Machine learning classifiers do not improve prediction of hospitalization > 2 days after fast-track hip and knee arthroplasty compared with a classical statistical risk model

Publikation: Bidrag til tidsskriftTidsskriftartikelForskningfagfællebedømt

Standard

Machine learning classifiers do not improve prediction of hospitalization > 2 days after fast-track hip and knee arthroplasty compared with a classical statistical risk model. / Johannesdottir, Katrin B.; Kehlet, Henrik; Petersen, Pelle B.; Aasvang, Eske K.; Sørensen, Helge B. D.; Jørgensen, Christoffer C.; Centre for Fast-track Hip and Knee Replacement Collaborative Group.

I: Acta Orthopaedica, Bind 93, 2022, s. 117-123.

Publikation: Bidrag til tidsskriftTidsskriftartikelForskningfagfællebedømt

Harvard

Johannesdottir, KB, Kehlet, H, Petersen, PB, Aasvang, EK, Sørensen, HBD, Jørgensen, CC & Centre for Fast-track Hip and Knee Replacement Collaborative Group 2022, 'Machine learning classifiers do not improve prediction of hospitalization > 2 days after fast-track hip and knee arthroplasty compared with a classical statistical risk model', Acta Orthopaedica, bind 93, s. 117-123. https://doi.org/10.2340/17453674.2021.843

APA

Johannesdottir, K. B., Kehlet, H., Petersen, P. B., Aasvang, E. K., Sørensen, H. B. D., Jørgensen, C. C., & Centre for Fast-track Hip and Knee Replacement Collaborative Group (2022). Machine learning classifiers do not improve prediction of hospitalization > 2 days after fast-track hip and knee arthroplasty compared with a classical statistical risk model. Acta Orthopaedica, 93, 117-123. https://doi.org/10.2340/17453674.2021.843

Vancouver

Johannesdottir KB, Kehlet H, Petersen PB, Aasvang EK, Sørensen HBD, Jørgensen CC o.a. Machine learning classifiers do not improve prediction of hospitalization > 2 days after fast-track hip and knee arthroplasty compared with a classical statistical risk model. Acta Orthopaedica. 2022;93:117-123. https://doi.org/10.2340/17453674.2021.843

Author

Johannesdottir, Katrin B. ; Kehlet, Henrik ; Petersen, Pelle B. ; Aasvang, Eske K. ; Sørensen, Helge B. D. ; Jørgensen, Christoffer C. ; Centre for Fast-track Hip and Knee Replacement Collaborative Group. / Machine learning classifiers do not improve prediction of hospitalization > 2 days after fast-track hip and knee arthroplasty compared with a classical statistical risk model. I: Acta Orthopaedica. 2022 ; Bind 93. s. 117-123.

Bibtex

@article{1c069ba740bb4f3084b3958ef93bf050,
title = "Machine learning classifiers do not improve prediction of hospitalization > 2 days after fast-track hip and knee arthroplasty compared with a classical statistical risk model",
abstract = "Background and purpose: Prediction of postoperative outcomes and length of hospital stay (LOS) of patients is vital for allocation of healthcare resources. We investigated the performance of prediction models based on machinelearning algorithms compared with a previous risk stratification model using traditional multiple logistic regression, for predicting the risk of a LOS of > 2 days after fast-track total hip and knee replacement. Patients and methods: 3 different machine learning classifiers were trained on data from the Lundbeck Centre for Fast-track Hip and Knee Replacement Database (LCDB) collected from 9,512 patients between 2016 and 2017. The chosen classifiers were a random forest classifier (RF), a support vector machine classifier with a polynomial kernel (SVM), and a multinomial Na{\"i}ve-Bayes classifier (NB). Results: Comparing performance measures of the classifiers with the traditional model revealed that all the models had a similar performance in terms of F1 score, accuracy, sensitivity, specificity, area under the receiver operating curve (AUC), and area under the precision-recall curve (AUPRC). A feature importance analysis of the RF classifier found hospital, age, use of walking aid, living alone, and joint operated on to be the most relevant input features. None of the classifiers reached a clinically relevant performance with the input data from the LCDB. Interpretation: Despite the promising prospects of machine-learning practices for disease and risk prediction, none of the machine learning models tested outperformed the traditional multiple regression model in predicting which patients in this cohort had a LOS > 2 days.",
author = "Johannesdottir, {Katrin B.} and Henrik Kehlet and Petersen, {Pelle B.} and Aasvang, {Eske K.} and S{\o}rensen, {Helge B. D.} and J{\o}rgensen, {Christoffer C.} and {Centre for Fast-track Hip and Knee Replacement Collaborative Group}",
year = "2022",
doi = "10.2340/17453674.2021.843",
language = "English",
volume = "93",
pages = "117--123",
journal = "Acta Orthopaedica",
issn = "1745-3674",
publisher = "Taylor & Francis",

}

RIS

TY - JOUR

T1 - Machine learning classifiers do not improve prediction of hospitalization > 2 days after fast-track hip and knee arthroplasty compared with a classical statistical risk model

AU - Johannesdottir, Katrin B.

AU - Kehlet, Henrik

AU - Petersen, Pelle B.

AU - Aasvang, Eske K.

AU - Sørensen, Helge B. D.

AU - Jørgensen, Christoffer C.

AU - Centre for Fast-track Hip and Knee Replacement Collaborative Group

PY - 2022

Y1 - 2022

N2 - Background and purpose: Prediction of postoperative outcomes and length of hospital stay (LOS) of patients is vital for allocation of healthcare resources. We investigated the performance of prediction models based on machinelearning algorithms compared with a previous risk stratification model using traditional multiple logistic regression, for predicting the risk of a LOS of > 2 days after fast-track total hip and knee replacement. Patients and methods: 3 different machine learning classifiers were trained on data from the Lundbeck Centre for Fast-track Hip and Knee Replacement Database (LCDB) collected from 9,512 patients between 2016 and 2017. The chosen classifiers were a random forest classifier (RF), a support vector machine classifier with a polynomial kernel (SVM), and a multinomial Naïve-Bayes classifier (NB). Results: Comparing performance measures of the classifiers with the traditional model revealed that all the models had a similar performance in terms of F1 score, accuracy, sensitivity, specificity, area under the receiver operating curve (AUC), and area under the precision-recall curve (AUPRC). A feature importance analysis of the RF classifier found hospital, age, use of walking aid, living alone, and joint operated on to be the most relevant input features. None of the classifiers reached a clinically relevant performance with the input data from the LCDB. Interpretation: Despite the promising prospects of machine-learning practices for disease and risk prediction, none of the machine learning models tested outperformed the traditional multiple regression model in predicting which patients in this cohort had a LOS > 2 days.

AB - Background and purpose: Prediction of postoperative outcomes and length of hospital stay (LOS) of patients is vital for allocation of healthcare resources. We investigated the performance of prediction models based on machinelearning algorithms compared with a previous risk stratification model using traditional multiple logistic regression, for predicting the risk of a LOS of > 2 days after fast-track total hip and knee replacement. Patients and methods: 3 different machine learning classifiers were trained on data from the Lundbeck Centre for Fast-track Hip and Knee Replacement Database (LCDB) collected from 9,512 patients between 2016 and 2017. The chosen classifiers were a random forest classifier (RF), a support vector machine classifier with a polynomial kernel (SVM), and a multinomial Naïve-Bayes classifier (NB). Results: Comparing performance measures of the classifiers with the traditional model revealed that all the models had a similar performance in terms of F1 score, accuracy, sensitivity, specificity, area under the receiver operating curve (AUC), and area under the precision-recall curve (AUPRC). A feature importance analysis of the RF classifier found hospital, age, use of walking aid, living alone, and joint operated on to be the most relevant input features. None of the classifiers reached a clinically relevant performance with the input data from the LCDB. Interpretation: Despite the promising prospects of machine-learning practices for disease and risk prediction, none of the machine learning models tested outperformed the traditional multiple regression model in predicting which patients in this cohort had a LOS > 2 days.

U2 - 10.2340/17453674.2021.843

DO - 10.2340/17453674.2021.843

M3 - Journal article

C2 - 34984485

AN - SCOPUS:85123228753

VL - 93

SP - 117

EP - 123

JO - Acta Orthopaedica

JF - Acta Orthopaedica

SN - 1745-3674

ER -

ID: 313045791