Clustering of critically ill patients using an individualized learning approach enables dose optimization of mobilization in the ICU

Abstract

Background: While early mobilization is commonly implemented in intensive care unit treatment guidelines to improve functional outcome, the characterization of the optimal individual dosage (frequency, level or duration) remains unclear. The aim of this study was to demonstrate that artificial intelligence-based clustering of a large ICU cohort can provide individualized mobilization recommendations that have a positive impact on the likelihood of being discharged home.

Methods: This study is an analysis of a prospective observational database of two interdisciplinary intensive care units in Munich, Germany. Dosage of mobilization is determined by sessions per day, mean duration, early mobilization as well as average and maximum level achieved. A k-means cluster analysis was conducted including collected parameters at ICU admission to generate clinically definable clusters.

Results: Between April 2017 and May 2019, 948 patients were included. Four different clusters were identified, comprising "Young Trauma," "Severely ill & Frail," "Old non-frail" and "Middle-aged" patients. Early mobilization (< 72 h) was the most important factor to be discharged home in "Young Trauma" patients (OR_adj 10.0 [2.8 to 44.0], p < 0.001). In the cluster of "Middle-aged" patients, the likelihood to be discharged home increased with each mobilization level, to a maximum 24-fold increased likelihood for ambulating (OR_adj 24.0 [7.4 to 86.1], p < 0.001). The likelihood increased significantly when standing or ambulating was achieved in the older, non-frail cluster (OR_adj 4.7 [1.2 to 23.2], p = 0.035 and OR_adj 8.1 [1.8 to 45.8], p = 0.010).

Conclusions: An artificial intelligence-based learning approach was able to divide a heterogeneous critical care cohort into four clusters, which differed significantly in their clinical characteristics and in their mobilization parameters. Depending on the cluster, different mobilization strategies supported the likelihood of being discharged home enabling an individualized and resource-optimized mobilization approach.

Trial registration: Clinical Trials NCT03666286, retrospectively registered 04 September 2018.

Keywords: Critical care; Critical illness; Early ambulation; Patient discharge; Physical therapy modalities.

Fig. 1

STROBE Diagram

Fig. 2

Biplot of the Cluster centers on the first two dimensions of a Principal Components Analysis (PCA). Arrows illustrate the strength and direction of the influence of the variables on the first and the second component of the PCA. The higher the value of a variable, the longer the arrow, and the stronger the influence in the direction of the arrow. The colored ellipses show the cluster centers. The dots in different colors indicate individual patients and their belonging to the clusters. First component of the PCA explains 15.5% of the variance in the data and is highly positively loaded with APACHE, SOFA and frailty scores and highly negatively loaded with Mobility-Transfer-Barthel, GCS, as well as the admission reasons non-traumatic brain injury, tic brain injury and polytrauma. The second component explains 11.8% of the variance in the data and is highly positive loaded with APACHE, non-traumatic brain injury and traumatic brain injury and highly negative loaded with GCS, department and the admission reason postoperative. The red cluster is mainly loaded with high APACHE, SOFA and Clinical Frailty Scale, which is why it is labeled “Severely ill & Frail”. The green cluster is mainly loaded with young age, high Mobility-Transfer-Barthel, and polytrauma, which is why it is labeled “Young Trauma”. The purple cluster is mainly loaded with allocation for postoperative treatment due to old age but low SOFA, APACHE and Clinical Frailty Scale, which is why it is labeled “Old non-frail”. The blue cluster has no specific load from the first or second principal component. Since the cluster’s median age is close to that of the total cohort, it is labeled “Middle-aged”. GCS Glasgow Coma Scale, SOFA Sepsis-related organ failure assessment score, APACHE Acute physiology and chronic health evaluation score

Fig. 3

Synthetic figure summarizing the main findings. The bar charts show the number of patients in each maximum achieved SOMS level according to their discharge disposition. The percentages below the columns show the frequency of patients discharged home of each SOMS level. Numbers are presented as n (%) or median [IQR]. Early mobilization is defined as mobilization within the first 72 h after ICU admission. The reference for early mobilization is “No Early Mobilization,” the reference for maximum SOMS level achieved is “0/1″. ^aModel was corrected for “Hospital admission,” “Body Mass Index (categories),” “Clinical Frailty Scale,” “Other ICU admission reasons,” “Postoperative care” and “SOFA.” ^bModel was corrected for “Hospital admission,” “APACHE,” “Body Mass Index (categories),” “Charlson Comorbidity Index,” “Clinical Frailty Scale,” “Other ICU admission reasons” and “SOFA.” ^cModel was corrected for “Hospital admission,” “Age (categories),” “APACHE,” “Mobility-Transfer-Barthel” and “Department.” ^dModel was corrected for “Clinical Frailty Scale” and “Postoperative care.” ICU Intensive care unit, IQR Interquartile range, SOMS-Score Surgical Intensive Care Unit Optimal Mobilization Score [–20]