Use of a multilayer perceptron to create a prediction model for dressing independence in a small sample at a single facility

Abstract

[Purpose] This study aimed to assess the accuracy of a prediction model for dressing independence created with a multilayer perceptron in a small sample at a single facility. [Participants and Methods] This retrospective observational study included 82 first-stroke patients. The prediction models for dressing independence at hospital discharge were created using a multilayer perceptron, logistic regression, and a decision tree, and compared for predictive accuracy. Age, dressing performance, trunk function, visuospatial perception, balance, and cognitive function at admission were used as variables. [Results] The area under the receiver operating characteristic curve, classification accuracy, sensitivity, specificity, positive-predictive value, and negative-predictive value for training data were highest with the multilayer perceptron model. Cochran’s Q and multiple comparison tests revealed a significant difference between logistic regression and multilayer perceptron models. Testing of data in 10-fold cross-validation yielded the same results, except for sensitivity. [Conclusion] The present study suggested that higher accuracy could be expected with a multilayer perceptron than with logistic regression and a decision tree when creating a prediction model for independence of activities of daily living in a small sample of stroke patients.

Relationship between the number of samples and the accuracy of the prediction model for dressing independence using artificial neural networks in stroke patients

“…Our results suggested that the lower limit on the number of samples to create a useful prediction model for dressing independence when using artificial neural networks is approximately 80 and that the advantage may be lost if the number of samples is <60. In a sample of approximately 80 patients, the artificial neural network successfully created a model with higher accuracy than logistic regression; this is consistent with the result of a previous study [5]. In the dataset of 100 patients, the difference in the classification accuracy between the artificial neural network and logistic regression was not significant compared with that in the datasets of 120 and 80 patients; however, the artificial neural network exceeded the logistic regression in all positive predictive values, negative predictive values, sensitivities, and specificities.…”

“…In recent years, a systematic review and meta-analysis on the outcomes of trauma patients [32] also reported that artificial neural network models have better performance than logistic regression. Furthermore, the authors reported that even in a small sample of 83 patients, artificial neural networks successfully created models with higher prediction accuracy than logistic regression in terms of predicting the dressing independence of stroke patients [5]. By contrast, no studies have verified the difference in accuracy between artificial neural networks and logistic regression by changing the number of samples, and only a few reports used a small number of samples (i.e., 100).…”

“…However, the number of usable data is small when creating a prediction model at each single facility. Fujita et al [5] examined a method that creates a highly accurate ADL prediction model even for small samples at each single facility and reported that artificial neural networks are useful. They also reported that when a prediction model for dressing independence was created using 83 samples, the accuracy of the artificial neural network model was higher than that of logistic regression and decision trees.…”

Predicting rectal cancer prognosis from histopathological images and clinical information using multi-modal deep learning

“…ResNeSt is an enhanced version of ResNet that incorporates a Split-Attention module into the architecture design, resulting in increased performance of the network ( 14 , 15 ). The Multi-Layer Perceptron (MLP) is a type of feedforward ANNs (Artificial Neural Network) that, which has a long history of implementation in medical research for image classification ( 29 , 30 ), detection ( 31 , 32 ) and prediction ( 33 , 34 ). The multi-head attention mechanism aims to prioritize the most relevant information for the current task from a large amount of input data.…”

Evaluation of Perioperative Nursing for Patients with Hepatobiliary Pancreatic Diseases Combined with Diabetes Mellitus Based on Multilayer Perceptron Neural Network

“…Network. MLP is one of the most widely used artificial neural network models, which has good nonlinear system modeling ability [11]. The MLP neural network is composed of multilayer nonlinear neurons (also known as nodes), which is divided into three parts: the input layer composed of a group of source nodes, the implicit layer of one or more layers of computing nodes, and the output layer of one layer of computing nodes.…”