Unsupervised machine learning analysis to identify patterns of ICU medication use for fluid overload prediction

Abstract

Introduction: Intravenous (IV) medications are a fundamental cause of fluid overload (FO) in the intensive care unit (ICU); however, the association between IV medication use (including volume), administration timing, and FO occurrence remains unclear.

Methods: This retrospective cohort study included consecutive adults admitted to an ICU ≥72 hours with available fluid balance data. FO was defined as a positive fluid balance ≥7% of admission body weight within 72 hours of ICU admission. After reviewing medication administration record (MAR) data in three-hour periods, IV medication exposure was categorized into clusters using principal component analysis (PCA) and Restricted Boltzmann Machine (RBM). Medication regimens of patients with and without FO were compared within clusters to assess for temporal clusters associated with FO using the Wilcoxon rank sum test. Exploratory analyses of the medication cluster most associated with FO for medications frequently appearing and used in the first 24 hours was conducted.

Results: FO occurred in 127/927 (13.7%) of the patients enrolled. Patients received a median (IQR) of 31 (13-65) discrete IV medication administrations over the 72-hour period. Across all 47,803 IV medication administrations, ten unique IV medication clusters were identified with 121-130 medications in each cluster. Among the ten clusters, cluster 7 had the greatest association with FO; the mean number of cluster 7 medications received was significantly greater in patients in the FO cohort compared to patients who did not experience FO (25.6 vs.10.9. p<0.0001). 51 of the 127 medications in cluster 7 (40.2%) appeared in > 5 separate 3-hour periods during the 72-hour study window. The most common cluster 7 medications included continuous infusions, antibiotics, and sedatives/analgesics. Addition of cluster 7 medications to a prediction model with APACHE II score and receipt of diuretics improved the ability for the model to predict fluid overload (AUROC 5.65, p =0.0004).

Conclusions: Using ML approaches, a unique IV medication cluster was strongly associated with FO. Incorporation of this cluster improved the ability to predict development of fluid overload in ICU patients compared with traditional prediction models. This method may be further developed into real-time clinical applications to improve early detection of adverse outcomes.

Keywords: critical care; fluid overload; machine learning; medication regimen complexity; prediction.

Figure 1.

Workflow for unsupervised analysis of medications for prediction of fluid overload

Figure 2.

Venn diagrams of medication overlap within the 10 clusters by medication name & timing Venn Diagrams Illustrating Medication Overlaps Between Clusters 1–5 and Clusters 6–10, with Numerical Values Indicating the Count of Shared Medication Administrations (Both medication name and time period of administration)

Figure 3.

Venn diagrams of medication overlap within the 10 clusters by medication name only Venn Diagrams Illustrating Medication Overlaps Between Clusters 1–5 and Clusters 6–10, with Numerical Values Indicating the Count of Shared Medication Names

Figure 4.

Cluster 7 medications organized by timing of medication administrations Medication Record of Cluster 7 Distribution Over 72 Hours, with boxes indicating administration of medication at specific time slot

Figure 5.

Cluster 7 medications organized by frequency and timing of administration Distribution of Medication Records for Cluster 7 Over 72 Hours. Horizontal Axis: Medication Names. Vertical Axis: Frequency of Appearance in Time Slots

Figure 6.

Cluster 7 medications organized by timing of administration and medication class

Figure 7.

Cluster 7 medications organized by proportion of medications from each 3-hour time period The Custer 7 medication administration is distributed over a span of 72 hours, divided into twenty-four three-hour time slots. These slots are arranged clockwise, starting from the 0–3 hour slot and ending at the 68–72 hour slot. The term “area” represents the quantity of medication detected within each respective time slot.

Figure 8

Logistic regression model for Cluster 7 Logistic regression for incidence of fluid overload, including Cluster 7, APACHE II score, and diuretic level

Figure 9.

Visualization of significance of cluster 7 proportion and APACHE II score at 24 hours in logistic regression model in predicting fluid overload.

Figure 10.

Distribution of patients in each group (non-fluid overload versus fluid overload) based on proportion of individual medications that appeared within Cluster 7

Figure 11.

Marginal effect of cluster 7 proportion on fluid overload Likelihood of an individual patient developing fluid overload, normalized to APACHE II score of 14 and no receipt of diuretics.