Machine-learning vs. logistic regression for preoperative prediction of medical morbidity after fast-track hip and knee arthroplasty—a comparative study

BMC Anesthesiology · 16 authors, 4 centres

AI SUMMARY

FIDELITY 100%

This summary was generated by AI from a single paper. It has not been reviewed by a clinician and is not clinical advice. Verify against the source before acting on it.

AGENT 01 · STRUCTURED SUMMARY

TRACE TO SOURCE

“

TL;DR

A cohort study compared machine-learning and logistic regression models for predicting postoperative medical complications in fast-track hip and knee replacement patients. The machine-learning model showed a slight improvement in identifying high-risk patients.

Full summary

806 CHARS

A machine-learning model (Boosted Decision Trees) using 33 preoperative variables was compared to a logistic regression model for predicting 'medical' morbidity defined as length of stay >4 days or 90-day readmission. When selecting the top 20% as 'risk-patients' from the test cohort, the full machine-learning model had a precision of 13.6% and AUROC of 76.3%, compared to 12.4% and 74.7% for the logistic regression model. The machine-learning model was slightly superior on most performance metrics. SHAP analysis revealed age and use of walking aids were the most important predictors in both models. The study concluded that a machine-learning algorithm slightly improves preoperative risk stratification for medical complications, which could help target patients for intensified perioperative care.

STUDY TYPE

cohort

SPECIALTY

surgery

SUMMARISED BY

AI pipeline

FIDELITY CHECK

100% · A

ORIGINAL PAPER

PMC10685559

Machine-learning vs. logistic regression for preoperative prediction of medical morbidity after fast-track hip and knee arthroplasty—a comparative study

BMC Anesthesiology

READ THE FULL PAPER

ABSTRACT

Background Machine-learning models may improve prediction of length of stay (LOS) and morbidity after surgery. However, few studies include fast-track programs, and most rely on administrative coding with limited follow-up and information on perioperative care. This study investigates potential benefits of a machine-learning model for prediction of postoperative morbidity in fast-track total hip (THA) and knee arthroplasty (TKA). Methods Cohort study in consecutive unselected primary THA/TKA between 2014–2017 from seven Danish centers with established fast-track protocols. Preoperative comorbidity and prescribed medication were recorded prospectively and information on length of stay and readmissions was obtained through the Danish National Patient Registry and medical records. We used a machine-learning model (Boosted Decision Trees) based on boosted decision trees with 33 preoperative variables for predicting “medical” morbidity leading to LOS > 4 days or 90-days readmissions and compared to a logistical regression model based on the same variables. We also evaluated two parsimonious models, using the ten most important variables in the full machine-learning and logistic regression models. Data collected between 2014–2016 (n:18,013) was used for model training and data from 2017 (n:3913) was used for testing. Model performances were analyzed using precision, area under receiver operating (AUROC) and precision recall curves (AUPRC), as well as the Mathews Correlation Coefficient. Variable importance was analyzed using Shapley Additive Explanations values. Results Using a threshold of 20% “risk-patients” (n:782), precision, AUROC and AUPRC were 13.6%, 76.3% and 15.5% vs. 12.4%, 74.7% and 15.6% for the machine-learning and logistic regression model, respectively. The parsimonious machine-learning model performed better than the full logistic regression model. Of the top ten variables, eight were shared between the machine-learning and logistic regression models, but with a considerable age-related variation in importance of specific types of medication. Conclusion A machine-learning model using preoperative characteristics and prescriptions slightly improved identification of patients in high-risk of “medical” complications after fast-track THA and TKA compared to a logistic regression model. Such algorithms could help find a manageable population of patients who may benefit most from intensified perioperative care. Supplementary Information The online version contains supplementary material available at 10.1186/s12871-023-02354-z.

CITATION EXPORT

Christian Michelsen, Christoffer C. Jørgensen, Mathias Heltberg et al.. Machine-learning vs. logistic regression for preoperative prediction of medical morbidity after fast-track hip and knee arthroplasty—a comparative study. BMC Anesthesiology. (2023). https://doi.org/10.1186/s12871-023-02354-z

AUTHORS

16 · 4 CENTRES

Christian Michelsen, Christoffer C. Jørgensen, Mathias Heltberg, Mogens H. Jensen, Alessandra Lucchetti, Pelle B. Petersen, Troels Petersen, Henrik Kehlet, Frank Madsen, Torben B. Hansen, Kirill Gromov, Thomas Jakobsen, Claus Varnum, Soren Overgaard, Mikkel Rathsach, Lars Hansen

1https://ror.org/035b05819grid.5254.60000 0001 0674 042XThe Niels Bohr Institute, University of Copenhagen, Blegdamsvej 17, 2100 Copenhagen, Denmark
2Department of Anesthesia and Intensive Care, Hospital of Northern Zealand, Dyrehavevej 29 3400, Hillerød, Denmark
3The Centre for Fast-Track Hip and Knee Replacement, 7621, Rigshospitalet, Blegdamsvej 9, 2100 Copenhagen, Denmark
4Section of Surgical Pathophysiology, 7621, Rigshospitalet, Blegdamsvej 9, 2100 Copenhagen, Denmark