Researchers Evaluate Machine Learning and Biomarkers for Predicting Mortality in Intensive Care Heart Failure Patients
A new retrospective study published in Scientific Reports investigates whether machine learning models and blood biomarker ratios can reliably predict in-hospital mortality among critically ill adults with heart failure and atrial fibrillation.
Key points
- Retrospective study examined machine learning performance in ICU patients with heart failure and atrial fibrillation.
- XGBoost model achieved an external AUC of 0.668, showing little advantage over standard logistic regression.
- The lactate-to-hemoglobin ratio correlated with mortality across cohorts, but overall findings are observational.
A research team led by investigators at Bezmialem Vakif University Faculty of Medicine in Istanbul has published a retrospective evaluation examining early prognostic assessment tools for intensive care unit patients suffering from both heart failure and atrial fibrillation. Published in Scientific Reports, the study analyzed deidentified records from large critical care databases, specifically the MIMIC-IV and eICU-CRD registries, to see how effectively machine learning algorithms and specific biomarker combinations can forecast in-hospital death rates.
The investigation focused on adult patients who had valid measurements for lactate, blood urea nitrogen, and hemoglobin within the first six hours of their ICU admission. Out of 3,925 patients in the primary development cohort, 760 died during hospitalization, representing 19.4% of the group, while an external validation cohort of 403 patients saw 125 deaths, or 31.0%. The authors found that an adjusted lactate-to-hemoglobin ratio showed an association with mortality across both cohorts, though a blood urea nitrogen-to-hemoglobin ratio proved less reliable during external testing. A locked XGBoost machine learning model achieved an external area under the receiver operating characteristic curve of 0.668, performing similarly to standard logistic regression models that utilized raw clinical variables and individual biomarkers.
Because this was a retrospective database analysis relying on existing medical records, the findings remain observational and early-stage. The authors note that the machine learning approach did not demonstrate a substantial predictive advantage over simpler, traditional logistic regression methods, highlighting the ongoing challenge of improving mortality risk stratification in complex critical care populations.
- What to know: Researchers tested machine learning and blood biomarker ratios to predict in-hospital mortality in ICU patients with heart failure and atrial fibrillation.
- Key finding: Advanced algorithms like XGBoost offered limited performance gains compared to conventional logistic regression models using raw clinical metrics.
- Next steps: Clinicians and researchers emphasize that database findings require further prospective validation before changing clinical practice. Always consult a qualified healthcare provider for personal medical decisions.
Source Nature Health Sciences — nature.com/articles/s41598-026-79598-5 ↗
Read next
🏥Public HealthPreventive Services Task Force Resumes Meetings Under New Leadership
The U.S. Preventive Services Task Force is set to reconvene this week for the first time in over a year following a significant membership overhaul by Health Secretary Ro…
❤️Conditions & CareBlood-Based Genetic Signature May Predict CAR T-Cell Therapy Success in Lymphoma
Researchers have identified a specific RNA signature in blood plasma that may help predict how well lymphoma patients respond to CAR T-cell therapy.
🧠AI AnalysisUltraprocessed Diets May Impact Brain Reward Systems in Younger Adults
A Virginia Tech study suggests that individuals aged 18 to 21 may experience changes in brain reward processing after consuming ultraprocessed foods, a response not obser…
🧠Mind & Mental HealthAddressing Body Image and the Reality of Gluteal Anatomy
A personal reflection on the social and psychological impact of body shape, specifically regarding the gluteal region, and the limitations of consumer solutions for perce…
❤️Conditions & CareMigraine and Smoking Linked to Increased Early Stroke Risk in Women
A large-scale study suggests that people who have given birth and experience migraines while also smoking face a significantly higher risk of stroke before age 60.
🏃FitnessExpert Advice on Hydration Gear Sparks Debate Among Recreational Runners
Olympic medalist Josh Kerr recently suggested that runners may not require hydration packs for distances under 12 miles, igniting a broader conversation about gear necess…