AI-Driven Models Enhance Prediction of In-Hospital Mortality for ICU Cancer Patients
Machine learning models, including the CatBoost classifier, show high accuracy in predicting in-hospital mortality among ICU patients with lymphoma, offering valuable insights for personalized risk assessment.
Recent advances in machine learning have demonstrated promising potential in predicting the risk of in-hospital death among ICU patients with lymphoma. A comprehensive study published in PLOS ONE details the development and validation of multiple machine learning models, utilizing data from the Medical Information Mart for Intensive Care IV database. The research involved 1,591 adult ICU patients diagnosed with lymphoma, of whom 342 (21.5%) experienced in-hospital mortality.
The study compared fifteen different machine learning algorithms through receiver operating characteristic and area under the curve (AUC) analyses. Among these, the CatBoost classifier exhibited the highest predictive accuracy with an AUC of 0.7766. Significant predictors identified by the models included blood urea nitrogen (BUN), platelet count, prothrombin time, heart rate, systolic blood pressure, activated partial thromboplastin time, oxygen saturation, and bicarbonate levels.
Particularly noteworthy was the prominent role of BUN, emphasized through SHapley Additive exPlanation (SHAP) analysis, which ranked it as the most influential factor in predicting mortality. The models also provided personalized risk assessments using SHAP force plots, enabling identification of high-risk patient subgroups.
According to the study authors, these machine learning models, especially the CatBoost classifier, could serve as valuable tools for clinicians in estimating mortality risk, supplementing traditional assessment methods. Their interpretability via SHAP analysis offers insights into patient-specific risks, potentially informing more targeted interventions and improving ICU care for lymphoma patients.
This innovative approach highlights the significant role of AI in personalized medicine, ultimately aiming to enhance patient outcomes and clinical decision-making in critical care settings.
Source: https://medicalxpress.com/news/2025-08-ai-hospital-mortality-icu-patients.html
Stay Updated with Mia's Feed
Get the latest health & wellness insights delivered straight to your inbox.
Related Articles
Innovative Approach to Stem Cell Transplant Significantly Improves Outcomes for Blood Cancer Patients
A groundbreaking Australian study introduces a new drug combination post-stem cell transplant that triples survival rates and significantly reduces GVHD risk in blood cancer patients, revolutionizing treatment standards.
Semaglutide Usage May Double the Risk of Developing Neovascular Macular Degeneration
Studies reveal that semaglutide, a popular diabetes medication, may double the risk of developing neovascular age-related macular degeneration in older adults, highlighting the need for eye health monitoring during long-term treatment.
Innovative Cell Cross-Talk Enhances CAR-T Therapy Against Glioblastoma
A novel cell cross-talk strategy boosts CAR-T therapy for glioblastoma by reprogramming the tumor microenvironment with targeted cytokine delivery, improving immune activation and survival in preclinical models.
Increasing Prevalence of Eczema Correlates with Severity of Alopecia Areata
Higher severity of alopecia areata is linked to increased prevalence and risk of eczema, underscoring the importance of monitoring for atopic dermatitis in affected patients.