Adverse Event Case Intake Agent · Use case
How can a drug maker use an AI agent to prepare adverse event cases at intake on its own infrastructure?
An adverse event case intake agent reads an incoming report, sets out the reporter, the patient, the product and the event, flags what is missing for a valid case, and prepares the case entry. A safety officer decides whether the case is valid, serious and expected, and whether the product caused the event. The agent decides none of these, enters no case and reports nothing to an authority.
The report arrives in any form.
The case entry is typed by hand.
- ✗ 01Reports arrive as emails, letters, call notes and forms, in many languages, and each is read and typed into the case entry by hand.
- ✗ 02A reporting deadline runs from the day the company first receives the information, so time spent typing is time taken from assessment.
Where it stops for a person
Always. A safety officer decides validity, seriousness, expectedness and causality.Always. Validity, seriousness, expectedness and causality are decided by a safety officer, and the safety team keeps the date of receipt and every reporting deadline.
It prepares the case entry.
A safety officer decides.
Step · 01
A case processor uploads the source document.
Step · 02
The agent sets out the reporter, the patient, the product and the event, each with the passage it came from.
Step · 03
It checks the product against your product list.
Step · 04
It flags which elements of a valid case are missing or unclear.
Step · 05
It prepares the case entry. A safety officer checks it, decides validity, seriousness, expectedness and causality, and enters the case.
Reads from · connectors
The source document, and your product list indexed as a knowledge source. Any passage it could not read with confidence is marked.
Acts through · MCP servers
Nothing. It writes no case to your safety database, sends no follow-up to a reporter and reports nothing to an authority.
What you set up
- Index the product list and your case intake procedure as a knowledge source.
- Give the agent your case entry form.
- Build the agent in Katonic Studio and test it on source documents your safety team has already processed.
- Publish it to case processors, one product group first.
Where it runs
The agent, its knowledge store and the model run on infrastructure you own or control. You choose the model and serve it yourself, so questions, documents and answers are not sent to a public AI service. Adverse event reports hold patients' health data and stay inside the company.
How you get it today
This page is labelled Use case: the agent is built in Katonic Studio. You describe the job, connect what is named above, add test questions and release the agent once it passes them. When the Catalog holds a tested template for this job, this page will carry the Template label and show how to take it.
Does it decide whether a case is valid or serious?
No. It sets out what the report says and flags what is missing. A safety officer decides validity, seriousness and expectedness.
Does it assess whether the medicine caused the event?
No. Causality is a medical judgement and stays with the safety team.
Does it submit the case?
No. It enters nothing in the safety database and reports nothing to an authority. A person enters and submits every case.