INTRODUCTION
The direct oral anticoagulants (DOACs) given to patients in the intensive care unit (ICU) are associated with various clinical outcomes, such as major bleeding episodes or intracranial hemorrhages (ICH). These pose major challenges for managing patients in ICUs. There is a need for early detection and intervention among patients taking DOACs to prevent adverse clinical outcomes. We hypothesized that machine-learning algorithms could be applied to develop a more accurate and user-friendly diagnostic tool that integrates various clinical and laboratory data and considers complex interactions.
METHODS
This was an observational, retrospective study using patients from the MIMIC-IV database. The study included patients who were admitted to the ICU with a DOAC (apixaban, rivaroxaban, dabigatran, or edoxaban) listed as one of the active medications at the time of hospital admission. ICD-9 and ICD-10 codes were utilized to identify patients with ICH (I61, 431). Descriptive statistics were reported for all variables of interest. Various machine-learning models were utilized to predict the outcome variable of ICH. Patients who had expired as discharge dispositions were excluded. Models accounted for variables from the demographic, laboratory, comorbidity, etc. domains. Model performance was evaluated by comparing the accuracy and Area Under the Receiver Operating Characteristic Curve (AUROC) values.
RESULTS
Overall, there were 3597 cases that met the inclusion criteria. Among these, 1517 (42.2%) were female. The majority of the sample was White (70.5%), and 2093 (58.2%) cases had Medicare insurance. The median [Q1, Q3] age of the sample was 74 [65, 83] years. ICH was reported in 245 (6.8%) cases. ICU mortality was reported in 297 cases (8.2%). The proportion of ER admissions among those cases who did not have ICH was 40.7%, and 64.1% among those who had ICH in the ICU (p < 0.05). ICU mortality among patients with an ICH was 24.1% vs. 7.1% among those without an ICH (p < 0.001). There were more females in the ICH group (49.8% vs. 41.6%; p < 0.015), and the patients were older (median age 76 vs. 74 years; p < 0.001). The machine learning model results showed that the XGBoost classifier performed best in predicting mortality in the ICU, with an AUROC value of 91.4%. The top ten variables associated with ICH by feature importance using the ‘Gain’ metric, which measures the improvement in accuracy brought by a feature to the branches it is on, were ICU length of stay, age, heart rate, respiratory rate, oxygen saturation, anion gap, mean arterial blood pressure, temperature, blood urea nitrogen, and platelet count.
CONCLUSION
The XGBoost classifier algorithm improved discrimination and calibration compared to other ML models for predicting ICH in the ICU among patients taking DOACs. It has the potential to reduce delayed diagnosis and overtreatment in clinical practice. Future studies will validate this model in broader clinical settings.
Disclosures
No relevant conflicts of interest to declare.
