stemedb/uat/consumer-health/glp1-preprint-vs-peer-review.md

UAT: Pre-print vs Peer Review (Multi-Sig Weighting)

Date: YYYY-MM-DD Feature: Multi-Signature Consensus Status: [ ] PASS / [ ] FAIL / [ ] BLOCKED

Scenario

A pre-print on bioRxiv claims GLP-1 drugs have cardiovascular risks. A peer-reviewed Lancet study with expert co-signatures says cardiovascular outcomes are improved.

The system should weight the Lancet study higher due to:

1. Higher source class (Clinical vs Observational)
2. Co-signatures from recognized reviewers

Acceptance Criteria

Criterion	Expected	Met?
Pre-print ingested	Hash returned	[ ]
Peer-reviewed ingested	Hash returned	[ ]
Co-signatures applied	3 signatures on Lancet	[ ]
Authority lens winner	Lancet study	[ ]
Weight differential	Lancet > 2x bioRxiv	[ ]

Test Matrix

Step	Action	Expected	Actual	Status
1	Ingest bioRxiv pre-print	Hash returned		[ ]
2	Ingest Lancet study	Hash returned		[ ]
3	Add co-signatures to Lancet	3 signatures recorded		[ ]
4	Query with Authority lens	Lancet wins		[ ]
5	Check weight ratio	>= 2x		[ ]

Setup Commands

# Start StemeDB
cargo run --bin stemedb-api &
sleep 2

Test Commands

Step 1: Ingest bioRxiv Pre-print (Observational, no co-signs)

curl -X POST http://localhost:18180/v1/assertions \
  -H "Content-Type: application/json" \
  -d '{
    "subject": "Semaglutide:CardiovascularOutcome",
    "predicate": "cv_risk_change",
    "object": {"Number": 1.15},
    "confidence": 0.65,
    "source_class": "Observational",
    "source_hash": "0000000000000000000000000000000000000000000000000000000000000010"
  }'

Expected: {"hash": "..."} Actual: Status: [ ]

Step 2: Ingest Lancet Study (Clinical, peer-reviewed)

curl -X POST http://localhost:18180/v1/assertions \
  -H "Content-Type: application/json" \
  -d '{
    "subject": "Semaglutide:CardiovascularOutcome",
    "predicate": "cv_risk_change",
    "object": {"Number": 0.86},
    "confidence": 0.92,
    "source_class": "Clinical",
    "source_hash": "0000000000000000000000000000000000000000000000000000000000000011"
  }'

Save the returned hash as <LANCET_HASH>.

Expected: {"hash": "..."} Actual: Status: [ ]

Step 3: Add Co-Signatures to Lancet Study

# Expert reviewer 1
curl -X POST http://localhost:18180/v1/votes \
  -H "Content-Type: application/json" \
  -d '{
    "assertion_hash": "<LANCET_HASH>",
    "agent_id": "expert_cardiologist_001",
    "weight": 0.95
  }'

# Expert reviewer 2
curl -X POST http://localhost:18180/v1/votes \
  -H "Content-Type: application/json" \
  -d '{
    "assertion_hash": "<LANCET_HASH>",
    "agent_id": "expert_cardiologist_002",
    "weight": 0.90
  }'

# Lancet editor
curl -X POST http://localhost:18180/v1/votes \
  -H "Content-Type: application/json" \
  -d '{
    "assertion_hash": "<LANCET_HASH>",
    "agent_id": "lancet_editor_review",
    "weight": 0.85
  }'

Expected: 3 votes recorded Actual: Status: [ ]

Step 4: Query with Authority Lens

curl "http://localhost:18180/v1/query?subject=Semaglutide:CardiovascularOutcome&predicate=cv_risk_change&lens=authority"

Expected: Winner is Lancet study (cv_risk_change = 0.86) Actual: Status: [ ]

Step 5: Verify Weight Differential

Query the skeptic lens to see weight shares:

curl "http://localhost:18180/v1/skeptic?subject=Semaglutide:CardiovascularOutcome&predicate=cv_risk_change"

Expected: Lancet weight_share >= 2x bioRxiv weight_share Actual: Status: [ ]

Sign-Off Checklist

• Pre-print assertion stored correctly
• Peer-reviewed assertion stored correctly
• Co-signatures increase effective weight
• Authority lens prefers higher-tier + more signatures
• Weight differential is meaningful (>= 2x)

Notes

Multi-sig validation happens via VoteStore aggregation, not signature count alone.

---

Tester: Date: Result: