What we released
On October 19, 2026, we released Open Altio, a cleaned version of the FDA Adverse Event Reporting System (FAERS) from 2004 to today. FDA publishes every version of every report it receives. Open Altio keeps one version per case, removes the cases FDA withdrew, and sets probable duplicates aside rather than deleting them. The 2026Q2 release holds 20,296,780 cases, with 291,749 probable duplicates set aside.
Drug names are split into active ingredients and mapped to RxNorm and WHO ATC classes. Each reaction is checked against the drug's current FDA label from DailyMed. Disproportionality statistics (EBGM with EB05 and EB95, PRR, ROR, RRR and IC) come precomputed, each with a methods note. Every cleaning step and its counts are on the methods page.
Two ways to use it
Researchers, students and safety teams can apply for access to the hosted service at open.altioai.com, for themselves or for their organization. Data scientists can run the whole pipeline on their own machine from the open-source code on GitHub, under the Apache 2.0 license, with no sign-in and no telemetry. Disagree with a rule? Change it in your own copy and rebuild.
Why we made it open
FAERS has been public for twenty years, yet most teams still spend weeks cleaning it before they can ask a question, and each team cleans it differently. We think the cleaning should be shared, inspectable and improved in the open, so that time goes into the science instead. Open Altio builds on the work of the DiAna project at the University of Bologna, the AEOLUS resource from Banda and colleagues, and Erasmus MC's faersdbstats and aioli projects.
What comes next
FAERS is the first public safety database we clean. VAERS, Canada Vigilance and Japan's JADER follow, with the same open method. FAERS reports do not establish that a product caused an event, and Open Altio is a starting point for analysis, not a conclusion.
