Preparing your experience...
Preparing your experience...
Capability Domain
Detect safety signals from real-world adverse event data using the same disproportionality methods regulators use — PRR, ROR, IC, and EBGM. All open source, all verifiable.
Signal detection answers a simple question: is this drug-event combination reported more often than expected? AlgoVigilance queries the FDA Adverse Event Reporting System (FAERS), computes a 2×2 contingency table, and runs four statistical methods that each answer the question from a different mathematical perspective.
PRR
Proportional Reporting Ratio
How many times more often?
Signal threshold: >= 2.0
ROR
Reporting Odds Ratio
What are the odds?
Signal threshold: >= 2.0
IC
Information Component
How many bits of surprise?
Signal threshold: >= 0.0
EBGM
Empirical Bayes Geometric Mean
What does Bayesian analysis say?
Signal threshold: EB05 >= 2.0
∃ = ∂(×(ς, ∅))
existence = boundary(product(observed, expected))
2×2 Contingency Table (semaglutide × pancreatitis, FAERS)
| Pancreatitis | No Pancreatitis | |
|---|---|---|
| Semaglutide | a = 2,068 | b = 76,216 |
| Other drugs | c = 75,999 | d = 19,852,706 |
Observed rate ς
a/(a+b) = 0.0264
2.64% of semaglutide reports mention pancreatitis
Expected rate ∅
c/(c+d) = 0.0038
0.38% background rate across all other drugs
Same inputs, four ∂ operators
∂_ratio
6.93
ς / ∅
How many times more?
PRR
∂_odds
7.09
[ς/(1−ς)] / [∅/(1−∅)]
What are the odds?
ROR
∂_info
2.76
log₂(ς / ∅)
How many bits of surprise?
IC
∂_bayes
6.7
E[λ | a,b,c,d]
What does the data believe?
EBGM
All four agree: strong signal. They differ on magnitude because ∂ is a perspective, not a fact.
Enter any drug-event combination. Results are computed live from the FDA FAERS database via mcp.nexvigilant.com
Try: metformin + lactic acidosis, or ibuprofen + gastrointestinal hemorrhage
Real data from the FDA FAERS database, queried live via AlgoVigilance Station on March 30, 2026.
| Metric | Value | 95% CI Lower | 95% CI Upper | Signal? |
|---|---|---|---|---|
| PRR | 6.93 | 6.63 | 7.23 | |
| ROR | 7.09 | 6.78 | 7.41 | |
| IC | 2.76 | 2.69 | 2.82 | |
| EBGM | 6.70 | 6.35 | 7.12 |
| Drug | Semaglutide (RxCUI 1991302) | RxNav |
| Event | Pancreatitis | FAERS |
| Cases | 2,047 serious reports | openFDA |
| PRR | 6.93 (CI: 6.63-7.23) | OpenVigil |
| ROR | 7.09 (CI: 6.78-7.41) | OpenVigil |
| IC | 2.76 (CI: 2.69-2.82) | AlgoVigilance Compute |
| EBGM | 6.70 (EB05: 6.35) | AlgoVigilance Compute |
| On Label | Yes — W&P 5.2, ADR 6.1, 6.2 | DailyMed |
| Literature | 14 case reports | PubMed |
| Consensus | 4/4 methods agree: strong signal | Pipeline |
Recommendation: Known safety signal — on label. For any new case: initiate causality assessment using Naranjo or WHO-UMC, review for expedited regulatory reporting.
19 FAERS Tools
Search, filter, compare adverse events from FDA's post-market surveillance database
6 Disproportionality Methods
PRR, ROR, IC, EBGM, chi-squared, and full contingency table analysis
5 Regulatory Sources
openFDA, DailyMed, PubMed, OpenVigil, RxNav — all queried live
Connect your AI agent to mcp.nexvigilant.com and run the full 6-step pipeline on any drug-event combination.