2164 Clustering and Subgroup Analysis of 91,000 Traumatic Subarachnoid Hemorrhage Hospitalizations Using Machine Learning

Le résumé fourni par la source

INTRODUCTION: Machine learning (ML) based clustering algorithms enable simultaneous consideration of multiple categorical comorbidity and clinical course variables. This study aims to identify patient groups at increased risk of deterioration within a sample of traumatic subarachnoid hemorrhage (SAH) hospitalizations, also providing new insights into length-of-stay and non-routine discharge in these groups. METHODS: The 2015-2021 National inpatient Sample was queried using ICD-10 CM/PCS coding to identify patients with traumatic SAH. An ML clustering analysis evaluated the population based on 49 comorbidities, complications and clinical covariates. Optimal number of clusters was determined using the Davies-Bouldin index (DBI) and Calinski-Harabasz index (CHI). Between-cluster multivariate logistic regression analysis was performed to assess risk of mortality and non-routine discharge. Kruskal-Wallis H-Testing was performed to assess variance in length-of-stay between clusters. Statistical analysis was performed using Python. RESULTS: 91,069 patients were grouped into 4 clusters according to Composite DBI-CHI scoring. Cluster sizes ranged from 2,063-83,921 patients. Mortality ranged from 6.16% in Cluster 1 to 36.00% in Cluster 4. Cluster 4 had the greatest mean prevalence of acute kidney injury (AKI), heart failure (HF) and sepsis than all other clusters, also displaying significantly higher risk of mortality [OR; 5.24, p < 0.001] relative to Cluster 1. Kruskall-Wallis H-testing and post-hoc pairwise testing of length-of-stay distributions showed significant (p < 0.001) differences between all clusters, with the greatest difference in medians occurring when comparing clusters 1 to 2. CONCLUSIONS: Clustering analysis of patients who experienced a traumatic SAH identified 4 distinct groups with unique comorbidity profiles. The group with the highest prevalence of AKI, HF and sepsis had a 5-fold increase in mortality. This approach allows consideration of complex patterns in comorbidities and clinical course features often overlooked by traditional statistical methods, providing a basis for more personalized clinical decision-making in traumatic SAH subpopulations.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.