AI models show promising results in ICU mortality predictions
Predicting mortality risk in intensive care unit (ICU) patients is a crucial part of medical treatment, and recent research has found artificial intelligence could help clinicians make more efficient and accurate decisions.
A recent study led by academics from Australian Catholic University and Charles Darwin University has demonstrated that artificial intelligence (AI) models can accurately predict mortality risk in intensive care unit (ICU) patients. The research, published in the journal BMJ Health & Care Informatics, compared the performance of machine-learning algorithms in identifying conditions and symptoms affecting ICU mortality.
Traditional tools like the Acute Physiology and Chronic Health Evaluation (APACHE) and the Simplified Acute Physiology Score (SAPS) have been used to estimate ICU mortality. However, these models struggle to capture evolving patient conditions and require frequent recalibration. In contrast, AI algorithms have consistently shown superior accuracy in predicting ICU patient mortality.
Two specific AI models, extra trees (ET) and gradient boosting (GB), achieved prediction accuracies of 98.33% and 98.23%, respectively. To enhance the interpretability of these AI models, the researchers employed additional analysis techniques. The ET model identified hypertension, tumors, endocrine disease, digestive disease, and cardiovascular disease as key factors in predicting mortality.
These findings align with established medical knowledge and could help health care practitioners understand and trust the AI model's decisions.
Lead author Professor Niusha Shafiabady, head of discipline for IT and director of the Women in AI for Social Good lab at Australian Catholic University, emphasized that these AI systems can assist clinicians in identifying high-risk patients who require urgent attention or targeted interventions. By continuously monitoring at-risk patients and supporting proactive care, AI tools can prevent deterioration or adverse events.
The researchers noted that future studies should expand the use of these models and algorithms with larger datasets and across diverse healthcare settings. This would assess the generalizability of the AI models and further solidify their clinical relevance.
The study involved collaboration between researchers from Amirkabir University of Technology in Tehran, the University of New England, the University of Technology Sydney, and Western Sydney University. The findings highlight the potential of AI models in improving mortality predictions and decision-making in ICU settings while addressing the need for explainable AI tools that complement clinical practice.
Written by urgent.news from Medical Xpress's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
