![[2602.22902] A Data-Driven Approach to Support Clinical Renal Replacement Therapy](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2602.22902] A Data-Driven Approach to Support Clinical Renal Replacement Therapy
Summary
This study explores a machine learning approach to predict membrane fouling in patients undergoing Continuous Renal Replacement Therapy (CRRT), demonstrating the effectiveness of a tabular data method over traditional temporal models.
Why It Matters
The research highlights the potential of interpretable machine learning in clinical settings, specifically for improving patient outcomes in CRRT. By accurately predicting membrane fouling, healthcare providers can make informed therapeutic adjustments, ultimately enhancing patient management and care quality.
Key Takeaways
- Machine learning can effectively predict membrane fouling in CRRT patients.
- A tabular data approach outperformed LSTM models, indicating explicit temporal modeling may not be necessary.
- Feature selection reduced complexity while maintaining accuracy, enhancing interpretability.
- ADASYN oversampling improved model performance for minority class representation.
- Counterfactual analysis provided actionable insights for clinical decision-making.
Computer Science > Machine Learning arXiv:2602.22902 (cs) [Submitted on 26 Feb 2026] Title:A Data-Driven Approach to Support Clinical Renal Replacement Therapy Authors:Alice Balboni, Luis Escobar, Andrea Manno, Fabrizio Rossi, Maria Cristina Ruffa, Gianluca Villa, Giordano D'Aloisio, Antonio Consolo View a PDF of the paper titled A Data-Driven Approach to Support Clinical Renal Replacement Therapy, by Alice Balboni and 7 other authors View PDF HTML (experimental) Abstract:This study investigates a data-driven machine learning approach to predict membrane fouling in critically ill patients undergoing Continuous Renal Replacement Therapy (CRRT). Using time-series data from an ICU, 16 clinically selected features were identified to train predictive models. To ensure interpretability and enable reliable counterfactual analysis, the researchers adopted a tabular data approach rather than modeling temporal dependencies directly. Given the imbalance between fouling and non-fouling cases, the ADASYN oversampling technique was applied to improve minority class representation. Random Forest, XGBoost, and LightGBM models were tested, achieving balanced performance with 77.6% sensitivity and 96.3% specificity at a 10% rebalancing rate. Results remained robust across different forecasting horizons. Notably, the tabular approach outperformed LSTM recurrent neural networks, suggesting that explicit temporal modeling was not necessary for strong predictive performance. Feature selection f...
