Ensemble Learning of Colorectal Cancer Survival Rates

TL;DR

This study develops an ensemble machine learning approach to predict colorectal cancer survival rates, leveraging a unique dataset of patient cellular and physical conditions to improve prognostic accuracy especially for challenging cases.

Contribution

It introduces an ensemble method that enhances survival prediction accuracy by combining multiple models on complex patient data, advancing colorectal cancer prognosis tools.

Findings

Ensemble models outperform individual models in survival prediction.

Agreement among models correlates with higher prediction accuracy.

Significant improvements observed on unseen test data for agreed predictions.

Abstract

In this paper, we describe a dataset relating to cellular and physical conditions of patients who are operated upon to remove colorectal tumours. This data provides a unique insight into immunological status at the point of tumour removal, tumour classification and post-operative survival. We build on existing research on clustering and machine learning facets of this data to demonstrate a role for an ensemble approach to highlighting patients with clearer prognosis parameters. Results for survival prediction using 3 different approaches are shown for a subset of the data which is most difficult to model. The performance of each model individually is compared with subsets of the data where some agreement is reached for multiple models. Significant improvements in model accuracy on an unseen test set can be achieved for patients where agreement between models is achieved.