Predicting Mammogram Screening Follow Through with Electronic Health Record and Geographically Linked Data

Area of Science:

• Medical Informatics
• Public Health
• Machine Learning

Background:

• Breast cancer is a leading cause of death, making screening crucial for mortality reduction.
• National guidelines recommend mammogram screening, yet adherence remains a challenge.
• Identifying patients who will not complete screening is vital for effective intervention.

Purpose of the Study:

• To develop an automated system using machine learning to predict mammogram screening non-completion.
• To identify key patient features influencing screening non-completion.
• To enable focused resource allocation for targeted screening interventions.

Main Methods:

• Application of machine learning classification models.
• Analysis of patient data to predict screening non-completion.
• Identification of predictive features for non-adherence.

Main Results:

• A specific patient subgroup at high risk for non-completion was identified.
• Machine learning models demonstrated predictive capability for screening adherence.
• Features influencing prediction were elucidated, enabling targeted outreach.

Conclusions:

• Machine learning can effectively predict mammogram screening non-completion.
• Early identification of at-risk patients allows for focused interventions.
• This approach can enhance screening program effectiveness and potentially reduce breast cancer mortality.