> ## Documentation Index
> Fetch the complete documentation index at: https://docs.roserx.ai/llms.txt
> Use this file to discover all available pages before exploring further.
# Adverse Event Management
> Review, manage, and export AI-detected adverse events for regulatory compliance in the RoseRx Platform.
The Adverse Events page is your pharmacovigilance command center. It surfaces all potential adverse drug reactions detected by the RoseRx Platform's AI across your agents and deployments, giving your team the tools to review, assess, and report them in line with regulatory obligations.
Adverse event management is a critical compliance function. The AI detection system is a support tool — it does not replace your organization's pharmacovigilance processes. Always follow your established procedures for assessment, reporting, and escalation.
## Accessing adverse events
Navigate to **Monitor → Adverse Events** in the left sidebar.
## Filtering events
Use the filter bar to find specific events:
* **Agent** — Filter by a specific agent.
* **Deployment** — Narrow to a specific widget deployment.
* **Severity** — Filter by Low, Medium, High, or Critical.
* **Status** — Filter by New, Reviewing, or Closed.
* **Time Range** — Select 7, 30, 60, or 90 days, a custom date range, or all time.
* **Search** — Free-text search across event details and trigger keywords.
## Adverse events table
The table displays detected events with the following columns:
| Column | Description |
| -------------------- | ------------------------------------------------------------------------------------ |
| **Detected** | When the event was flagged by the AI system |
| **Deployment** | Which widget deployment the conversation occurred in |
| **Agent** | The agent that was involved in the conversation |
| **Conversation** | A link to the full conversation in [Conversations](/guides/monitoring/conversations) |
| **Severity** | Color-coded badge indicating event severity |
| **Status** | Current review status (New, Reviewing, or Closed) |
| **Trigger Keywords** | The words or phrases that triggered the detection |
| **Actions** | Options to review or update the event |
Adverse events link directly to the relevant conversation thread for full context review. Click the conversation link to see the entire conversation that triggered the event.
## Severity levels
Each adverse event is assigned a severity level based on the AI's assessment of the reported reaction:
| Severity | Color | Meaning |
| ------------ | ------ | ------------------------------------------------------------ |
| **Critical** | Red | Life-threatening event or reaction |
| **High** | Orange | Requires hospitalization or significant medical intervention |
| **Medium** | Yellow | Other medically important condition or event |
| **Low** | Blue | Non-serious event that does not meet the above criteria |
## Status workflow
Every adverse event follows a three-stage review workflow:
The event has been detected by the AI system and is awaiting review. All newly detected events start in this status.
A team member has acknowledged the event and is actively assessing it. Use this status to indicate the event is being investigated.
The review is complete. The event has been assessed, documented, and any required actions have been taken.
## Reviewing an event
Click on any event row to open the **adverse event detail sheet**, which includes:
* **Event ID** — Unique identifier for the event.
* **Severity** — The assigned severity level.
* **Confidence score** — A percentage indicating how confident the AI is in its detection (e.g. 85%).
* **Trigger keywords** — The specific words or phrases that triggered the detection.
* **Status management** — Update the status from New → Reviewing → Closed.
* **Reviewer notes** — Add internal notes documenting your assessment and any actions taken.
## Bulk actions
For efficient management of multiple events:
1. Use the checkboxes in the table to select multiple events.
2. Use the **bulk status update** action to change the status of all selected events at once.
This is particularly useful for triaging a large number of new events or closing a batch of reviewed events.
## Exporting adverse events
The RoseRx Platform supports two types of export for adverse events.
### CSV export
Click the **Export CSV** button to download event data in spreadsheet format. You can export:
* **Selected items** — Only the events you've checked in the table.
* **All matching** — All events matching your current filters (maximum 200 events per export).
### Regulatory export
Regulatory exports must be reviewed by qualified pharmacovigilance personnel before submission to regulatory authorities. These exports are formatted for regulatory systems but require human review and validation.
The RoseRx Platform can generate adverse event reports in four internationally recognized regulatory formats:
| Format | Description | Use case |
| ------------------------ | ---------------------------------------------------------- | ------------------------------------------------ |
| **E2B R2 (XML)** | ICH SGML format for individual case safety reports | Legacy systems still using the R2 standard |
| **E2B R3 (HL7 XML)** | Current ICH standard for electronic safety reporting | Modern regulatory submissions |
| **MedWatch (FDA 3500A)** | US FDA adverse event reporting form | Reporting to the US Food and Drug Administration |
| **TGA Blue Card** | Australian Therapeutic Goods Administration reporting form | Reporting to the Australian TGA |
Select the appropriate export format based on the regulatory authority you need to report to. Your pharmacovigilance team will be familiar with which format is required for each jurisdiction.
## Pagination
The adverse events table displays **50 items per page**. Use the pagination controls at the bottom of the table to navigate between pages.
## Tips for adverse event management
Review new adverse events as soon as possible. Timely assessment is a regulatory requirement in most jurisdictions.
Document your assessment in the reviewer notes field. This creates an audit trail that supports your compliance obligations.
Always review the full conversation before making a severity determination. Context matters.
If you're seeing too many false positives, consider adjusting the adverse event sensitivity in your [agent's AE monitoring settings](/guides/agents/adverse-event-monitoring).