An Artificial Intelligence Algorithm for Identifying Gynecologic Cancer Patients in Need of Outpatient Palliative Care

Piloting an Artificial Intelligence Algorithm Used to Identify Patients in Need of Outpatient (or Ambulatory) Palliative Care in an Oncology Population

This clinical trial tests an artificial intelligence (AI) algorithm for its ability to identify patients who may benefit from a palliative care consult for gynecologic cancer that has spread from where it first started to nearby tissue, lymph nodes, or distant parts of the body (advanced). A significant delay in referral to palliative care often occurs among patients with cancer. This delay can lead to poorer symptom management, decreased quality of life, and care that does not align with patient goals or values. AI algorithms are computer programs that use step-by-step procedures to solve a problem. In this trial, an AI algorithm is applied to patients' medical records in order to identify patients with a high burden of disease. Information gathered from this study may help researchers learn whether this AI algorithm is useful for identifying patients who could benefit from outpatient palliative care consultation.

Study Overview

• Status: Completed
• Conditions: Advanced Malignant Female Reproductive System Neoplasm
• Intervention / Treatment: Other: Electronic Health Record Review; Other: Internet-Based Intervention

Detailed Description

PRIMARY OBJECTIVE:

I. To pilot an oncology risk prediction model to identify patients who may benefit from outpatient palliative care consultation to improve symptom management and goal-concordant care in this population.

OUTLINE:

Patients' medical records are reviewed for consideration of palliative care consult using AI algorithm once a week (QW) for 6 months.

• Study Type: Interventional
• Enrollment (Actual): 221
• Phase: Not Applicable

Contacts and Locations

Study Locations

• United States
  • Minnesota
    • Rochester, Minnesota, United States, 55905
      • Mayo Clinic in Rochester

Participation Criteria

Eligibility Criteria

• Ages Eligible for Study: Adult; Older Adult
• Accepts Healthy Volunteers: No

Description

Inclusion Criteria:

• Adult patient in Enhanced, Electronic health record (EHR)-facilitated Cancer Symptom Control (E2C2) with a diagnosis of advanced gynecologic malignancy (International Classification of Diseases [ICD] codes C51 through C58)
• Weekly the reviewers will select patients by looking at patients in sorted order starting with the highest score and proceeding down the list and evaluating each patient for exclusion criteria

Exclusion Criteria:

• Patients that have been seen by palliative care will be excluded for 75 days
• Patients under the age of 18 years
• Patients currently enrolled with hospice

Study Plan

How is the study designed?

Design Details

• Primary Purpose: Screening
• Allocation: N/A
• Interventional Model: Single Group Assignment
• Masking: None (Open Label)

Number of Arms

1

Arms and Interventions

Participant Group / Arm	Intervention / Treatment
Experimental: Screening (AI algorithm); Patients' medical records are reviewed for consideration of palliative care consult using AI algorithm QW for 6 months.	Other: Electronic Health Record Review; Undergo medical record review; Other: Internet-Based Intervention; Use AI algorithm

What is the study measuring?

Primary Outcome Measures

Outcome Measure	Measure Description	Time Frame
Timely identification for need of palliative care	Will be measured as time to the electronic record of consult by the palliative care team in the outpatient setting.	Up to 6 months

Secondary Outcome Measures

Outcome Measure	Measure Description	Time Frame
Number of palliative care consultations	Number of palliative care consultations will be assessed as the number of participants who receive palliative care consultations.	Up to 6 months
Number of advanced care planning notes documented in the electronic health record	Participant electronic health records will be reviewed for the number of advanced care planning notes listed.	Up to 6 months
Number of billing codes International Classification of Diseases, 10th Revision for palliative care	Participant electronic health records will be reviewed for the number of International Classification of Diseases, 10th Revision (ICD-10) billing codes for palliative care.	Up to 6 months
Positive predictive value of screened patients	Will be assessed as the number of patients identified by Artificial Intelligence algorithm who actually received palliative care consultation.	Up to 6 months
Performance metrics on reviewer/oncologist handoff	Will be assessed by agreement statistics and descriptive statistics on time between oncology contact and oncology response.	Up to 6 months

Collaborators and Investigators

• Sponsor: Mayo Clinic
• Investigators: Principal Investigator: Rachel D. Havyer, MD, Mayo Clinic in Rochester

Publications and helpful links

Helpful Links

• Mayo Clinic Clinical Trials

Study record dates

Study Major Dates

• Study Start (Actual): December 11, 2023
• Primary Completion (Actual): June 26, 2024
• Study Completion (Actual): July 26, 2024

Study Registration Dates

• First Submitted: December 8, 2023
• First Submitted That Met QC Criteria: December 13, 2023
• First Posted (Actual): December 26, 2023

Study Record Updates

• Last Update Posted (Actual): April 4, 2025
• Last Update Submitted That Met QC Criteria: April 1, 2025
• Last Verified: April 1, 2025

More Information

Terms related to this study

• Additional Relevant MeSH Terms: Urogenital Diseases; Genital Diseases; Urogenital Neoplasms; Neoplasms by Site; Neoplasms; Female Urogenital Diseases; Female Urogenital Diseases and Pregnancy Complications; Genital Diseases, Female; Genital Neoplasms, Female
• Other Study ID Numbers: 23-008371 (Other Identifier: Mayo Clinic in Rochester); NCI-2023-09943 (Registry Identifier: CTRP (Clinical Trial Reporting Program))

Drug and device information, study documents

• Studies a U.S. FDA-regulated drug product: No
• Studies a U.S. FDA-regulated device product: No
• product manufactured in and exported from the U.S.: No