AI-Generated Instructions for Patient Education

Overseen By Jonah Zaretsky

Age: 18+

Sex: Any

Trial Phase: Academic

Sponsor: NYU Langone Health

Disqualifiers: Inpatient hospice, home hospice

No Placebo Group All trial participants will receive the active study treatment (no placebo)

Compensation: Varies

Number of Visits: 1

What You Need to Know Before You Apply

What is the purpose of this trial?

This trial tests whether AI-generated discharge instructions can help patients better understand their medical care upon leaving the hospital. Patients will receive either these AI-crafted instructions or the usual discharge information. The researchers aim to determine if the AI-generated instructions simplify post-hospital care for patients. Individuals currently in the inpatient medicine unit at NYU Langone Health who can read and understand English may qualify for this trial. As an unphased trial, this study provides a unique opportunity to contribute to innovative research that could enhance patient care.

Will I have to stop taking my current medications?

The trial information does not specify whether you need to stop taking your current medications.

What prior data suggests that AI-generated discharge instructions are safe for patient education?

Research has shown that AI-created discharge instructions are mostly safe, though some safety concerns exist. One study found problems in 18 out of 100 AI-generated instructions. While most instructions were satisfactory, some had issues requiring correction. Efforts are underway to study and address these problems, aiming to ensure that AI-generated instructions remain clear and helpful for patients. ^{1 2 3 4 5}

Why are researchers excited about this trial?

Researchers are excited about AI-Generated Discharge Instructions because they have the potential to significantly improve how patients understand and follow their post-discharge care. Unlike standard discharge instructions, which can often be complex and difficult to interpret, these AI-generated instructions use advanced natural language processing to create patient-friendly, easy-to-understand content. This could lead to better patient compliance and outcomes, as patients are more likely to follow instructions they fully comprehend. By making discharge information more accessible, the AI approach could reduce readmission rates and enhance overall patient satisfaction with their healthcare experience.

What evidence suggests that AI-generated discharge instructions are effective for patient education?

Research has shown that AI-generated discharge instructions can make medical information easier to read and understand. In this trial, some participants will receive discharge instructions generated by GPT-4, an advanced AI model, to enhance readability and understandability. One study found that AI tools like GPT-4 can transform complicated doctors' notes into clear, patient-friendly language. Surveys reveal that many patients find these AI-generated instructions easy to understand. Overall, using AI to create discharge summaries helps patients better grasp their care after leaving the hospital. ^{2 3 4 6 7}

Who Is on the Research Team?

Jonah Zaretsky

Principal Investigator

NYU Langone Health

Are You a Good Fit for This Trial?

This trial is for patients who are being discharged from the hospital and need discharge instructions. There aren't specific inclusion or exclusion criteria provided, so it's likely open to a broad range of patients.

Inclusion Criteria

Admitted to Inpatient Medicine unit at NYULH

Able to read and understand English

Willing and able to consent to this research

Exclusion Criteria

I am currently receiving hospice care at home.

I am currently in a hospice program as an inpatient.

Timeline for a Trial Participant

Screening

Participants are screened for eligibility to participate in the trial

1-2 weeks

Intervention

Patients receive either patient-friendly discharge instructions generated by GPT-4 or standard discharge instructions at the time of discharge

Day 1 (At Discharge)

1 visit (in-person)

Follow-up

Participants are monitored for the readability and understandability of discharge instructions

1 week

What Are the Treatments Tested in This Trial?

Interventions

• AI-Generated Discharge Instructions

Trial Overview The study is testing if AI-generated discharge instructions help patients better than standard ones. Patients will be randomly placed into two groups: one gets the AI-generated info, while the other receives usual care.

How Is the Trial Designed?

Group I: Generative AI Discharge Instructions Experimental Treatment 1 Intervention

At the time of discharge, patients will receive patient-friendly discharge instructions generated by GPT-4, an advanced natural language processing model, with the aim of enhancing patient readability and understandability in the discharge information.

Group II: Standard of Care Discharge Instructions Active Control 1 Intervention

At the time of discharge, patients will receive standard discharge instructions (i.e., not generated by GPT-4).

Who Is Running the Clinical Trial?

NYU Langone Health

Lead Sponsor

Trials

1,431

Recruited

838,000+

Published Research Related to This Trial

The SAFER Care framework, which includes standardized written and verbal discharge counseling, significantly improved caregiver comprehension of discharge instructions from 37% to 62% after implementation, based on surveys from 433 caregivers.

This study highlights the effectiveness of structured discharge counseling in a pediatric inpatient setting, suggesting that similar approaches could benefit other patient populations, especially those with limited English proficiency.

NCBI (Pubmed)

pubmed.ncbi.nlm.nih.gov

SAFER Care: Improving Caregiver Comprehension of Discharge Instructions. Uong, A., Philips, K., Hametz, P., et al.[2022]

In a study of 2,498 acute myocardial infarction patients, those who adhered very carefully to discharge instructions, particularly regarding medication, reported significantly less angina one year later compared to those with poor adherence, indicating that following discharge instructions can improve health outcomes.

Despite high adherence to medication instructions (94% reported very careful adherence), there was no clear link between overall risk factor management behaviors and quality of life, physical functioning, rehospitalization, or mortality, suggesting that while adherence is important, other factors may also influence these outcomes.

NCBI (Pubmed)

pubmed.ncbi.nlm.nih.gov

Risk factor management after myocardial infarction: reported adherence and outcomes. Decker, C., Ahmad, H., Moreng, KL., et al.[2009]

Citations

1. pmc.ncbi.nlm.nih.gov pmc.ncbi.nlm.nih.gov/articles/PMC10928500/

Generative Artificial Intelligence to Transform Inpatient ... This cross-sectional study measures the readability and understandability of artificial intelligence (AI)-generated patient-friendly discharge summaries.

2. clinicaltrials.gov clinicaltrials.gov/study/NCT06711458

Using Generative Artificial Intelligence (AI) to Create ... Patient-friendly hospital discharge instructions generated by GPT-4, an advanced natural language processing model. Participant Group/Arm, Active Comparator ...

3. nyulangone.org nyulangone.org/news/artificial-intelligence-model-makes-doctors-notes-more-patient-friendly

An Artificial Intelligence Model Makes Doctors' Notes More ... An artificial intelligence (AI) tool can convert the text of doctors' notes summarizing patients' hospital visits into accurate lay language, a new study found.

4. jmir.org jmir.org/2024/1/e60336/

Patient-Representing Population's Perceptions of GPT ... Of the 156 survey respondents, GPT-generated discharge instructions were assigned favorable ratings, “agree” and “strongly agree,” more ...

5. asra.com asra.com/news-publications/asra-newsletter/newsletter-item/asra-news/2024/05/02/does-ai-have-a-role-in-discharge-instructions-exploring-the-clinical-utility-of-ai-generated-discharge-instructions-following-interventional-spine-procedures

Does AI Have a Role in Discharge Instructions? Exploring ... To create each set of DI, we provided ChatGPT-3.5 with the following prompts: “Can you write discharge instructions for a patient seen in a ...

6. nature.com nature.com/articles/s41746-024-01336-w

The quality and safety of using generative AI to produce ... Safety issues attributable to the AI-generated response were identified in 18% (18 of 100) of the patient-centred discharge instructions. Other ...

7. shmabstracts.org shmabstracts.org/abstract/enhancing-patient-discharges-with-ai-driven-ehr-integrated-patient-friendly-discharge-summaries-usability-perspectives-from-a-large-academic-health-system/

ENHANCING PATIENT DISCHARGES WITH AI-DRIVEN, ... We conducted a usability study for patients and clinicians to provide feedback on these AI-generated patient friendly discharge summaries.