It is not a disease-screening test, medical diagnosis, prescription service or substitute for clinical evaluation.
What the Explorer Is Designed to Show
The output is deliberately layered so an observation cannot silently become a diagnosis or prescription.
Visible Observations
Supported image-level features only—described as what the image appears to show.
Traditional Constitution Context
Educational context inside the defined nine-constitution framework.
Confidence & Limitations
Image quality, uncertainty, unsupported features and model version.
Learning & Routines
Optional educational routes, not “your tongue says buy product X.”
Privacy, Method & Image Quality Come First
Consent should not be hidden after a photo is transmitted.
Review Data Choices
See storage, retention, training choice, processors, deletion and export.
AI Privacy & Consent →Review Validation Status
See model version, dataset governance, metrics, failures and limitations.
Methodology & Limitations →Understand the Framework
The nine constitutions are traditional educational categories, not biomedical diagnoses.
Nine Constitutions →How to Photograph Your Tongue
The exact capture specification belongs to the current model version. General principles help avoid obviously unusable images.
Neutral Light
Avoid colored lighting and heavy color cast.
Focus
Keep the required visible surface in focus.
No Beauty Filters
Avoid filters, color correction and degraded screenshots.
Follow Versioned Guide
Current capture specification not yet published.
Assessment Workspace
This static page does not pretend a live upload/AI backend exists when the deployment state is not supplied.
Upload / Start Assessment
The production application should render the upload interface only when the current model, privacy notice, consent version, failure handling and access rules are active.
What Happens to the Photo?
P070 only shows a short summary. P072 is the controlling AI-feature privacy notice.
Confidence Is Not Disease Probability
If a future result displays a score such as 0.82 or 82%, P074 must define what that score is, how it is calibrated and what it does not mean.
Advisors & Evidence Owners
Project roles do not substitute for technical validation.
Changen Zhou / 周常恩
Verified role: TCM AI / AI Tongue Assessment Academic Advisor.
Xikui Zhang / 张喜奎
Verified role: Chinese Medicine / Product Research Medical Advisor.
Frequently Asked Questions
Short answers preserve the page boundary and route deeper safety or product questions to the correct owner.
Is AI tongue analysis a medical diagnosis?
No.
Can a tongue photo diagnose disease?
Not through the current SanYuan tool.
What can the tool tell me?
Supported visible-feature observations and traditional constitution-oriented educational context, plus confidence and limitations.
Can it prescribe herbs?
No.
Is the result always accurate?
No. Use P074 for measured performance and failure modes.
Can the system refuse my image?
Yes. Poor or unsupported images should be rejected rather than guessed.
Is my photo used to train AI?
Only according to the current explicit training/consent choice in P072.
Can products be recommended?
Only as transparent general wellness discovery—not prescription-like matching.