Development and validation of two aspiration prediction models in patients receiving nasogastric feeding

Abstract

Aim To develop and validate two aspiration prediction models in patients receiving nasogastric feeding. Background Aspiration is one of the most serious complications of nasogastric feeding. However, there is a lack of aspiration prediction models for nasogastric feeding. Methods A total of 515 patients receiving nasogastric feeding were randomly selected for this unmatched case–control study, with 103 patients in the case group and 412 patients in the control group. Logistic regression was used to develop nomogram and Classification And Regression Tree (CART) models. The performances of the models were internally validated using 1,000 bootstrapped samples. Results The predictive accuracy of the CART model (94.5%) was higher than that of the nomogram model (89.1%). The area under the receiver operating characteristic curve of the CART model (0.96) was slightly higher than that of the nomogram model (0.93). Conclusions The intubation depth, number of comorbidities, aspiration history, indwelling days, food type and the use of sedative‐hypnotics may be used to identify aspiration risk. Implications for Nursing Management Two aspiration prediction models are provided for nurses to evaluate aspiration risk and increase the quality of nursing management.

A machine learning model for the prediction of peripherally inserted central catheter-related venous thrombosis among high-risk adult patients

“…This study does not present a variety of complex mathematical equations but provides a visual way to explain the model. Model visualization can support clinicians and nurses in making decisions and making treatment recommendations for patients [30]. The signi cant in uence of each feature on the relevant prediction for each particular patient can be seen from the SHAP value, and the risk of thromboembolism for each patient is evident when compared to the baseline value.…”

Sample size requirements are not being considered in studies developing prediction models for binary outcomes: a systematic review

“…The ratio of cases to controls was 1:4. Considering that a larger sample size would yield better performances of the prediction models, 515 patients were randomly selected for inclusion in the study, with 103 patients in the aspiration group and 412 patients in the non-aspiration group’) [ 29 ]. One study justified their sample size by reporting ‘BIMS occurred in 2782 patients in our dataset.…”