receipt
pre-intervention median weekly time to first action was 11.3 hours [IQR: 5.6, 15.1] (Site A)Abstract, ResultsmedRxiv: AI Tool Implementation to Assist with Management of Patient Portal Messages: Auto-Categorization and Routing · medrxiv.org
falling post-intervention to 1.0 hour [IQR: 0, 3.1] (Site A)Abstract, ResultsmedRxiv: AI Tool Implementation to Assist with Management of Patient Portal Messages: Auto-Categorization and Routing · medrxiv.org
Because of an AI tool that categorizes and routes patient portal messages, the median time to first action in a Boston epilepsy clinic fell from 11.3 hours to 1.0 hour.
receipt
pre-intervention median weekly time to first action was 11.3 hours [IQR: 5.6, 15.1] (Site A)Abstract, ResultsmedRxiv: AI Tool Implementation to Assist with Management of Patient Portal Messages: Auto-Categorization and Routing · medrxiv.org
falling post-intervention to 1.0 hour [IQR: 0, 3.1] (Site A)Abstract, ResultsmedRxiv: AI Tool Implementation to Assist with Management of Patient Portal Messages: Auto-Categorization and Routing · medrxiv.org
Two Mass General Brigham neurology sites switched on Epic's AI message categorization and routing in their epilepsy divisions in April 2026. A controlled interrupted time series against volume-matched neurology divisions found much faster first action on patient messages, but no significant change in time to closure.
Sources
- primarymedRxiv: AI Tool Implementation to Assist with Management of Patient Portal Messages: Auto-Categorization and Routingmedrxiv.org
Editor's notes
verify: publication date taken from the medRxiv DOI (2026.10.19); the page printed no posting date. verify: the control divisions had no AI tool, but the study is not randomized. verify: the outcome is a process measure (message response time), not a patient health outcome; no patient outcomes were measured. verify: Site B's first-action effect was a downward trend (−43.5% per week), not a significant immediate level change; the headline uses Site A.
History
- Added by the backfill run.
Verification
- Drafted by
- Anthropic
claude-sonnet-5-5 - Verified by
- OpenAI
gpt-6.1-sol· pass - How sure
- Strict
- Record
- data/verifications/mgb-neurology-portal-message-routing/2026-11-13.json · The preprint supports faster initial action on real patient messages against concurrent matched controls, including all quoted numerical results. Message closure did not significantly improve, and the stated follow-up durations conflict with the explicit calendar dates.
Two models from different vendors read the primary source independently; this entry is published because they agreed on every receipt and every grade. The models grading an AI registry have an interest in the answer. Cross-vendor checking and the receipts on this page are how the registry handles that, not a claim to have removed it. How the gate works · Dispute this entry
Cite this entry
“Because of an AI tool that categorizes and routes patient portal messages, the median time to first action in a Boston epilepsy clinic fell from 11.3 hours to 1.0 hour.” Because of AI, updated Nov 13, 2026. https://becauseofai.org/entry/mgb-neurology-portal-message-routing@misc{becauseofai_mgb_neurology_portal_message_routing,
title = {Because of an AI tool that categorizes and routes patient portal messages, the median time to first action in a Boston epilepsy clinic fell from 11.3 hours to 1.0 hour.},
howpublished = {\url{https://becauseofai.org/entry/mgb-neurology-portal-message-routing}},
note = {Because of AI registry, updated Nov 13, 2026},
year = {2026}
}