Data preparation method for machine learning-based breast cancer risk prediction: A Cuban case study

Area of Science:

• Epidemiology
• Public Health
• Machine Learning

Background:

• Breast cancer risk assessment is crucial for public health.
• Developing accurate predictive models requires high-quality, accessible datasets.
• Existing datasets may lack specific demographic or methodological rigor.

Purpose of the Study:

• To present a curated dataset of breast cancer risk factors from Cuban women.
• To facilitate the development and validation of predictive models for breast cancer risk.
• To support machine learning applications in public health and epidemiology.

Main Methods:

• Collected data from 1697 Cuban women between 2001 and 2018.
• Implemented a reproducible methodology for data quality control and variable enrichment.
• Ensured data integrity and compatibility with machine learning techniques.

Main Results:

• An open dataset of breast cancer risk factors is now available.
• The preprocessing methodology ensures data quality, traceability, and consistency.
• Consistent prediction model performance was achieved across multiple metrics post-preprocessing.

Conclusions:

• The dataset serves as a valuable tool for epidemiological studies and risk assessment.
• The implemented methodology ensures the dataset's suitability for machine learning applications.
• This resource can enhance public health strategies for breast cancer prevention and early detection.