AI and Patient Portal Messages: Drafting Replies That Sound Human and Are Reviewed

End-of-lesson check

15 questions · take it digitally for instant feedback at tendril.neural-forge.io/learn/quiz/end-creators-healthcare-AI-and-patient-portal-replies-r13a6-adults

1. A clinician receives an AI-drafted reply to a patient message and notices the draft suggests a medication change that contradicts the patient's allergies listed in the chart. What should happen next?

   1. Edit the medication suggestion to match the chart and then send
   2. Delete the draft and type a new reply from scratch
   3. Send the draft as-is since the AI generated it
   4. Report the error to Epic technical support

2. What does the lesson identify as the leading driver of physician burnout?

   1. Virtual video visits
   2. Administrative billing tasks
   3. Inbox patient messages
   4. Electronic health record documentation

3. A patient sends a message about new chest pain. The AI draft suggests 'This sounds like acid reflux; try antacids.' Why is this problematic?

   1. The AI does not have enough information to make any treatment suggestion
   2. The AI is not allowed to suggest any treatments
   3. Acid reflux is never an appropriate suggestion for chest pain
   4. Chest pain could indicate urgent cardiac issues requiring clinical judgment

4. What is the recommended target percentage for sending AI drafts completely unedited?

   1. At least 80%
   2. Between 50-70%
   3. Less than 30%
   4. As close to 0% as possible

5. Why is 'The AI drafted it' not a legal defense if a patient is harmed by content in a message?

   1. Healthcare organizations are immune from AI-related claims
   2. The clinician's name is on the message and bears responsibility
   3. AI companies hold all liability for healthcare applications
   4. AI-generated content is not subject to medical malpractice law

6. Which of the following is something AI can do well when drafting patient message replies?

   1. Access the patient's full medical history from all healthcare systems
   2. Determine whether a symptom requires emergency care
   3. Pull recent labs and visit history into the reply for context
   4. Diagnose conditions based on patient-reported symptoms

7. A patient mentions they visited an emergency room at a different hospital last week. How might this affect AI-drafted replies?

   1. The patient must manually upload those records to the portal
   2. The AI will ask the patient to confirm the visit before drafting
   3. The AI cannot access records from other healthcare systems
   4. The AI will automatically incorporate that information

8. Why might patients be able to tell when no one has actually read their message?

   1. AI cannot fully replace the clinician-patient relationship; patients sense authenticity
   2. The portal automatically notifies patients of AI involvement
   3. Patients can view AI usage logs in the portal
   4. AI drafts always include a visible marker

9. When editing an AI-drafted message, what specific type of content must clinicians verify against the patient's chart?

   1. Signature and credentials
   2. Spelling and grammar
   3. Clinical claims and recommendations
   4. Formatting and layout

10. Which of the following is explicitly listed as something AI cannot do when drafting patient messages?

    1. Generate empathetic acknowledgment of patient concerns
    2. Decide what is urgent
    3. Convert a short reply into a warmer one
    4. Suggest a triage decision

11. A clinician wants to track their use of AI drafting features. What metric should they monitor?

    1. Time spent editing each draft
    2. Number of patients who respond to AI-drafted messages
    3. Number of messages drafted by AI per day
    4. Percentage of drafts sent completely unedited

12. What type of information might an AI incorrectly include in a patient message draft?

    1. An acknowledgment of the patient's concern
    2. A friendly greeting
    3. A request for additional information
    4. Fabricated clinical advice

13. What is one reason the lesson gives for enabling AI drafting features in patient portals?

    1. To eliminate the need for any clinician review
    2. To ensure all patient messages receive immediate responses
    3. To reduce the time clinicians spend on inbox messages
    4. To automatically schedule patient appointments

14. What should a clinician do if an AI draft converts their brief reply into a warm four-sentence message but includes inaccurate information?

    1. Send it anyway since it sounds more empathetic
    2. Edit the inaccurate portions and send the corrected version
    3. Refuse to use the AI drafting feature for complex messages
    4. Request a new draft from the AI system

15. In the context of patient messaging, what does the lesson mean by 'the message becomes discoverable'?

    1. The message can be requested and used in legal proceedings
    2. The message is automatically shared with other providers
    3. The message is visible to the patient in their portal
    4. The message is stored in the cloud backup system