Breast cancer risk assessment for screening: a hybrid artificial intelligence approach
Clinical Scorecard: Evaluating Breast Cancer Risk for Screening Using a Combined Artificial Intelligence Method
At a Glance
| Category | Detail |
|---|
| Condition | Breast cancer risk prediction |
| Key Mechanisms | Integration of clinical risk factors and mammographic image analysis using machine learning and AI models |
| Target Population | Women aged 50–69 undergoing mammographic screening |
| Care Setting | Population-based breast cancer screening programs in clinical radiology settings |
Key Highlights
- Mammographic screening reduces breast cancer mortality and is interpreted by trained radiologists.
- AI models, including CNNs, improve diagnostic performance and reduce radiologist workload.
- Hybrid AI models combining clinical data and mammographic images enhance short-term (2-year) breast cancer risk prediction.
Guideline-Based Recommendations
Diagnosis
- Use full-field digital mammography (FFDM) with double reading and arbitration for screening interpretation.
- Define cases as women with negative mammograms who develop breast cancer within two years confirmed by pathology.
- Include interval cancers diagnosed between screenings via record linkage.
Management
- Apply AI-based risk prediction models to identify high-risk women for targeted screening interventions.
- Consider integrating clinical risk factors (age, family history, breast density) with mammographic features for personalized risk assessment.
Monitoring & Follow-up
- Use biennial screening intervals for women aged 50–69 with negative assessments.
- Monitor high-risk individuals identified by AI models more frequently or with additional diagnostics.
Risks
- Exclude women with prior breast cancer or breast implants from screening risk prediction models.
- Ensure data anonymization and ethical approval for retrospective data use.
Patient & Prescribing Data
Women aged 50–69 participating in biennial breast cancer screening programs without prior breast cancer
AI models can stratify short-term breast cancer risk to guide personalized screening frequency and diagnostic follow-up.
Clinical Best Practices
- Employ combined AI models integrating clinical and mammographic data for improved risk prediction accuracy.
- Use stratified cross-validation to validate AI model performance robustly.
- Maintain adherence to ethical standards and data protection in retrospective studies.
References
