Is the Random Forest Algorithm Suitable for Predicting Parkinson’s Disease with Mild Cognitive Impairment out of Parkinson’s Disease with Normal Cognition?

Abstract

Because it is possible to delay the progression of dementia if it is detected and treated in an early stage, identifying mild cognitive impairment (MCI) is an important primary goal of dementia treatment. The objectives of this study were to develop a random forest-based Parkinson’s disease with mild cognitive impairment (PD-MCI) prediction model considering health behaviors, environmental factors, medical history, physical functions, depression, and cognitive functions using the Parkinson’s Dementia Clinical Epidemiology Data (a national survey conducted by the Korea Centers for Disease Control and Prevention) and to compare the prediction accuracy of our model with those of decision tree and multiple logistic regression models. We analyzed 96 subjects (PD-MCI = 45; Parkinson’s disease with normal cognition (PD-NC) = 51 subjects). The prediction accuracy of the model was calculated using the overall accuracy, sensitivity, and specificity. Based on the random forest analysis, the major risk factors of PD-MCI were, in descending order of magnitude, Clinical Dementia Rating (CDR) sum of boxes, Untitled Parkinson’s Disease Rating (UPDRS) motor score, the Korean Mini Mental State Examination (K-MMSE) total score, and the K- Korean Montreal Cognitive Assessment (K-MoCA) total score. The random forest method achieved a higher sensitivity than the decision tree model. Thus, it is advisable to develop a protocol to easily identify early stage PDD based on the PD-MCI prediction model developed in this study, in order to establish individualized monitoring to track high-risk groups.

Exploring the Predictors of Rapid Eye Movement Sleep Behavior Disorder for Parkinson’s Disease Patients Using Classifier Ensemble

“…Another finding of this study was that the prediction performance of RF was superior to that of CART or RBF artificial neural network as well as that of traditional statistical techniques such as discriminant analysis and regression analysis. The results of this study agreed with a previous study [ 36 ] that predicted the high-risk group of cognitive impairment in old age using RF. It is believed that the prediction performance of RF was better than that of traditional statistical techniques (e.g., regression analysis) or that of CART because RF was based on bootstrap aggregating (bagging), generating diverse trees from multiple bootstrap samples.…”

“…In other words, while CART has a risk of overfitting because even an outlier is highly likely to be constructed as a node without exception, RF can prevent overfitting because it generates multiple bootstrap samples and has better prediction accuracy than decision tree [ 37 ]. RF showed high prediction performance even when the binomial classification was conducted using imbalanced data like disease data [ 36 ].…”

Application of Machine Learning Technique to Distinguish Parkinson’s Disease Dementia and Alzheimer’s Dementia: Predictive Power of Parkinson’s Disease-Related Non-Motor Symptoms and Neuropsychological Profile

“…Health surveys were performed using computer-assisted personal interviews. The data were composed of sociodemographic factors, health behaviors, disease history, motor characteristics related to PD, REM sleep behavior disorders, and neuropsychological test results [ 32 ]. Patients with idiopathic PD were diagnosed by the neurologists according to the diagnostic criteria of the United Kingdom Parkinson’s disease Society Brain Bank [ 33 ].…”

“…However, pruning is minimized while constructing models. Random forest can be free from overfitting theoretically and is not affected by noise or outliers much [ 32 ]. Moreover, it can generate high accuracy results by reducing generalization errors.…”

“…Moreover, it can generate high accuracy results by reducing generalization errors. However, random forest has a higher probability that each tree will be more complex when an unimportant explanatory variable is selected [ 32 ]. Therefore, this study improved the accuracy of the model by considering the number of mtry, the number of candidate explanatory variables, in advance.…”