- View
2.3: Monitor and Respond to Identity Risks [15 minutes]
Building on the identity controls established earlier in this section, the next step is to determine whether those controls are working—and what to do when identity activity indicates danger. Monitoring turns authentication, access, and account events into risk signals; response converts those signals into timely, proportionate action.
Turn identity activity into useful risk signals
Identity monitoring is the continuous collection and analysis of events involving users, service accounts, devices, applications, credentials, roles, and sessions. Its purpose is not simply to create more logs. It is to identify activity that could indicate account compromise, misuse, excessive access, or a breakdown in identity controls.
In a healthcare environment, identity events may come from:
-
Identity providers and single sign-on systems
-
Electronic health record systems
-
Clinical, billing, pharmacy, and laboratory applications
-
Privileged access management systems
-
Cloud platforms and software-as-a-service applications
-
Virtual private networks and remote access gateways
-
Human resources and workforce management systems
-
Medical devices or supporting systems that authenticate users
-
Help desk and credential recovery systems
Combining these sources provides context. For example, an identity provider may show that a nurse signed in successfully, while the clinical application shows that the same identity accessed an unusually large number of patient records. Neither event alone necessarily proves misuse, but together they may justify investigation.
A successful sign-in confirms that a credential or authentication method was accepted. It does not prove that the person using it was the legitimate account owner.
Signals that deserve attention
A risk signal is an event or pattern that changes the likelihood that an identity is being misused. Useful signals generally fall into several categories.
| Signal category | Examples | Possible concern |
|---|---|---|
| Authentication | Repeated failures, sign-in from an unusual location, unexpected multifactor authentication prompts | Credential theft, password spraying, or push-notification abuse |
| Account lifecycle | Use of a dormant account, activity after termination, unexplained account creation | Incomplete deprovisioning or unauthorized provisioning |
| Authorization | Sudden role elevation, new administrative privileges, access outside normal duties | Privilege escalation or excessive access |
| Behavior | Unusual working hours, rapid access to many records, use of unfamiliar applications | Compromise, misuse, or operational anomaly |
| Session | Concurrent sessions from distant locations, unexpected token reuse, long-lived session activity | Session theft or shared credentials |
| Recovery | Multiple password resets, repeated help desk verification failures, changes to recovery methods | Account takeover attempt |
| Machine identity | Service account used interactively, secret used from a new system, abnormal workload activity | Stolen secret or misconfigured automation |
Context determines whether a signal represents genuine risk. A physician accessing records late at night may be responding to an emergency. A traveling executive may sign in from a new location. A system administrator may receive temporary privileges during an approved maintenance window.
For that reason, monitoring should compare identity activity with information such as:
-
The person’s role, department, and employment status
-
Assigned facility and normal working schedule
-
Device ownership and security status
-
Application sensitivity
-
Patient-care responsibilities
-
Approved change or maintenance records
-
Current access reviews and temporary access approvals
-
Previous activity associated with the identity
Clinical operations also affect response decisions. An action that is appropriate for a billing application may be unsafe if applied without review to an account supporting active patient care.
Events, alerts, incidents, and false positives
These terms describe different stages of monitoring and response:
-
An event is a recorded action, such as a sign-in or role change.
-
An alert is a notification that an event or pattern matched a detection rule.
-
An incident is a confirmed or sufficiently credible security issue requiring coordinated response.
-
A false positive is an alert that appears suspicious but is explained by legitimate activity.
Not every event should generate an alert, and not every alert should become an incident. If teams treat every anomaly as equally urgent, analysts become overloaded and important warnings may be overlooked.
A common misconception is that eliminating all false positives should be the goal. In practice, a detection process that never produces false positives may be too narrow to identify subtle attacks. The better objective is to make alerts sufficiently accurate, contextual, and prioritized so analysts can make dependable decisions.
Assess and prioritize identity risk
Identity risk is commonly evaluated by considering both likelihood and impact. Likelihood asks how strongly the available evidence suggests misuse. Impact asks what could happen if the identity is compromised or abused.
A basic assessment can be expressed conceptually as:
Identity risk = likelihood of misuse × potential impact
This does not require a mathematically perfect score. The purpose is to support consistent decisions.
Evaluate likelihood
Signals that may increase likelihood include:
-
Authentication from an unexpected country followed by local activity minutes later
-
A newly registered authentication method immediately before sensitive access
-
A service account signing in through an interactive user interface
-
Repeated failed sign-ins across many accounts from the same source
-
Privilege elevation without an associated approval
-
An identity exhibiting behavior far outside its established pattern
-
Multiple weak signals occurring close together
Confidence should rise when independent signals reinforce one another. For example, a new device alone may be ordinary. A new device combined with a password reset, a new authentication method, and access to restricted patient data is more concerning.
Evaluate impact
Potential impact depends on what the identity can reach and what business or clinical function it supports. Important questions include:
-
Is the account privileged?
-
Can it access protected health information?
-
Can it modify medication, diagnosis, or treatment information?
-
Can it create users, assign roles, or disable security controls?
-
Is it used by an automated clinical or operational process?
-
Can it access many facilities or applications?
-
Could suspending it interrupt patient care?
-
Is the identity shared, making individual accountability difficult?
An alert affecting a low-privilege training account may be lower priority than the same alert affecting an EHR administrator. Similarly, a compromised service account with broad system access may create greater risk than a normal workforce account even if the service account never accesses a user interface.
Assign a response priority
A practical prioritization model uses severity levels rather than treating all alerts alike.
| Priority | Typical characteristics | Initial response |
|---|---|---|
| Critical | Strong evidence of compromise involving privileged access, clinical safety, or broad sensitive-data exposure | Immediate containment and incident escalation |
| High | Multiple credible signals or suspicious activity involving sensitive systems | Rapid investigation; restrict risky access where appropriate |
| Medium | Meaningful anomaly with limited supporting evidence or impact | Validate context and increase monitoring |
| Low | Weak signal, expected variation, or low-impact identity | Record, correlate, and review if repeated |
Priority should be adjusted as new evidence becomes available. A medium-priority alert may become critical if investigators discover that the identity exported patient data or created a new administrator. A high-priority alert may be closed if the activity matches an approved emergency access procedure.
Risk scoring supports human judgment; it does not replace it. Healthcare response decisions must account for both security exposure and continuity of care.
Investigate and respond proportionately
A dependable response process follows a consistent sequence: validate the signal, determine scope, contain the threat, restore trustworthy access, and learn from the incident.

[[flowchart-placeholder]]
Validate and enrich the alert
The first task is to establish what occurred and whether the alert is credible. Investigators should preserve relevant records and build a timeline before making conclusions.
Useful questions include:
- Which identity, device, application, and resource were involved?
- When did the activity begin, and is it still occurring?
- Was authentication successful, and which method was used?
- Did the account’s privileges or recovery methods change?
- What data or functions did the identity access?
- Is there an approved business, clinical, or maintenance explanation?
- Are other identities showing the same pattern?
- Could an automated process or configuration change explain the activity?
Verification should use trusted communication channels. If an employee’s email account may be compromised, sending a verification message to that same mailbox provides weak assurance. Depending on organizational procedures, verification may instead use a known phone number, the person’s manager, an in-person check, or an approved help desk workflow.
Investigators should also avoid revealing sensitive details unnecessarily. A request such as “Did you sign in at approximately 2:15 a.m.?” is safer than sending a complete list of patient records or security findings to an unverified person.
Determine the scope
A compromised identity can affect more than the system that produced the initial alert. Investigation should consider:
-
Other active sessions and authentication tokens
-
Connected applications using single sign-on
-
Passwords or secrets reused elsewhere
-
Newly created accounts, roles, or access keys
-
Changes to multifactor authentication methods
-
Patient records viewed, changed, printed, or exported
-
Mailbox rules or forwarding settings
-
Actions performed through administrative tools
-
Service accounts or delegated permissions controlled by the identity
Scope also includes time. Investigators may need to review activity before the triggering alert because attackers often test credentials, establish persistence, or collect information before performing an obvious action.
Select a containment action
Containment should reduce immediate risk without causing avoidable operational harm. Available actions range from low disruption to complete access suspension.
| Response action | Appropriate use | Important consideration |
|---|---|---|
| Require reauthentication | Session appears questionable but compromise is not confirmed | Provides limited protection if the credential is already stolen |
| Require stronger authentication | Access conditions have become riskier | The authentication method must be trustworthy |
| Revoke sessions or tokens | Active sessions may be stolen or unauthorized | The user may need to sign in again across applications |
| Reset credentials | Password or secret may be exposed | Reused credentials and connected systems must be considered |
| Remove elevated privileges | Administrative access is unexpected or no longer justified | Preserve enough access for legitimate duties |
| Block a device or location | Risk is associated with a particular source | Avoid simplistic rules that block legitimate clinical access |
| Temporarily disable an account | Strong evidence indicates active compromise | Assess patient-care and system dependencies first |
| Isolate a service identity | A machine credential or workload appears compromised | Determine which applications or devices depend on it |
Complete account disablement is not always the safest first step. If an identity supports an active clinical workflow, abruptly disabling it could interrupt care. The response team may instead revoke suspicious sessions, remove elevated permissions, issue a controlled replacement identity, or coordinate a temporary operational process before disabling the account.
This does not mean that clinical urgency should excuse unsafe access. It means that security, clinical operations, privacy, identity administration, and application owners may need to coordinate containment.
Recover trustworthy access
Recovery is more than unlocking an account. The organization must restore confidence that the correct person or workload controls the identity.
Recovery may include:
-
Verifying the user through an approved identity-proofing process
-
Resetting compromised passwords, keys, certificates, or secrets
-
Removing unauthorized authentication and recovery methods
-
Revoking existing sessions and refresh tokens
-
Reviewing and correcting role assignments
-
Removing unauthorized forwarding rules or delegated access
-
Rebuilding or securing an affected device
-
Reissuing credentials through a trusted channel
-
Confirming that dependent applications function correctly
-
Monitoring the restored identity for recurring activity
For a service account, recovery may require rotating its secret across every dependent system. Changing the secret in the identity platform without updating applications can cause outages. Leaving the old secret active for too long, however, extends the attacker’s opportunity. Coordination and a documented rotation sequence are therefore essential.
Build a repeatable healthcare response process
Consider a nurse account that signs in from a managed workstation at its assigned hospital. Fifteen minutes later, the same identity begins a remote session from an unfamiliar device and accesses an unusually broad set of patient records.
A proportionate response could proceed as follows:
- Correlate the events. Confirm that both sessions belong to the same account and compare devices, authentication methods, locations, and applications.
- Check operational context. Determine whether the nurse changed shifts, moved facilities, used an approved remote workflow, or received temporary responsibilities.
- Verify the user safely. Contact the nurse or supervisor through an established channel rather than through the suspicious session.
- Contain the remote activity. Revoke the unfamiliar session while preserving the verified clinical session if organizational policy and the evidence support doing so.
- Inspect for persistence. Check for newly added authentication methods, changed recovery information, delegated access, or role modifications.
- Assess patient-data exposure. Identify which records were viewed or exported and involve privacy personnel according to the organization’s procedures.
- Restore confidence. Reset affected credentials, require approved authentication, and monitor the identity for further anomalies.
- Document the outcome. Record the evidence, decisions, affected resources, clinical considerations, and final classification.
The result may be a confirmed account takeover, legitimate activity, or an unresolved event requiring additional monitoring. Each outcome still provides useful information for improving future detection.
Define ownership and escalation
Identity response often crosses organizational boundaries. Responsibilities should be established before an incident occurs.
-
The security operations team monitors alerts and coordinates investigation.
-
The identity team revokes sessions, resets credentials, and corrects account configurations.
-
Application owners explain system-specific activity and dependencies.
-
Clinical leadership evaluates patient-care impact.
-
Privacy and compliance teams assess possible exposure of protected information.
-
Human resources supports cases involving workforce status or suspected insider misuse.
-
The help desk follows approved identity-verification and recovery procedures.
-
Incident leadership coordinates severe or widespread events.
Escalation criteria should be explicit. Examples include evidence of privileged account compromise, access to many patient records, alteration of clinical information, compromise of authentication infrastructure, or activity affecting multiple facilities.
Measure whether monitoring is effective
A monitoring program should be evaluated by the quality of its outcomes, not by the volume of alerts it produces. Useful measures include:
-
Time from risky activity to detection
-
Time from detection to initial containment
-
Percentage of alerts receiving the required investigation
-
Percentage of alerts confirmed as incidents
-
Frequency of repeat alerts caused by the same unresolved issue
-
Number of dormant or terminated identities generating activity
-
Completeness of identity-related logs from critical applications
-
Percentage of cases with documented evidence and response decisions
Metrics need context. A lower number of alerts may indicate better tuning, but it could also mean that a log source failed. A higher number of confirmed incidents may reflect worsening risk, improved detection, or both. Trends should therefore be reviewed with changes in systems, staffing, detection rules, and organizational activity.
Post-incident review closes the monitoring loop. Teams should ask whether the alert arrived early enough, whether responders had sufficient context, whether containment created unexpected disruption, and whether identity controls should be changed. Improvements might include adjusting a rule, reducing unnecessary privilege, correcting deprovisioning, strengthening recovery procedures, or adding a missing log source.
Recap
Identity monitoring identifies suspicious authentication, authorization, lifecycle, session, and behavioral activity. Effective response combines those signals with role, device, application, clinical, and operational context.
The essential sequence is to:
-
Validate and enrich the alert
-
Assess likelihood and potential impact
-
Determine the full scope
-
Contain the risk proportionately
-
Restore trustworthy access
-
Document, measure, and improve the process
The central principle is balance: respond quickly enough to limit identity misuse while accounting for the healthcare organization’s responsibility to maintain safe and continuous patient care.