Predicting risk of cervical cancer

For a safer, more effective screening programme.

Partner: Kreftregisteret

Description

Applied AI collaborated with the Norwegian Cancer Registry on the Decipher project, which aimed to improve the national cervical cancer screening programme by using machine learning to identify women at higher risk based on their screening history.

The project’s objective was to personalize screening intervals for high-risk individuals, ensuring more frequent monitoring and thereby reducing the chances of cervical cancer development.

Our role included transforming the researchers' experimental code into a robust, maintainable Python package that implemented the machine learning methods for risk prediction, making it accessible and easy to use by medical staff. Additionally, we developed an interactive data visualization tool using Bokeh, which allowed researchers to explore and analyze the complex screening data efficiently. The tool featured advanced filtering capabilities and synchronized visualizations, enabling deeper insights into the data across various dimensions.

Impact

The Python package we delivered made the research code more usable and maintainable, enabling smoother integration of risk prediction models into the screening process.

The interactive visualization tool significantly enhanced the researchers’ ability to explore large datasets, quickly filter for specific criteria, and uncover meaningful patterns.

It streamlined data exploration, reduced the time needed to answer complex research questions, and provided a more intuitive interface for both technical and clinical users. Overall, these contributions helped lay the foundation for a more robust cervical cancer screening program by supporting more targeted and timely follow-up screenings.

Tools & Methods

• Machine Learning in Healthcare
• Software Development
• Advanced data visualisation

Developed by Simula's Department of Applied AI.