Optimizing Post-Cancer Treatment Prognosis: A Study of Machine Learning and Ensemble Techniques

2025-04-21

AI Summary

This study addresses the challenge of inaccurate prediction of post-chemotherapy recovery duration in cancer patients. Leveraging real-world clinical data from Sylhet, Bangladesh—comprising 15 features including age, cancer stage, and hematologic biomarkers—we systematically evaluate 13 machine learning algorithms and propose, for the first time in a regional cancer cohort, a majority-voting ensemble classifier. Results demonstrate that the ensemble model achieves 77% accuracy, substantially outperforming individual models (e.g., decision tree: 59%); LightGBM and random forest follow closely at ~76%. The findings highlight the limitations of single-model approaches in complex clinical settings and empirically validate the necessity of ensemble strategies to enhance prognostic robustness. This work delivers a deployable, evidence-based tool to support personalized treatment decisions and optimize nursing care in resource-constrained healthcare settings.

Technology Category

• Machine Learning: Ensemble Methods
• Humans and AI: Human-in-the-loop Machine Learning
• Intelligent Robots: State Estimation

Application Category

• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets

Abstract

The aim is to create a method for accurately estimating the duration of post-cancer treatment, particularly focused on chemotherapy, to optimize patient care and recovery. This initiative seeks to improve the effectiveness of cancer treatment, emphasizing the significance of each patient's journey and well-being. Our focus is to provide patients with valuable insight into their treatment timeline because we deeply believe that every life matters. We combined medical expertise with smart technology to create a model that accurately predicted each patient's treatment timeline. By using machine learning, we personalized predictions based on individual patient details which were collected from a regional government hospital named Sylhet M.A.G. Osmani Medical College&Hospital, Sylhet, Bangladesh, improving cancer care effectively. We tackled the challenge by employing around 13 machine learning algorithms and analyzing 15 distinct features, including LR, SVM, DT, RF, etc. we obtained a refined precision in predicting cancer patient's treatment durations. Furthermore, we utilized ensemble techniques to reinforce the accuracy of our methods. Notably, our study revealed that our majority voting ensemble classifier displayed exceptional performance, achieving 77% accuracy, with LightGBM and Random Forest closely following at approximately 76% accuracy. Our research unveiled the inherent complexities of cancer datasets, as seen in the Decision Tree's 59% accuracy. This emphasizes the need for improved algorithms to better predict outcomes and enhance patient care. Our comparison with other methods confirmed our promising accuracy rates, showing the potential impact of our approach in improving cancer treatment strategies. This study marks a significant step forward in optimizing post-cancer treatment prognosis using machine learning and ensemble techniques.

Problem

• Accurately predict post-chemotherapy treatment duration for patients
• Improve cancer care using machine learning and ensemble techniques
• Personalize treatment timelines based on individual patient data

Innovation

• Combined medical expertise with smart technology
• Used 13 machine learning algorithms
• Applied majority voting ensemble classifier