Enhancing trainee performance in obstetric ultrasound through an artificial intelligence system: randomized...

Area of Science:

• Medical Imaging
• Artificial Intelligence in Healthcare
• Obstetric Ultrasound Training

Background:

• Performing obstetric ultrasound scans requires specialized skills, posing challenges for inexperienced operators.
• Developing effective training methods is crucial for improving diagnostic accuracy and patient care in obstetrics.

Purpose of the Study:

• To evaluate the effectiveness of the Prenatal Screening Artificial Intelligence System (PSAIS) in enhancing obstetric ultrasound training.
• To assess whether PSAIS improves the efficiency of acquiring and interpreting standard basic obstetric views for trainees.

Main Methods:

• A prospective, randomized controlled study involving 32 residents with no prior obstetric ultrasound experience.
• Trainees were divided into PSAIS-assisted and conventional training groups, undergoing four cycles of practical scanning.
• Proficiency was assessed by expert evaluation of test scans, with primary outcome being the number of cycles to reach ≥80% proficiency.

Main Results:

• The PSAIS-assisted group required significantly fewer training cycles to meet proficiency standards (P=0.037).
• Expert ratings showed superior image acquisition ability in the PSAIS group during later training cycles (P=0.012 and P<0.001).
• AI assistance led to a statistically significant reduction in trainee-expert rating discordance from the first cycle onwards (P<0.013).

Conclusions:

• Artificial intelligence-assisted systems can enhance the effectiveness of obstetric ultrasound training for inexperienced operators.
• PSAIS shows potential to improve trainee efficiency and accuracy in acquiring and interpreting standard obstetric ultrasound views.