Ovarian Cancer Survival Prediction

Followed on Feb 2024

Image from Leon_02 Gitee

In a project aimed at improving cancer patient outcomes, we sought to enhance prediction precision by integrating multiple data sources.

The goal was to use Graph Neural Networks (GNNs) to analyze cancer patient data, leveraging attention layers to integrate and process diverse data sources effectively.

Action:

• Data Integration: We collected and preprocessed various data sources, including genomic data and clinical records. This comprehensive dataset provided a rich foundation for our analysis.
• Graph Neural Networks: I implemented GNNs to model the complex relationships between different data points. GNNs are particularly well-suited for this task due to their ability to capture the dependencies and interactions within the data.
• Attention Layers: To further enhance the model’s performance, I integrated graph attention layers. These layers allowed the model to focus on the most relevant features and data points, improving its ability to make accurate predictions.
• Model Training and Evaluation: I trained the GNN model on the integrated dataset and evaluated its performance using various metrics. The graph attention layers significantly improved the model’s precision and recall.

The project successfully demonstrated that using GNNs with attention layers can significantly enhance prediction precision for cancer patient outcomes. This approach provided more accurate and reliable predictions, potentially leading to better-informed treatment decisions and improved patient care.