Predictive models for hospital readmission risk: A systematic review of methods

Modelling hospital readmissions under frailty conditions for healthy aging

“…We have carried out computational validation experiments using several machine learning (Witten et al, ; Haykin, ; Kuhn & Johnson, ) approaches for predictive model building. These models have been reported in the literature about readmission prediction for adult patients (Artetxe et al, ; Kansagara et al, ). We have used the caret package in R (Kuhn & Johnson, ), because to provides a widely recognized standard implementation for the cross‐validation of diverse classification and regression techniques.…”

“…The reported applications of deep learning to readmission prediction are restricted to a specific disease, that is, lupus patients (Reddy & Delen, ), for which there are long clinical histories per patient accessible through the EHR, so that the abundance of data allows for the training of deep models. Therefore, we focus on the following well‐known machine‐learning classification methods (Artetxe, Beristain, Graña and Besga ; Artetxe, Ayerdi, Graña & Rios ; Artetxe, Ayerdi, Graña, & Beristain ; Artetxe et al, ; Garmendia et al, ; Garmendia et al, ): •Linear discrimination analysis (LDA) and variants quadratic discriminant analysis (QDA), and mixture discriminant analysis (MDA) are the most standard linear models that provide a baseline result from linear discriminant theory, which is well grounded and accepted by the medical researchers. •Support vector machines (SVM) are the most standard machine‐learning algorithm in the biosciences literature used for predictive analysis, we explore both linear and non‐linear approaches, the later using the so‐called kernel trick. They have been tested extensively on readmission studies. •Multilayer perceptrons (MLP) are the classical artificial neural network approach to build nonlinear discriminant models.…”

“…An internationally extended standard period to count a patient return as readmission is 30 days, but it may change for political reasons (Kansagara et al, ). Readmission prediction in the case of adult patients has been tackled with diverse statistical approaches (Artetxe et al, ; Kansagara et al, ) such as logistic regression (Garmendia et al, ; Ottenbacher et al, ) and survival analysis (Garmendia et al, ). Recent works favor the application of predictive machine‐learning approaches, formulating readmission prediction as a binary classification problem (Artetxe et al, ; Ottenbacher et al, ).…”

“…Most of the studies about readmissions are focused in high‐risk populations of middle age adults (Artetxe et al, ; Kansagara et al, ), with specific chronic conditions so that readmission is potentially preventable or predictable, the latter implying that it is possible to manage resource planning in order to minimize the impact of unexpected readmissions. There are few studies that consider explicitly the frail aged individuals in their cohorts (Besga et al, ).…”

“…An internationally extended standard period to count a patient return as readmission is 30 days, but it may change for political reasons (Kansagara et al, 2011). Readmission prediction in the case of adult patients has been tackled with diverse statistical approaches (Artetxe et al, 2018;Kansagara et al, 2011) such as logistic regression (Garmendia et al, 2017;Ottenbacher et al, 2001) and survival analysis (Garmendia et al, 2019). machines (SVM) (Besga et al, 2015;Futoma et al, 2015;Zheng et al, 2015;Cui et al, 2018) deep learning (Reddy & Delen, 2018;Xiao et al, 2018), artificial neural network (Ottenbacher et al, 2001), and naïve Bayes (Vukicevic et al, 2015;Shameer et al, 2017).…”