The Care Signals (Smart Notes) feature provides automated analysis of shift notes and form responses, helping you identify potential risks and incidents that might otherwise go unnoticed.
The system scans form responses and event notes (such as progress notes, feedback, and injury notes) and flags concerning entries based on customisable rules.
🤖 This feature utilises Artificial Intelligence (AI) to analyse notes and generate text. While designed to be highly accurate, AI can occasionally misinterpret context or make mistakes. Always review all flagged content and verify any AI-generated details before confirming an action.
How Care Signals Works
When staff members submit or update notes and forms via the web or mobile app, the system scans the text in the background.
If a concern is detected, the note is flagged and assigned a severity level and a confidence percentage:
High (Red): Incident/Emergency, Injury/Fall, Aggression/Violence, Medication Error, Self-Harm, Safeguarding, Staff Injury.
Moderate (Orange): Behavioural Concern, Damage/Maintenance, Medication Concern, Safety Hazard, Transport Issue, Client Distress, Health Concern, Injury Signs, Fall Risk, Empty Mandatory Section*, Insufficient Detail*.
Low (Green): Missing Items, Shift Issue, Equipment Issue, Refusal of Care, Admin Issue, Environmental Concern, Health, Generic/Templated Content*.
*In addition to detecting risks, the AI also checks shift notes for completeness. It cross-references the carer's note against the client's profile data (like care plans, medications, and support needs). If the AI notices that crucial expected information is missing, it will flag it as a documentation gap.
Note: If a staff member edits a previously flagged note to remove the concerning text, the system will re-scan it and automatically remove the flag from your queue.
This feature ensures that important details buried in long progress notes or critical gaps in reporting are brought to the immediate attention of your admin team, even if a formal incident report was not raised.
Navigating Care Signals
To view notes flagged by the system, navigate to Care Signals from the sidebar menu. The number displayed in the menu indicates the total number of high and moderate risk flags.
The page is divided into two distinct views (tabs):
Risk Signals: Focuses on safety, behaviour, incidents, and your top at-risk clients.
Documentation Quality: Focuses on completeness rules (e.g., missing details) and per-staff completeness metrics.
Both dashboards share the same date window selector and filters. If you apply a filter on one tab, it automatically carries over to the other.
By default, the dashboard displays a rolling 30-day window. You can use the quick-select buttons at the top right of the page to switch between 7 days, 30 days, or 90 days.
All the metrics, summary cards, and tables below will instantly update to reflect your selected time window.
Understanding Dashboard Metrics
The metrics on your dashboard calculate the live health of your organisation's reporting.
If an Admin reviews a flag and clicks 'Clear Flag' (marking it as a false positive), it is excluded from all dashboard metrics, percentages, and response times.
Risk Signals Dashboard
Severity Trends (High/Moderate/Low): Count of non-cleared flags per severity in the table.
AVG Response Time: The average and median time it takes your Admin team to action a flag (e.g., confirming it as an incident or clearing it).
Oldest Unactioned: The oldest High Severity flag still sitting in your Open queue.
Confirmed Incidents: The total count of flags you have officially converted into Incident Tickets.
Top at Risk Clients: Lists the clients with the most at-risk flags, along with the specific categories they've been tagged with. You can click on a specific client's name to instantly filter the entire dashboard to show only their data. Click the client's name again to go back to the main view.
If a flagged progress note was written on a multi-client group shift, the flag counts against every client on that shift.
Documentation Quality Dashboard
Rule Trends (Insufficient Detail, Empty Mandatory Section): Count of flags matching each rule tag in the table.
Account Completeness: A score showing what percentage of progress notes written across your entire account passed without a completeness flag.
Most Complete / Incomplete Staff: Ranks your staff members based on their audit readiness. This list excludes staff members who wrote fewer than 5 progress notes in the selected time window. It only lists the Most Incomplete staff if they have received a completeness flag.
Managing Flagged Notes
Below the dashboard metrics, a detailed table lists each flagged entry, including its severity, category, client and staff details, a preview, and the confidence score.
The table is divided into five tabs:
Open: Flags that are awaiting admin review.
Needs Follow up: A queue to park flags that require you to check back in with a staff member or investigate further before making a final decision.
Incident: Flags that an admin has confirmed and converted into an Incident Ticket.
Acknowledged: A queue for flags that are valid and you want to acknowledge, but do not necessarily warrant creating a formal Incident Ticket.
Cleared: Flags that an admin has reviewed and determined were false positives.
Click on any record in the list to view its details. The detail view will show the AI's reasoning for the flag.
If the source was a Shift Note, you will see the full text content.
If the source was a Form Response, you will see the specific Question and Answer pairs that triggered the flag.
You can click 'View Original Content' to open the actual shift note or form response page.
At the top of the page, you can perform several actions:
If the flag was a false positive or requires no action, select 'Clear Flag'. The flag will move to the Cleared tab.
If the note describes an issue that requires formal tracking, click 'Confirm as Incident'. This will convert the flag directly into a new Incident Ticket. Incident Management is a Premium feature. If you are on a standard plan, confirming an incident will open an upgrade modal.
Click 'Acknowledge' or 'Needs Follow Up' to move the flag to the relevant tab.
These actions can also quickly be done from the table.
You can also bulk-manage multiple flags using the checkboxes in the list and clicking 'Bulk Actions'.
Filtering the List
You can click 'More Filters' to drill down into the list. Available filters include specific custom date ranges, severity levels, source types, categories, staff members, clients, teams, and facilities.
Any filter you apply here will automatically update all the metrics and charts on your dashboard to match.
Generating an Incident Ticket
You can create an incident ticket directly from a flagged note in your list.
Click on an open flag and select 'Confirm as Incident', or you can directly click 'Incident' from the Care Signals dashboard.
An incident creation modal will open. The AI will read the original shift note and automatically populate the new Incident Ticket with the following information:
Status: Open
Client and Staff Member
Priority Level
Category
Ticket Name
Description (including the specific flags, confidence percentages, and the contents of the note)
Select the assignee, due date, and attach the associated form and any attachments if needed. Private mode is enabled by default, which can be disabled.
Once you confirm the flag as an incident, you can only attach an answered form. You cannot attach a blank or unanswered form.
Click 'Enhance with AI' to improve the ticket name and description (formatting it cleanly with structured sections: What Happened, Who was involved, Context, and Actions Taken). All prefilled fields remain fully editable. If you are not satisfied with the AI-generated text, click 'Undo' on the AI banner to revert all fields to their original blank state.
Click 'Save' to create an incident ticket and trigger the Incident Report process.
The original flag will move to the Incidents tab on your Care Signals dashboard. If your account has Incident Management enabled (Premium only), you can click 'Manage Incident' to open the new ticket.
Please always review the AI's output and manually add or modify any missing details before saving the ticket.
Customising Detection Rules
The system uses a set of predefined rules to determine what qualifies as a risk. You can customise these rules by clicking 'Configure Flags' to better suit your organisation's needs.
Here are the various customisations you can perform on this page:
Enable/Disable Flag Rules: Toggle specific rules On or Off. Disabling a rule prevents that category from being flagged in future scans, but existing flags are unaffected.
Edit Rule Descriptions: You can edit the text description of each rule (up to 500 characters) to tell the AI exactly what to look for. If you make a mistake while editing a rule description, you can revert it to the original ShiftCare phrasing by clicking 'Restore to default description'.
Adjust Detection Sensitivity: You can set how strict the AI should be. Setting a higher threshold means only very high-confidence matches will create a flag.
Sensitive: Detect more possible cases
Balanced: Recommended default
Strict: Only strong matches
Custom Keywords (English and Spanish): Add or disable specific keywords for the AI to scan.
Add Actions (Email Alerts): You can configure which rules trigger automated email alerts by clicking '+ Add Action'.
Email Alert: Select this action to send an email to admin users subscribed to Care Signals notifications. If a rule has no action configured, no emails will be sent when it is triggered.
Email Care Signal Author: Particularly useful for Completeness rules, this action will automatically email the staff member who wrote the flagged note.
Customise Settings: Select the risk severity of this flag, add a description, and select specific roles that will be notified when this flag is detected.
Please don't forget to click 'Save' after making the necessary changes.
All changes made to these rules are tracked in an audit log, so you can always see which user made a change and when.
Tip: If you are unsure how to write an effective AI prompt, it is best to leave the rule descriptions on their default settings.
Notifications
A highly-detailed email notification will be sent when a note is flagged, provided the specific rule has the 'Email Alert' action configured in the settings.
You can opt out of receiving email notifications by unchecking 'Care Signal Alerts' in your Subscribed Email Categories (staff details page). Read this guide for in-depth instructions.
You can also choose to receive an instant message via chat by toggling on 'Send Care Signal Alerts via Chat' in Account > Feature Settings.
Stream Chat alerts include a 'View Care Signal' link, allowing you to jump straight to the flag detail page from your messages. Notifications are only sent for Risk flags, not Completeness flags.
You can disable the Care Signals feature at any time in Account > AI Settings. Please read this in-depth guide for more information.
Frequently Asked Questions
Can I enable Smart Notes for just one client or staff member?
No. The feature is account-wide. All notes matching the scanned categories are processed.
Does clearing a flag delete it?
No. Clearing a flag simply moves it to the Cleared tab, and it remains in the database for audit purposes.
Can a flag be reopened after being cleared?
Not currently. Once marked as cleared, it stays in the Cleared tab. However, flagged notes in the Cleared tab can still be confirmed as incidents.
Why is the dashboard showing the same percentage after I cleared a flag?
The total-notes count behind the completeness percentages is cached for up to 1 hour to improve performance. While the individual flags update instantly, the percentages (like Most Complete) may take up to an hour to refresh.
What happens if I edit a note that was already flagged?
The note is automatically re-scanned. If the updated text still contains risk indicators, the existing flag is updated. If the risk content was removed, the flag is deleted and disappears from your list.
Are notes from the mobile app also scanned?
Yes. Any note created or updated through the web or mobile app that belongs to the scanned categories (progress notes, enquiry, feedback, injury, incident) is processed.
Which roles have access to the Care Signals Feature?
Currently, only users with the Admin role can access and manage the Care Signals feature (no custom permissions yet).
For further assistance with Care Signals: AI Smart Note Review, please contact our Support team using online chat when you're logged into ShiftCare or email our Support team at support@shiftcare.com.