Classifications of Neurodegenerative Disorders Using a Multiplex Blood Biomarkers-Based Machine Learning Model

Abstract

Easily accessible biomarkers for Alzheimer's disease (AD), Parkinson's disease (PD), frontotemporal dementia (FTD), and related neurodegenerative disorders are urgently needed in an aging society to assist early-stage diagnoses. In this study, we aimed to develop machine learning algorithms using the multiplex blood-based biomarkers to identify patients with different neurodegenerative diseases. Plasma samples (n = 377) were obtained from healthy controls, patients with AD spectrum (including mild cognitive impairment (MCI)), PD spectrum with variable cognitive severity (including PD with dementia (PDD)), and FTD. We measured plasma levels of amyloid-beta 42 (Aβ42), Aβ40, total Tau, p-Tau181, and α-synuclein using an immunomagnetic reduction-based immunoassay. We observed increased levels of all biomarkers except Aβ40 in the AD group when compared to the MCI and controls. The plasma α-synuclein levels increased in PDD when compared to PD with normal cognition. We applied machine learning-based frameworks, including a linear discriminant analysis (LDA), for feature extraction and several classifiers, using features from these blood-based biomarkers to classify these neurodegenerative disorders. We found that the random forest (RF) was the best classifier to separate different dementia syndromes. Using RF, the established LDA model had an average accuracy of 76% when classifying AD, PD spectrum, and FTD. Moreover, we found 83% and 63% accuracies when differentiating the individual disease severity of subgroups in the AD and PD spectrum, respectively. The developed LDA model with the RF classifier can assist clinicians in distinguishing variable neurodegenerative disorders.

Keywords: Alzheimer’s disease; Parkinson’s disease; biomarkers; classification; deep learning model; frontotemporal dementia; linear discriminant analysis; multivariate imputation by chained equations; neurodegenerative disorders.

Figure 1

Individual plasma biomarker levels of normal controls and in different disease groups. The plasma Aβ42 (a) and Aβ40 (b) levels significantly increased in patients with AD and FTD, especially when compared to the normal control and other disease groups (p < 0.01). The plasma total tau (c) and p-tau181 (d) level significantly increased in patients with FTD and then followed by AD, MCI, and PDD group (p < 0.01). The plasma α-synuclein (e) levels were highest in the PDD group than other disease groups and controls. The mean ± one standard deviation (SD) was illustrated as horizontal lines in each disease group. * p < 0.05; ** p < 0.01.

Figure 2

The correlation matrix between each biomarker from four groups, including healthy controls, as well as patients with Alzheimer’s disease (AD) spectrum, Parkinson’s Disease (PD) spectrum, and Frontotemporal Dementia (FTD). The upper triangular part of the matrix is the correlation coefficients between any two biomarkers. The lower triangular part of the matrix is the scattered-plot graphs of any two biomarkers. The main diagonal part of the matrix is the distribution graphs of each biomarker. log(α) is the log of α-synuclein. * p < 0.05; *** p < 0.001.

Figure 3

The 3D scatter plot demonstrates that LDA separated sample points of different dementia groups. AD: AD spectrm, PD: PD spectrum.

Figure 4

The correlation matrix between each biomarker from three groups, including normal controls, patients with mild cognitive impairment (MCI), and AD. The upper triangular part of the matrix is the correlation coefficients between any two biomarkers. The lower triangular part of the matrix is the scattered-plot graphs of any two biomarkers. The main diagonal part of the matrix is the distribution graphs of each biomarker. log(α) is the log of α-synuclein. * p < 0.05; *** p < 0.001.

Figure 5

The 2D scatter plot demonstrates that LDA separated sample points of the AD spectrum.

Figure 6

The correlation matrix between each biomarker from four groups, including normal controls, PD-NC, PD-MCI, and PDD. The lower triangular part of the matrix is the scattered-plot graphs of any two biomarkers. The main diagonal part of the matrix is the distribution graphs of each biomarker. log(α) is the log of α-synuclein. ** p < 0.01; *** p < 0.001.

Figure 7

The 3D scatter plot demonstrates that LDA separated sample points of different subgroups of the PD spectrum.

Figure 8

The estimated accuracy for models classifying different dementia syndrome (a), AD spectrum (b), and PD spectrum (c). LDA1: the first function of the linear combinations based on the original five biomarker features using linear discriminant analysis; LDA2: the second function of the linear combinations based on the original five biomarker features using linear discriminant analysis; LDA3: the third function of the linear combinations based on the original five biomarker features using linear discriminant analysis.

Figure 9

Data processing flow chart in this study; MICE: multivariate imputation by the chained equation; LOOCV: leave-one-out cross-validation; NB: naïve Bayes; kNN: k-nearest neighbor; SVM: support vector machine; C4.5: C4.5 decision tree; CART: classification and regression trees; RF: random forest; LogReg: logistic regression.