Artificial Intelligence in Breast Cancer Screening Programs

Not Applicable

Completed

• Conditions: Breast Cancer; Screening

• Registration Number: NCT04949776
• Lead Sponsor: Maimónides Biomedical Research Institute of Córdoba

Brief Summary

The use of artificial intelligence software in breast screening (Transpara®) makes it possible to identify studies with a very low probability of cancer.

The hypothesis raised in this work is that reading strategies based on artificial intelligence (single or double reading only of cases with a score\> 7 with Transpara®), allow reducing the workload of a screening program by more than 50 % with respect to the standard reading of the program (double reading of all cases without Transpara®), without presenting inferiority in terms of detection rates and recalls of the program, both with the use of 2D digital mammography and with the use of tomosynthesis or 3D mammogram.

Detailed Description

Not available

Study Timeline

• Study First Submitted: 2021-06-15
• Study First Submitted QC: 2021-06-30
• Study First Posted: 2021-07-02
• Study Start: 2022-03-15
• Primary Completion: 2024-01-11
• Study Completion: 2024-01-11
• Status Verified: 2025-08
• Last Update Submitted: 2025-08-11
• Last Update Posted: 2025-08-14

Recruitment & Eligibility

• Status: COMPLETED
• Sex: Female
• Target Recruitment: 31301

Inclusion Criteria

Not provided

Exclusion Criteria

1. Women invited to the program who do not agree to participate in the research study by signing the informed consent form.
2. Women with breast prostheses.
3. Women with signs or symptoms of suspected breast cancer.

Study & Design

• Study Type: INTERVENTIONAL
• Study Design: SINGLE_GROUP

Primary Outcome Measures

Name	Time	Method
Detection rate	At the end of the study, at 2 years.	Proportion of women diagnosed with breast cancer among those screened.
Recall or referral rate	At the end of the study, at 2 years.	Proportion of women who, after the screening test, are referred to the breast diagnosis unit.
Assessment of Workload of each strategy	At the end of the study, at 2 years.	The workload of each strategy shall be assessed by multiplying the average time for a reading of that strategy by the total number of readings of that strategy.; The average reading time of a case in each strategy shall be calculated from the measurement of the individual reading time in a sample of 500 cases in each strategy.

Secondary Outcome Measures

Name	Time	Method
Positive predictive value of biopsies	At the end of the study, at 2 years.	Proportion of women with breast cancer among all women undergoing biopsy.
Positive predictive value of Transpara® scores	At the end of the study, at 2 years.	Proportion of breast cancers diagnosed among women with a given score.
Positive predictive value of referrals	At the end of the study, at 2 years.	Proportion of women diagnosed with breast cancer among those referred to the hospital.

Trial Locations

Locations (1)

Hospital Universitario Reina Sofia; 🇪🇸 Córdoba, Spain