Human-AI Collaborative Intelligence for Improving Fetal Flow Management

NCT06371859 | Recruiting DMC

• 1 other identifier: 3-3031-2915/1
• Study Type: interventional
• Target: 92
• Locations: 1 country
• Sites: 2

Brief Summary

This randomized controlled study evaluates the effectiveness of explainable AI (XAI) in improving clinicians' interpretation of Doppler ultrasound images (UA and MCA) in obstetrics. It involves 92 clinicians, randomized into intervention and control groups. The intervention group receives XAI feedback, aiming to enhance accuracy in ultrasound interpretation and medical decision-making. Objectives:

1. 1.To develop an interpretable model for commonly used Doppler flows, specifically the Pulsatility Index (PI) of the umbilical artery (UA) and middle cerebral artery (MCA), with the aim to provide quality feedback on Doppler spectrum images and suggest potential gate placements.
2. 2.To test the effects of providing Explainable AI (XAI)-feedback for clinicians compared with no feedback on their accuracy in ultrasound interpretation and management.

Conditions

Healthy

Outcome Measures

Primary Outcomes (1)

• Responses will be reviewed independently by two fetal medicine sonographers, and in case of disagreement between the two experts, a consensus will be reached.

  The accuracy in each group (AI-feedback and without AI-feedback group) was defined as the percentage difference in the number of correctly managed flow images between the two groups, assessed by two fetal medicine sonographers. Correct management was defined as: Correct gate placement (multiple sites possible) AND Correct identification of flow curves that were of adequate quality to allow medical decision-making.

  1 months

Secondary Outcomes (1)

• Accuracy of flow image management among competence groups

  1 months

Study Arms (2)

"XAI feedback on MCA/UA Doppler spectral curves and gate placement suggestions"

EXPERIMENTAL

The XAI feedback group will place a gate on MCA/UA images and evaluate the Doppler spectrum with AI feedback. N=46 clinicians (Clinicians will be divided into two groups (XAI feedback \& No XAI feedback groups) of 46 each, matched for experience levels across hospitals)

Behavioral: "XAI feedback on MCA/UA Doppler spectral curves and gate placement suggestions"

"No XAI feedback"

NO INTERVENTION

The control group will place a gate on MCA/UA images and evaluate the Doppler spectrum without AI feedback. N=46 clinicians (Clinicians will be divided into two groups (XAI feedback \& No XAI feedback groups) of 46 each, matched for experience levels across hospitals)

Interventions

"XAI feedback on MCA/UA Doppler spectral curves and gate placement suggestions" BEHAVIORAL

This study includes 1840 ultrasound images, split into UA and MCA flow and spectrum images, each duplicated for a total of 3680 images to compare explainable AI (XAI) feedback vs. no feedback. The investigators will provide matched sets of 40 images (one for the XAI group and one for the non-XAI group) to participants. Participants are matched based on their level of experience within each hospital (Resident physicians, obstetricians, and gynecologists with obstetric ultrasound experience). All participants are instructed to place gates on the flow images of the umbilical artery and the middle cerebral artery and to assess the quality of the resulting flow curves. Specifically, for flow images, participants must identify the most appropriate gate placement. For spectral flow curves, they are to decide if the curves are of sufficient quality to guide medical management decisions.

"XAI feedback on MCA/UA Doppler spectral curves and gate placement suggestions"

Eligibility Criteria

• Sex: all
• Healthy Volunteers: Yes
• Age Groups: Child (0-17), Adult (18-64), Older Adult (65+)

Sponsors & Collaborators

• Rigshospitalet, Denmark: lead
• Slagelse Hospital: collaborator
• Technical University of Denmark: collaborator

Study Sites (2)

Rigshospitalet

Copenhagen, Capital Region of Denmark, 2100, Denmark

RECRUITING

Slagelse Hospital

Slagelse, Region Sjælland, 4200, Denmark

RECRUITING

Central Study Contacts

Zahra Bashir, MD

CONTACT

71871407; zahra.bashir@regionh.dk

Study Design

• Study Type: interventional
• Phase: not applicable
• Allocation: RANDOMIZED
• Masking: DOUBLE
• Who Masked: PARTICIPANT, INVESTIGATOR
• Masking Details: Sealed envelopes will be used and opened at the time of inclusion of each participant. Each participant will be randomized to either AI support or no support. The allocation of participants will be performed by (XYZ) who does not have access to the randomization order.
• Purpose: HEALTH SERVICES RESEARCH
• Intervention Model: PARALLEL
• Sponsor Type: OTHER
• Responsible Party: PRINCIPAL INVESTIGATOR
• PI Title: Ph.d.student, medical doctor and primary investigator

Model Details: Randomized controlled trial

Study Record Dates

• First Submitted: May 9, 2023
• First Posted: April 17, 2024
• Study Start: April 29, 2024
• Primary Completion: December 1, 2024
• Study Completion: December 1, 2025
• Last Updated: May 6, 2024

Record last verified: 2024-05

Locations

DK (2)