2019
DOI: 10.1007/s00134-019-05790-z
|View full text |Cite
|
Sign up to set email alerts
|

ICU staffing feature phenotypes and their relationship with patients’ outcomes: an unsupervised machine learning analysis

Abstract: To study whether ICU staffing features are associated with improved hospital mortality, ICU length of stay (LOS) and duration of mechanical ventilation (MV) using cluster analysis directed by machine learning. Methods:The following variables were included in the analysis: average bed to nurse, physiotherapist and physician ratios, presence of 24/7 board-certified intensivists and dedicated pharmacists in the ICU, and nurse and physiotherapist autonomy scores. Clusters were defined using the partition around me… Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
2
1
1

Citation Types

1
66
0
4

Year Published

2020
2020
2024
2024

Publication Types

Select...
8

Relationship

2
6

Authors

Journals

citations
Cited by 71 publications
(71 citation statements)
references
References 28 publications
1
66
0
4
Order By: Relevance
“…In another study that analyzed 167 confirmed patients with severe COVID-19, the LDH concentration was higher and the albumin concentration was lower in these patients, with significant differences [ 34 ]. Recently, Liang et al [ 22 ] used Least Absolute Shrinkage and Selection Operator (LASSO) and logistic regression to construct a predictive risk score (COVID-GRAM), in which the AUC in the development cohort was 0.88 (95% CI 0.85-0.91) and the AUC in the validation cohort was 0.88 (95% CI 0.84-0.93) for predicting patients’ risk of developing critical illness. The results in that study coincide with our results to some extent, and the machine learning algorithm can identify potential indicators better than the conventional algorithm.…”
Section: Discussionmentioning
confidence: 99%
See 1 more Smart Citation
“…In another study that analyzed 167 confirmed patients with severe COVID-19, the LDH concentration was higher and the albumin concentration was lower in these patients, with significant differences [ 34 ]. Recently, Liang et al [ 22 ] used Least Absolute Shrinkage and Selection Operator (LASSO) and logistic regression to construct a predictive risk score (COVID-GRAM), in which the AUC in the development cohort was 0.88 (95% CI 0.85-0.91) and the AUC in the validation cohort was 0.88 (95% CI 0.84-0.93) for predicting patients’ risk of developing critical illness. The results in that study coincide with our results to some extent, and the machine learning algorithm can identify potential indicators better than the conventional algorithm.…”
Section: Discussionmentioning
confidence: 99%
“…The theoretical core of machine learning analysis is the data mining algorithm. Various data mining algorithms based on different data types and formats can more scientifically represent the characteristics of the data and can better penetrate the data trends and recognized values [ 22 ]. On this basis, one of the most important application areas is predictive analysis, which involves identifying features (in machine learning, “features” refers to individual characteristics of the data) from mechanical learning, establishing models through science, and then running new data through the models to predict future data [ 23 ].…”
Section: Introductionmentioning
confidence: 99%
“…In Brazil, ICU nurse staffing varies considerably depending on local regulations. However, a study in 93 Brazilian ICUs showed that ICUs where nurses have higher autonomy, including start of weaning from ventilation and the titrating FiO 2 , have better outcomes compared to ICUs where there is less nurse autonomy [36]. After implementation of the structured checklist in the second phase of the CHECKLIST-ICU study, the positive effect of one nurse for every 10 patients in all shifts on the use of LTVV was no longer associated with compliance with LTVV.…”
Section: Discussionmentioning
confidence: 99%
“…This was a retrospective study on prospectively collected data from two databases. The first is from the Orchestra study [15], a multicenter study performed in 93 ICUs from 55 hospitals in several Brazilian states from January 2014 to December 2015, and the second, from the A.C. Camargo Cancer Center, a dedicated cancer center in São Paulo Brazil, with 50 ICU beds, from January 2011 to December 2017. A.C. Camargo Cancer Center Local Ethics Committees (CAAE: 86761718.0.0000.5432) and the Brazilian National Ethics Committee (CAAE: 19687113.8.1001.5249) approved the study without the need for informed consent, since all data were fully anonymized before researches could access them.…”
Section: Design and Settingmentioning
confidence: 99%