Individualized Bayesian Risk Assessment for Cervical Squamous Neoplasia

Area of Science:

• Oncology
• Biostatistics
• Public Health

Background:

• Current cervical cancer screening guidelines primarily rely on recent test results.
• There is a need for improved risk stratification to personalize patient follow-up.
• Individualized risk assessment could enhance the detection of cervical squamous neoplasia.

Purpose of the Study:

• To explore the application of a Bayesian decision science model for quantitative risk stratification.
• To assess individual risk for the development of cervical squamous neoplasia.

Main Methods:

• Developed a dynamic multivariate Bayesian network model.
• Utilized 13 years of cervical screening and histopathologic data from 389,929 women.
• Incorporated liquid-based cytology, high-risk human papilloma virus (HPV) testing, and surgical procedure data.

Main Results:

• Historical data significantly impacted 5-year cumulative risk for cervical intraepithelial neoplasia 3 (CIN3) and squamous cell carcinoma (SCC).
• Prior high-grade squamous intraepithelial lesion cytology results were associated with the highest risk.
• Persistent abnormal cytologic or HPV results and previous precancer diagnoses increased squamous neoplasia risk.

Conclusions:

• Bayesian modeling enables individualized quantitative risk assessments for cervical squamous neoplasia.
• This approach can identify patients at higher risk for significant diagnoses, including rare outcomes like SCC.
• The model offers a powerful tool for personalized risk stratification in cervical cancer prevention.