Next Article in Journal
Combined Depth and Subdural Electrodes for Lateralization of the Ictal Onset Zone in Mesial Temporal Lobe Epilepsy with Hippocampal Sclerosis
Previous Article in Journal
Walk Locomotion Kinematic Changes in a Model of Penetrating Hippocampal Injury in Male/Female Mice and Rats
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Brain Sciences
Volume 13
Issue 11
10.3390/brainsci13111546
brainsci-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Applications of Machine Learning to Diagnosis of Parkinson’s Disease

by
Hong Lai
1,2,
Xu-Ying Li
1,
Fanxi Xu
1,
Junge Zhu
1,
Xian Li
1,
Yang Song
1,
Xianlin Wang
1,
Zhanjun Wang
1 and
Chaodong Wang
1,*
1
Department of Neurology, Xuanwu Hospital of Capital Medical University, National Clinical Research Center for Geriatric Diseases, Beijing 100053, China
2
Department of Neurology, The First Affiliated Hospital of Gannan Medical University, Ganzhou 341000, China
*
Author to whom correspondence should be addressed.
Brain Sci. 2023, 13(11), 1546; https://doi.org/10.3390/brainsci13111546
Submission received: 27 September 2023 / Revised: 28 October 2023 / Accepted: 31 October 2023 / Published: 3 November 2023
(This article belongs to the Section Neurodegenerative Diseases)
Browse Figures
Versions Notes

Abstract

:
Background: Accurate diagnosis of Parkinson’s disease (PD) is challenging due to its diverse manifestations. Machine learning (ML) algorithms can improve diagnostic precision, but their generalizability across medical centers in China is underexplored. Objective: To assess the accuracy of an ML algorithm for PD diagnosis, trained and tested on data from different medical centers in China. Methods: A total of 1656 participants were included, with 1028 from Beijing (training set) and 628 from Fuzhou (external validation set). Models were trained using the least absolute shrinkage and selection operator–logistic regression (LASSO-LR), decision tree (DT), random forest (RF), eXtreme gradient boosting (XGboost), support vector machine (SVM), and k-nearest neighbor (KNN) techniques. Hyperparameters were optimized using five-fold cross-validation and grid search techniques. Model performance was evaluated using the area under the curve (AUC) of the receiver operating characteristic (ROC) curve, accuracy, sensitivity (recall), specificity, precision, and F1 score. Variable importance was assessed for all models. Results: SVM demonstrated the best differentiation between healthy controls (HCs) and PD patients (AUC: 0.928, 95% CI: 0.908–0.947; accuracy: 0.844, 95% CI: 0.814–0.871; sensitivity: 0.826, 95% CI: 0.786–0.866; specificity: 0.861, 95% CI: 0.820–0.898; precision: 0.849, 95% CI: 0.807–0.891; F1 score: 0.837, 95% CI: 0.803–0.868) in the validation set. Constipation, olfactory decline, and daytime somnolence significantly influenced predictability. Conclusion: We identified multiple pivotal variables and SVM as a precise and clinician-friendly ML algorithm for prediction of PD in Chinese patients.

1. Introduction

Parkinson’s disease (PD) is the second most prevalent neurodegenerative disorder among the elderly population, primarily characterized by motor symptoms such as bradykinesia, rigidity, resting tremor, and postural instability. Alongside these, a range of non-motor symptoms (NMS) including constipation, hyposmia, REM sleep behavior disorders (RBD), depression, and cognitive impairment further complicate the clinical picture [1]. As the global population ages, PD prevalence continues to increase, affecting approximately 1.7% of individuals aged 65 years and older and 4.0% of those aged 80 years and older [2]. PD significantly impacts patients’ quality of life, social functioning, and family dynamics, imposing a substantial financial burden on individuals and society [3,4,5]. Current diagnostic criteria for PD primarily rely on clinical manifestations, and there is a lack of disease-modifying therapies available. This may be partly due to the loss of more than 50% of dopaminergic cells in the substantia nigra at the time of clinical diagnosis [6]. Early and accurate diagnosis of PD is crucial for implementing effective treatments and improving patient prognosis.
Extensive research has identified potential risk factors and protective factors for PD. Family history [7], pesticide exposure [8], occupational solvent exposure [8], and prodromal symptoms of PD [9] (including hyposmia, constipation, depression, rapid eye movement sleep behavior disorder (RBD), global cognitive deficit, daytime somnolence, and orthostatic hypotension) have been linked to an elevated risk of PD. Conversely, smoking [10], tea consumption [11], coffee intake [12], and physical activity [13] have been associated with a reduced risk of PD. Genome-wide association studies (GWAS) have identified numerous single-nucleotide polymorphisms (SNPs) at loci such as SNCA, GBA, LRRK2, PARK16, BST1, and MAPT which can modulate the risk of PD [14,15]. Leveraging these factors, high-risk populations can be identified, and targeted disease prevention measures can be deployed. However, an efficacious classification model that accurately discerns PD patients from healthy controls (HCs) remains elusive.
The advent of artificial intelligence has propelled machine learning (ML) to the forefront as an indispensable tool for disease diagnosis, progression evaluation, and prognosis assessment [16,17,18]. ML algorithms possess the potential to identify intricate data patterns, automate data analysis, and classify patient-specific data, which can be harnessed for precision medicine applications in PD. In recent years, ML has been applied to PD diagnosis using diverse data modalities, including speech and phonation evaluation [19], handwriting patterns [20], gait analysis [21], neuroimaging [22], cerebrospinal fluid (CSF) [23], and genetic and transcriptomic data [24].
Studies evaluating clinical and imaging biomarkers for the diagnosis of PD have been widely reported. For instance, Kang et al. [25] highlighted the prognostic and diagnostic potential of measures such as CSF Aβ1-42, T-tau, P-tau181, and α-synuclein in early-stage PD. Silveira-Moriyama et al. [26], through their research on smell identification tests in Brazil, underscored the importance of olfactory dysfunction in diagnosing PD. Additionally, Shinde et al. [27] employed neuromelanin-sensitive MRI to identify predictive markers for PD, while Armañanzas et al. [28] leveraged machine learning to pinpoint significant non-motor symptoms associated with PD severity. However, many studies have focused solely on individual biomarkers or imaging indicators, neglecting the potential of integrating diverse data types. Furthermore, the application of advanced machine learning techniques in PD research, despite their proven efficacy in other domains, remains underexplored. These gaps underscore the need for a more comprehensive approach, which our study aims to address by enhancing PD diagnostic accuracy through diverse machine learning algorithms, particularly tailored for the Chinese population.
The crucial questions for ML-based diagnostic models for PD include: what the pivotal variables are for PD, what the best ML algorithm is in the model construction, and how the models perform in different cohorts. To address these questions, we performed a comprehensive analysis using multi-faceted variables including demographic data, environmental factors, lifestyle habits, NMS, and genetic factors, constructed the models using six algorithms, and validated the models using a dataset from a separate (southern) population. Our research pinpointed several key variables and identified SVM as the most accurate and user-friendly ML algorithm for the prediction of PD in the Chinese population. We then present a more precise population. Consequently, we introduce a refined and pragmatical diagnostic model, providing a novel perspective for both the research and treatment of Parkinson’s.

2. Methods

2.1. Study Design and Data Source

The study was designed for training and validation tests. The study was approved by the Medical Ethics Committee of Xuanwu Hospital of Capital Medical University. Informed consent was obtained from all participants in the study. Two populations of participants were included: a training set to train the algorithms and a validation set to independently evaluate each algorithm’s performance. PD patients in the training set were recruited from Xuanwu Hospital of Capital Medical University, while those in the validation set were recruited from Fujian Medical University Union Hospital. Concurrently, healthy controls (HCs) in the training set were selected from among the residents of Beijing, and HCs in the validation set were recruited from Fuzhou communities, matching proportionally (within 5-year strata) to cases by gender and age. Patients were diagnosed by senior neurologists specializing in movement disorders based on the MDS clinical diagnostic criteria for PD released in 2015 [29]. Patients were excluded from the study using the following criteria: (i) Parkinson’s plus syndromes, including multiple system atrophy (MSA), progressive supranuclear palsy (PSP), dementia with Lewy bodies (DLB), and corticobasal degeneration (CBD); (ii) Parkinsonian syndromes caused by cerebrovascular diseases, brain trauma, hypoxic diseases, infectious diseases, and metabolic diseases, affecting the central nervous system; (iii) malignant tumors or other serious systemic diseases; and (iv) dementia, rendering them unable to cooperate with the questionnaire. The control participants who were enrolled fulfilled the following criteria: (i) lack of PD-related motor symptoms (bradykinesia, tremor, postural instability and rigidity); (ii) no history of dementia, PD, or other neurodegenerative diseases; and (iii) no history of malignant tumors or other serious systemic diseases. A total of 1882 participants were recruited from December 2016 to September 2021. After excluding 226 participants due to incomplete gene detection or critical assessments, we had a final count of 1656 participants. The training set comprised 1028 individuals (524 PD patients and 504 HCs), while the external validation set included 628 individuals (305 PD patients and 323 HCs) (Figure 1). All participants were assessed for demographic information, lifestyle behaviors, and environmental exposures. Participants were also assessed for motor and NMS of PD using diagnostic scales (additional information provided in Supplementary Methods). All assessments were performed via face-to-face interviews by clinical investigators with unified training.

2.2. Candidate Variables

The accuracy of any model largely depends on the variables it considers. Here, we outline the variables chosen in our study. Drawing on the previously published literature [2,30,31], we selected 26 potential predictive variables for inclusion in the prediction model: demographic variables (age, sex, and education level), family history of PD, head injury with unconsciousness, general anesthesia, PD-related NMS (possible RBD, olfactory dysfunction, constipation, global cognitive deficit, depression, and daytime somnolence), lifestyle factors (smoking, alcohol consumption, and tea and coffee intake), environmental exposures (pesticides, organic solvents, and heavy metals), as well as 7 PD-associated SNPs in Chinese and other populations (MMRN1 rs6532194, RAB7L1 rs823144, SNCA rs356182 and rs356219, LRRK2 rs34778348, MCCC1 rs12637471, and GBA rs421016) [32,33,34,35,36,37].

2.3. Genotype Analysis and Classification

Genetic factors play a pivotal role in many diseases, including Parkinson’s. This subsection elaborates on the genotype analysis we conducted. Genomic DNA was extracted from peripheral blood leukocytes using a standard protocol. Polymerase chain reaction (PCR) assays and extension primers for variants were designed employing the MassARRAY Assay Design software version 4.0 (Sequenom, Inc., San Diego, CA, USA) and the Primer 6.0 software (Premier, Vancouver, BC, Canada). The primer sequences for amplifying each SNP are listed in Supplementary Table S1. PCR products were purified and sequenced using an ABI3730xl DNA analyzer (Applied Biosystems, Inc., Waltham, MA, USA). Sequence readings were performed with the Chromas 2.22 software.

2.4. Machine Learning Algorithm

Machine learning stands as the cornerstone of our research. In this section, we delve into the algorithms we have harnessed and the metrics we utilize for evaluation. Six supervised ML algorithms were employed: least absolute shrinkage and selection operator–logistic regression (LASSO-LR), decision tree (DT), random forest (RF), eXtreme gradient boosting (XGboost), support vector machine (SVM), and k-nearest neighbor (KNN). In the training set, we applied five-fold cross-validation to minimize each model’s overfitting and utilized the grid search technique to select the optimal combination of hyperparameters for each ML algorithm. Algorithm performances were evaluated based on the area under the curve (AUC) of the receiver operating characteristic (ROC) curve, accuracy, sensitivity (recall), specificity, precision, and F1 score. The model with the highest accuracy was chosen for evaluation using the validation set. In addition to performance comparisons, we also analyzed the importance of variable factors in the models to identify which variables were critical in distinguishing PD patients from HCs in the training set. While deep learning techniques have shown significant promise in various applications, we opted not to use them in this study due to several considerations. Firstly, deep learning models typically require large datasets for effective training and to mitigate overfitting. Secondly, the intricacies of parameter tuning and model selection can be challenging. Furthermore, the complexity of deep learning models might compromise the interpretability of results, which was a key focus for our study. Given our dataset’s size and our emphasis on clear variable interpretation, we chose the machine learning algorithms described above. Detailed information about the ML algorithms used in this study is provided in the Supplementary Materials.

2.5. Statistical Analysis

This subsection dives into the statistical techniques we used to validate and interpret our findings. All statistical analyses were performed using the R software (version 4.1.1; R Foundation for Statistical Computing Vienna, Austria; http://www.R-project.org/, accessed on 9 October 2023). Categorical variables were expressed as frequencies and percentages, while continuous variables were described as medians and interquartile ranges (IQRs). Mann–Whitney U tests (for continuous variables with skewed distributions) and the chi-square test (for categorical variables) were utilized to compare differences in characteristics between groups. The R package “glmnet” was employed to perform the LASSO regression. The R packages “pROC”, “plotROC”, and “rmda” were applied to generate the ROC. The R package “caret” was utilized to run the ML algorithms. All tests were two-tailed, and p < 0.05 was defined as statistically significant.

3. Results

3.1. Clinical and Demographic Characteristics

Here, we present an overview of the clinical and demographic traits of our dataset. In the training set, the median age for PD patients was 66 years (interquartile range, 61–71.25 years), with 49% being male. The HCs had a median age of 67 years (interquartile range, 63–70 years), with 45% male representation. In the external validation set, PD patients had a median age of 69 years (interquartile range, 61–75 years), with 56.7% being male, while HCs had a median age of 68 years (interquartile range, 65–73 years), with 50.2% being male. Detailed demographic and clinical characteristics of the training and validation sets are provided in Table 1.

3.2. Comparison of Algorithms

With multiple algorithms being considered, it is essential to understand their comparative strengths and weaknesses. This section provides a side-by-side analysis of the algorithms, highlighting their performance metrics. Table 2 presents the performances of various ML algorithms with the training and validation sets, while Figure 2A,B illustrates the ROC of all six models with the training set and validation set.
LASSO–logistic regression: Utilizing 10-fold cross-validation, a LASSO regression model was employed to select predictive variables from among the preliminary factors. Following LASSO regression selection (Supplementary Figure S1), 15 variables with non-zero coefficients that minimized the overall lambda were identified as potential optimal variables for distinguishing PD. The optimal value of lambda, the ridge penalty, was 0.021. The tuned model in the training set accurately classified 433 out of 524 PD patients and 452 out of 504 HCs, achieving an AUC of 0.930 (95% CI: 0.915–0.945), an F1 score of 0.858 (95% CI: 0.834–0.879), and an overall accuracy of 0.861 (95% CI: 0.840–0.881). In the validation set, the model accurately classified 247 out of 305 PD patients and 282 out of 323 HCs, yielding an AUC of 0.925 (95% CI: 0.905–0.945), an F1 score of 0.833 (95% CI: 0.799–0.864), and an overall accuracy of 0.842 (95% CI: 0.812–0.869).
Decision tree: Employing five-fold cross-validation and the grid search method (Supplementary Figure S2A), the optimal complexity parameter (CP) for DT was determined to be 0.006. The tuned model in the training set accurately classified 437 out of 524 PD patients and 438 out of 504 HCs, resulting in an AUC of 0.888 (95% CI: 0.868−0.909), an F1 score of 0.851 (95% CI: 0.828–0.875), and an overall accuracy of 0.851 (95% CI: 0.829–0.873) In the validation set, the model accurately classified 229 out of 305 PD patients and 265 out of 323 HCs, yielding an AUC of 0.831 (95% CI: 0.80–0.862), an F1 score of 0.774 (95% CI: 0.731–0.811), and an overall accuracy of 0.787 (95% CI: 0.753–0.820).
Random forest: Utilizing five-fold cross-validation and the grid search method (Supplementary Figure S2B), 500 decision trees and three predictor variables selected at each node (mtry) provided optimal performance. In the training set, the tuned model accurately classified 477 out of 524 PD patients and 459 out of 504 HCs, achieving an AUC of 0.963 (95% CI: 0.952–0.973), an F1 score of 0.912 (95% CI: 0.894–0.930), and an overall accuracy of 0.911(95% CI: 0.892–0.928). In the validation set, the tuned model accurately classified 264 out of 305 PD patients and 260 out of 323 HCs, yielding an AUC of 0.912 (95% CI: 0.890–0.934), an F1 score of 0.835 (95% CI: 0.802–0.867), and an overall accuracy of 0.834 (95% CI: 0.804–0.863).
EXtreme gradient boosting: Employing five-fold cross-validation and the grid search method, the optimal parameters for the XGBoost model were as follows: learning rate (eta) of 0.3, colsample_bytree of 0.6, gamma of 0, maximum tree depth (max_depth) of 1, subsample of 0.875, and minimum leaf node sample weight (min_child_weight) of 1 (Supplementary Figure S2C). In the training set, the tuned model accurately classified 461 out of 524 PD patients and 448 out of 504 HCs, achieving an AUC of 0.946 (95% CI: 0.946–0.959), an F1 score of 0.886 (95% CI: 0.866–0.905), and an overall accuracy of 0.884 (95% CI: 0.865–0.904). In the validation set, the tuned model accurately classified 271 out of 305 PD patients and 243 out of 323 HCs, yielding an AUC of 0.908 (95% CI: 0.886–0.931), an F1 score of 0.826 (95% CI: 0.794–0.858), and an overall accuracy of 0.819 (95% CI: 0.788–0.849).
Support vector machine: A five-fold cross-validation and grid search determined that the optimal SVM cost choice C was 0.25, and the optimal kernel smoothing parameter σ was 0.04 (Supplementary Figure S2D). In the training set, the tuned model accurately classified 439 out of 524 PD patients and 459 out of 504 HCs, achieving an AUC of 0.941 (95% CI: 0.927–0.955), an F1 score of 0.871 (95% CI: 0.849–0.895), and an overall accuracy of 0.874 (95% CI: 0.854–0.896). In the validation set, the tuned model accurately classified 252 out of 305 PD patients and 278 out of 323 HCs, yielding an AUC of 0.928 (95% CI: 0.908–0.947), an F1 score of 0.837 (95% CI: 0.803–0.868), and an overall accuracy of 0.844 (95% CI: 0.814–0.871).
K-nearest neighbor: Based on five-fold cross-validation and grid search (Supplementary Figure S2E), the optimal number of nearest neighbors (k) was found to be nine. In the training set, the tuned model accurately classified 467 out of 524 PD patients and 427 out of 504 HCs, achieving an AUC of 0.941 (95% CI: 0.928–0.954), an F1 score of 0.875 (95% CI: 0.854–0.896), and an overall accuracy of 0.870 (95% CI: 0.849–0.890). In the validation set, the tuned model accurately classified 253 out of 305 PD patients and 256 out of 323 HCs, yielding an AUC of 0.896 (95% CI: 0.871–0.921), an F1 score of 0.810 (95% CI:0.774–0.842), and an overall accuracy of 0.811 (95% CI: 0.779–0.841).
Variable importance: The significance of features, as demonstrated by effect sizes, was calculated (Figure 3A–F). Olfactory dysfunction and constipation exhibited the highest frequencies among the top predictors across all six models, while daytime somnolence, global cognitive deficit, and depression also displayed large effect sizes in more than half of the models.
The SVM model, exhibiting the most remarkable discrimination capabilities, outperformed other models by achieving the highest AUC, overall accuracy, and F1 score in the validation set.

4. Discussion

The primary objective of this study was to develop a non-invasive, cost-effective, and accurate classification model to differentiate patients with PD and HCs by employing genetic factors, lifestyle factors, environmental exposures, and NMS. The current study combined analyses of multiple variable types and ML algorithms, and the model validation used datasets from distinct geographic centers. The results help to establish a comprehensive approach to the early and precise diagnosis of PD for the Chinese population.
Previous studies have typically relied on individual features for PD prediction [25,26,27,28]. However, models incorporating multiple factors have demonstrated increased accuracy compared to those based on single risk factors. In recent years, the application of ML algorithms has gained increasing attention. Table 3 [38,39,40,41,42] provides a comparison of several models. R. Prashanth et al. utilized a combination of NMS features, cerebrospinal fluid (CSF), and imaging markers to differentiate early PD subjects from normal individuals using the naïve Bayes, SVM, boosted trees, and RF methods, observing that the SVM classifier delivered the most optimal performance [38]. Govindu et al. employed machine learning techniques in telemedicine, using classifiers such as SVM, RF, KNN, and logistic regression on collected audio data. Among these, the random forest (RF) classifier stood out, achieving a remarkable 91.83% accuracy in early PD detection [42]. However, many existing models generated overfitting, especially when they were trained on limited datasets. Additionally, the unclear interpretability of some complex models makes it challenging for clinicians to trust and adopt these models in practice. Furthermore, several models were constructed using isolated datasets, and their generalizability across diverse populations and medical centers has yet to be thoroughly evaluated.
Our study addressed some of these gaps by evaluating and comparing multiple ML algorithms for PD detection, emphasizing both performance and interpretability. First, our data originated from different medical centers, which not only adds diversity to the datasets but also enhances the model’s generalizability. Second, in contrast to studies that rely on expensive radiomics or invasive cerebrospinal fluid collection, our approach is non-invasive and cost-effective. This holds tremendous potential for large-scale screening and clinical applications. Moreover, our model took into account a wider range of factors, including genetic factors, lifestyle factors, environmental exposure, and non-motor symptoms, contributing to a more accurate diagnosis of PD. Lastly, our data collection process was straightforward and user-friendly, and thus is easier for clinical doctors. In summary, our study offered a cost-effective, non-invasive, comprehensive, and easily accessible approach to the prediction of PD, and it is suitable for large-scale screening and clinical practice.
Currently, researchers are exploring novel biomarkers for diagnosing and predicting PD [43,44,45,46]. Although numerous studies have been conducted, no widely recognized and easy-to-use predictive tool currently exists. In this study, we employed ML algorithms and discovered that genetic factors, lifestyles, environmental exposures, and NMS serve as robust predictors of PD, largely aligning with previous findings [2,47]. Notably, we identified a strong association between PD and NMS. The variable importance analysis of the six ML models demonstrated that NMS were among the top predictors with the highest frequencies.
At present, the diagnosis of PD predominantly relies on clinical motor symptoms; however, NMS could potentially serve as an early warning for PD progression, as they may precede motor symptoms by decades [48]. This period, known as the prodromal stage, is crucial for identifying individuals with early-stage PD, potentially paving the way for disease-modifying treatments that could delay or even prevent the progression to manifest Parkinson’s disease if administered early.
Among the prodromal PD indicators in our study, olfactory dysfunction emerged as the strongest association, followed by constipation and daytime somnolence. This relationship aligns with previous findings on PD risk. Olfactory dysfunction, one of the most common and typical NMS associated with PD [49], is found in approximately 50–90% of PD patients [50], often manifesting as one of the disease’s earliest symptoms [51]. Most patients develop olfactory dysfunction 4–6 years before the onset of motor impairment [47]. This is supported by Braak et al., who identified the presence of Lewy bodies and Lewy neurites in the olfactory bulb during the premotor stage of PD [52]. Constipation, a marker of early PD, affects approximately half of patients before the emergence of overt motor symptoms [53]. The neuropathology of α-synuclein in the enteric nervous system may precede typical changes in the midbrain and limbic regions [49]. Prior research has shown a higher prevalence of daytime somnolence in PD (35.1%) compared to the general population (9.0–16.8%) [54]. The Honolulu–Asia Aging study revealed that men reporting subjective daytime somnolence had a 2.8-fold increased relative risk of developing PD [55]. Patients newly diagnosed with PD may already exhibit mild cognitive deficits, with nearly 50% developing PD dementia within 10 years of disease onset [56]. Depression, one of the earliest NMS in PD, has an incidence in PD patients three times higher than in the general population [57]. A large-scale retrospective cohort analysis of 32,415 individuals suggested that most patients developed depression approximately 10 years before motor symptoms [58]. RBD is a frequent and significant prodromal symptom of PD [59]. Approximately 50% of idiopathic RBD patients convert to PD within a decade, and ultimately, over 80% develop some form of neurodegenerative disease [59,60].
NMS bear clinical significance for early diagnosis. For instance, PD patients in the prodromal phase may present with NMS such as sleep disorders, olfactory dysfunction, mood disorders, constipation, and limb and trunk pain, but lack overt motor symptoms. These patients often visit sleep clinics, otorhinolaryngology clinics, psychological clinics, gastroenterology clinics, and orthopedics clinics, leading to delayed diagnosis. Furthermore, patients with limb pain may be mistakenly diagnosed with cervical or lumbar issues, and surgical intervention often fails to alleviate the pain. Employing an ML model can aid non-neurological clinicians in rapidly distinguishing PD patients from HCs, facilitating timely referrals to neurology departments for professional diagnosis and treatment.
There are several limitations to this study. First, as with all retrospective studies, potential selection biases and unknown confounding factors may not be accounted for in the analysis. To address these limitations, future work will include a prospective cohort study with a larger sample size. Second, as only one center was used for external validation, a larger sample size from multiple research centers is necessary to verify and improve the study’s results. Third, due to time and resource constraints, we were unable to examine the participants’ imaging indicators.

5. Conclusions

In conclusion, we identified multiple pivotal variables and SVM as a precise and clinician-friendly ML algorithm for the prediction of PD in Chinese patients. We believe that these ML algorithms, with further optimization, can serve as invaluable adjuncts to clinicians in the PD diagnostic process, potentially facilitating earlier and more accurate detection of the disease.

Supplementary Materials

The following supporting information can be downloaded at: https://www.mdpi.com/article/10.3390/brainsci13111546/s1. Table S1: The primer sequences for amplifying each SNPs. Figure S1: Feature selection using Lasso regression. (A) Lasso coefficient profiles of the candidate features. (B) Selection of the optimal parameter (lambda) through 10-fold cross-validation. The left and right dotted vertical lines represent the optimal lambda values using the minimum error criterion and one standard error (1-SE) of the minimum criterion, respectively. Figure S2: Hyperparameter optimization for each machine learning model using five-fold cross-validation and a grid search technique. (A) Hyperparameter optimization for the decision tree model. (B) Hyperparameter optimization for the random forest model. (C) Hyperparameter optimization for the eXtreme gradient boosting model. (D) Hyperparameter optimization for the support vector machine model. (E) Hyperparameter optimization for the k-nearest neighbor model.

Author Contributions

H.L. designed the study, analyzed the data, interpreted the results, and wrote the manuscript. X.-Y.L., F.X., J.Z. and X.L. collected the data. Y.S., X.W. and Z.W. supervised the study. C.W. designed the study, supervised the study, and revised the manuscript. All authors have read and agreed to the published version of the manuscript.

Funding

This work was supported by grants from the National Natural Science Foundation of China (NSFC) (82171412) to Chaodong Wang and the National Key Research and Development Program (2022YFC2009600) to Xu-Ying Li.

Institutional Review Board Statement

The study was approved by the Medical Ethics Committee of Xuanwu Hospital of Capital Medical University ([2020]060).

Informed Consent Statement

Informed consent was obtained from all subjects involved in the study. Written informed consent was obtained from all patients to publish this paper.

Data Availability Statement

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

Conflicts of Interest

The authors declare no conflict of interest.

Abbreviations

PD: Parkinson’s disease; REM: rapid eye movement; RBD: REM sleep behavior disorder; NMS: non-motor symptoms; SNPs: single-nucleotide polymorphisms; iRBD: idiopathic RBD; pRBD: possible REM sleep behavior disorder; GWAS: genome-wide association studies; CI: confidence interval; OR: odds ratio; IQRs: interquartile ranges; RSS: residual sum of squares; Lasso: least absolute shrinkage and selection operator; ROC: receiver operating characteristic; AUC: area under the ROC curve; DCA: decision curve analysis; NRI: net reclassification improvement; IDI: integrated discrimination improvement; ML: machine learning; DT: decision tree; RF: random forest; XGboost: eXtreme gradient boosting; SVM: support vector machine; KNN: k-nearest neighbor; HCs: healthy controls; MSA: multiple system atrophy; PSP: progressive supranuclear palsy; DLB: dementia with Lewy bodies; CBD: corticobasal degeneration; PCR: polymerase chain reaction.

References

  1. De Lau, L.M.L.; Breteler, M.M.B. Epidemiology of Parkinson’s disease. Lancet Neurol. 2006, 5, 525–535. [Google Scholar] [CrossRef]
  2. Ascherio, A.; Schwarzschild, M.A. The epidemiology of Parkinson’s disease: Risk factors and prevention. Lancet Neurol. 2016, 15, 1257–1272. [Google Scholar] [CrossRef] [PubMed]
  3. Yang, J.-X.; Chen, L. Economic Burden Analysis of Parkinson’s Disease Patients in China. Park. Dis. 2017, 2017, 8762939. [Google Scholar] [CrossRef] [PubMed]
  4. Johnson, S.J.; Diener, M.D.; Kaltenboeck, A.; Birnbaum, H.G.; Siderowf, A.D. An economic model of Parkinson’s disease: Implications for slowing progression in the United States. Mov. Disord. Off. J. Mov. Disord. Soc. 2013, 28, 319–326. [Google Scholar] [CrossRef] [PubMed]
  5. Wilczyński, J.; Ścipniak, M.; Ścipniak, K.; Margiel, K.; Wilczyński, I.; Zieliński, R.; Sobolewski, P. Assessment of Risk Factors for Falls among Patients with Parkinson’s Disease. BioMed Res. Int. 2021, 2021, 5531331. [Google Scholar] [CrossRef] [PubMed]
  6. Fearnley, J.M.; Lees, A.J. Ageing and Parkinson’s disease: Substantia nigra regional selectivity. Brain J. Neurol. 1991, 114 Pt 5, 2283–2301. [Google Scholar] [CrossRef] [PubMed]
  7. Rocca, W.A.; McDonnell, S.K.; Strain, K.J.; Bower, J.H.; Ahlskog, J.E.; Elbaz, A.; Schaid, D.J.; Maraganore, D.M. Familial aggregation of Parkinson’s disease: The Mayo Clinic family study. Ann. Neurol. 2004, 56, 495–502. [Google Scholar] [CrossRef] [PubMed]
  8. Pezzoli, G.; Cereda, E. Exposure to pesticides or solvents and risk of Parkinson disease. Neurology 2013, 80, 2035–2041. [Google Scholar] [CrossRef] [PubMed]
  9. Heinzel, S.; Berg, D.; Gasser, T.; Chen, H.; Yao, C.; Postuma, R.B.; The MDS Task Force on the Definition of Parkinson’s Disease. Update of the MDS research criteria for prodromal Parkinson’s disease. Mov. Disord. Off. J. Mov. Disord. Soc. 2019, 34, 1464–1470. [Google Scholar] [CrossRef]
  10. Noyce, A.J.; Bestwick, J.P.; Silveira-Moriyama, L.; Hawkes, C.H.; Giovannoni, G.; Lees, A.J.; Schrag, A. Meta-analysis of early nonmotor features and risk factors for Parkinson disease. Ann. Neurol. 2012, 72, 893–901. [Google Scholar] [CrossRef]
  11. Hu, G.; Bidel, S.; Jousilahti, P.; Antikainen, R.; Tuomilehto, J. Coffee and tea consumption and the risk of Parkinson’s disease. Mov. Disord. Off. J. Mov. Disord. Soc. 2007, 22, 2242–2248. [Google Scholar] [CrossRef] [PubMed]
  12. Liu, R.; Guo, X.; Park, Y.; Huang, X.; Sinha, R.; Freedman, N.D.; Hollenbeck, A.R.; Blair, A.; Chen, H. Caffeine Intake, Smoking, and Risk of Parkinson Disease in Men and Women. Am. J. Epidemiol. 2012, 175, 1200–1207. [Google Scholar] [CrossRef] [PubMed]
  13. Fang, X.; Han, D.; Cheng, Q.; Zhang, P.; Zhao, C.; Min, J.; Wang, F. Association of Levels of Physical Activity With Risk of Parkinson Disease: A Systematic Review and Meta-analysis. JAMA Netw. Open 2018, 1, e182421. [Google Scholar] [CrossRef] [PubMed]
  14. Foo, J.N.; Chew, E.G.Y.; Chung, S.J.; Peng, R.; Blauwendraat, C.; Nalls, M.A.; Mok, K.Y.; Satake, W.; Toda, T.; Chao, Y.; et al. Identification of Risk Loci for Parkinson Disease in Asians and Comparison of Risk between Asians and Europeans: A Genome-Wide Association Study. JAMA Neurol. 2020, 77, 746–754. [Google Scholar] [CrossRef] [PubMed]
  15. Nalls, M.A.; Blauwendraat, C.; Vallerga, C.L.; Heilbron, K.; Bandres-Ciga, S.; Chang, D.; Tan, M.; Kia, D.A.; Noyce, A.J.; Xue, A.; et al. Identification of novel risk loci, causal insights, and heritable risk for Parkinson’s disease: A meta-analysis of genome-wide association studies. Lancet Neurol. 2019, 18, 1091–1102. [Google Scholar] [CrossRef] [PubMed]
  16. Kawakami, E.; Tabata, J.; Yanaihara, N.; Ishikawa, T.; Koseki, K.; Iida, Y.; Saito, M.; Komazaki, H.; Shapiro, J.S.; Goto, C.; et al. Application of Artificial Intelligence for Preoperative Diagnostic and Prognostic Prediction in Epithelial Ovarian Cancer Based on Blood Biomarkers. Clin. Cancer Res. Off. J. Am. Assoc. Cancer Res. 2019, 25, 3006–3015. [Google Scholar] [CrossRef] [PubMed]
  17. Angraal, S.; Mortazavi, B.J.; Gupta, A.; Khera, R.; Ahmad, T.; Desai, N.R.; Jacoby, D.L.; Masoudi, F.A.; Spertus, J.A.; Krumholz, H.M. Machine Learning Prediction of Mortality and Hospitalization in Heart Failure with Preserved Ejection Fraction. JACC Heart Fail. 2020, 8, 12–21. [Google Scholar] [CrossRef]
  18. Wang, H.-H.; Wang, Y.-H.; Liang, C.-W.; Li, Y.-C. Assessment of Deep Learning Using Nonimaging Information and Sequential Medical Records to Develop a Prediction Model for Nonmelanoma Skin Cancer. JAMA Dermatol. 2019, 155, 1277–1283. [Google Scholar] [CrossRef]
  19. Ali, L.; Zhu, C.; Zhang, Z.; Liu, Y. Automated Detection of Parkinson’s Disease Based on Multiple Types of Sustained Phonations Using Linear Discriminant Analysis and Genetically Optimized Neural Network. IEEE J. Transl. Eng. Health Med. 2019, 7, 2000410. [Google Scholar] [CrossRef]
  20. Pereira, C.R.; Pereira, D.R.; Rosa, G.H.; Albuquerque, V.H.; Weber, S.A.; Hook, C.; Papa, J.P. Handwritten dynamics assessment through convolutional neural networks: An application to Parkinson’s disease identification. Artif. Intell. Med. 2018, 87, 67–77. [Google Scholar] [CrossRef]
  21. Caramia, C.; Torricelli, D.; Schmid, M.; Munoz-Gonzalez, A.; Gonzalez-Vargas, J.; Grandas, F.; Pons, J.L. IMU-Based Classification of Parkinson’s Disease From Gait: A Sensitivity Analysis on Sensor Location and Feature Selection. IEEE J. Biomed. Health Inform. 2018, 22, 1765–1774. [Google Scholar] [CrossRef] [PubMed]
  22. Zhang, J. Mining imaging and clinical data with machine learning approaches for the diagnosis and early detection of Parkinson’s disease. NPJ Park. Dis. 2022, 8, 13. [Google Scholar] [CrossRef] [PubMed]
  23. Maass, F.; Michalke, B.; Willkommen, D.; Leha, A.; Schulte, C.; Tönges, L.; Mollenhauer, B.; Trenkwalder, C.; Rückamp, D.; Börger, M.; et al. Elemental fingerprint: Reassessment of a cerebrospinal fluid biomarker for Parkinson’s disease. Neurobiol. Dis. 2020, 134, 104677. [Google Scholar] [CrossRef] [PubMed]
  24. Su, C.; Tong, J.; Wang, F. Mining genetic and transcriptomic data using machine learning approaches in Parkinson’s disease. NPJ Park. Dis. 2020, 6, 24. [Google Scholar] [CrossRef]
  25. Kang, J.-H.; Irwin, D.J.; Chen-Plotkin, A.S.; Siderowf, A.; Caspell, C.; Coffey, C.S.; Waligórska, T.; Taylor, P.; Pan, S.; Frasier, M.; et al. Association of Cerebrospinal Fluid β-Amyloid 1-42, T-tau, P-tau181, and α-Synuclein Levels with Clinical Features of Drug-Naive Patients with Early Parkinson Disease. JAMA Neurol. 2013, 70, 1277–1287. [Google Scholar] [CrossRef] [PubMed]
  26. Silveira-Moriyama, L.; Carvalho, M.d.J.; Katzenschlager, R.; Petrie, A.; Ranvaud, R.; Barbosa, E.R.; Lees, A.J. The use of smell identification tests in the diagnosis of Parkinson’s disease in Brazil. Mov. Disord. Off. J. Mov. Disord. Soc. 2008, 23, 2328–2334. [Google Scholar] [CrossRef] [PubMed]
  27. Shinde, S.; Prasad, S.; Saboo, Y.; Kaushick, R.; Saini, J.; Pal, P.K.; Ingalhalikar, M. Predictive markers for Parkinson’s disease using deep neural nets on neuromelanin sensitive MRI. NeuroImage Clin. 2019, 22, 101748. [Google Scholar] [CrossRef]
  28. Armañanzas, R.; Bielza, C.; Chaudhuri, K.R.; Martinez-Martin, P.; Larrañaga, P. Unveiling relevant non-motor Parkinson’s disease severity symptoms using a machine learning approach. Artif. Intell. Med. 2013, 58, 195–202. [Google Scholar] [CrossRef]
  29. Postuma, R.B.; Berg, D.; Stern, M.; Poewe, W.; Olanow, C.W.; Oertel, W.; Obeso, J.; Marek, K.; Litvan, I.; Lang, A.E.; et al. MDS clinical diagnostic criteria for Parkinson’s disease. Mov. Disord. Off. J. Mov. Disord. Soc. 2015, 30, 1591–1601. [Google Scholar] [CrossRef]
  30. Belvisi, D.; Pellicciari, R.; Fabbrini, A.; Costanzo, M.; Pietracupa, S.; De Lucia, M.; Modugno, N.; Magrinelli, F.; Dallocchio, C.; Ercoli, T.; et al. Risk factors of Parkinson disease: Sim-ultaneous assessment, interactions, and etiologic subtypes. Neurology 2020, 95, e2500–e2508. [Google Scholar] [CrossRef]
  31. Belvisi, D.; Pellicciari, R.; Fabbrini, G.; Tinazzi, M.; Berardelli, A.; Defazio, G. Modifiable risk and protective factors in disease development, progression and clinical subtypes of Parkinson’s disease: What do prospective studies suggest? Neurobiol. Dis. 2020, 134, 104671. [Google Scholar] [CrossRef] [PubMed]
  32. Wang, C.; Cai, Y.; Zheng, Z.; Tang, B.-S.; Xu, Y.; Wang, T.; Ma, J.; Chen, S.-D.; Langston, J.W.; Tanner, C.M.; et al. Penetrance of LRRK2 G2385R and R1628P is modified by common PD-associated genetic variants. Park. Relat. Disord. 2012, 18, 958–963. [Google Scholar] [CrossRef] [PubMed]
  33. Han, W.; Liu, Y.; Mi, Y.; Zhao, J.; Liu, D.; Tian, Q. Alpha-synuclein (SNCA) polymorphisms and susceptibility to Parkinson’s disease: A meta-analysis. Am. J. Med. Genet. Part B Neuropsychiatr. Genet. Off. Publ. Int. Soc. Psychiatr. Genet. 2015, 168, 123–134. [Google Scholar] [CrossRef] [PubMed]
  34. Chang, X.-L.; Mao, X.-Y.; Li, H.-H.; Zhang, J.-H.; Li, N.-N.; Burgunder, J.-M.; Peng, R.; Tan, E.-K. Association of GWAS loci with PD in China. Am. J. Med. Genet. Part B Neuropsychiatr. Genet. Off. Publ. Int. Soc. Psychiatr. Genet. 2011, 156, 334–339. [Google Scholar] [CrossRef] [PubMed]
  35. International Parkinson Disease Genomics Consortium. Imputation of sequence variants for identification of genetic risks for Parkinson’s disease: A meta-analysis of genome-wide association studies. Lancet 2011, 377, 641–649. [Google Scholar] [CrossRef] [PubMed]
  36. Wang, L.; Cheng, L.; Lu, Z.-J.; Sun, X.-Y.; Li, J.-Y.; Peng, R. Association of three candidate genetic variants in RAB7L1/NUCKS1, MCCC1 and STK39 with sporadic Parkinson’s disease in Han Chinese. J. Neural Transm. 2016, 123, 425–430. [Google Scholar] [CrossRef]
  37. Zhao, Y.; Qin, L.; Pan, H.; Liu, Z.; Jiang, L.; He, Y.; Zeng, Q.; Zhou, X.; Zhou, X.; Zhou, Y.; et al. The role of genetics in Parkinson’s disease: A large cohort study in Chinese mainland population. Brain J. Neurol. 2020, 143, 2220–2234. [Google Scholar] [CrossRef] [PubMed]
  38. Prashanth, R.; Roy, S.D.; Mandal, P.K.; Ghosh, S. High-Accuracy Detection of Early Parkinson’s Disease through Multimodal Features and Machine Learning. Int. J. Med. Inform. 2016, 90, 13–21. [Google Scholar] [CrossRef]
  39. Shu, Z.; Pang, P.; Wu, X.; Cui, S.; Xu, Y.; Zhang, M. An Integrative Nomogram for Identifying Early-Stage Parkinson’s Disease Using Non-motor Symptoms and White Matter-Based Radiomics Biomarkers from Whole-Brain MRI. Front. Aging Neurosci. 2020, 12, 548616. [Google Scholar] [CrossRef]
  40. Karabayir, I.; Butler, L.; Goldman, S.M.; Kamaleswaran, R.; Gunturkun, F.; Davis, R.L.; Ross, G.W.; Petrovitch, H.; Masaki, K.; Tanner, C.M.; et al. Predicting Parkinson’s Disease and Its Pathology via Simple Clinical Variables. J. Park. Dis. 2022, 12, 341–351. [Google Scholar] [CrossRef]
  41. Lin, S.; Gao, C.; Li, H.; Huang, P.; Ling, Y.; Chen, Z.; Ren, K.; Chen, S. Wearable sensor-based gait analysis to discriminate early Parkinson’s disease from essential tremor. J. Neurol. 2023, 270, 2283–2301. [Google Scholar] [CrossRef] [PubMed]
  42. Govindu, A.; Palwe, S. Early detection of Parkinson’s disease using machine learning. Procedia Comput. Sci. 2023, 218, 249–261. [Google Scholar] [CrossRef]
  43. He, S.; Huang, L.; Shao, C.; Nie, T.; Xia, L.; Cui, B.; Lu, F.; Zhu, L.; Chen, B.; Yang, Q. Several miRNAs derived from serum extracellular vesicles are potential biomarkers for early diagnosis and progression of Parkinson’s disease. Transl. Neurodegener. 2021, 10, 25. [Google Scholar] [CrossRef] [PubMed]
  44. Gopar-Cuevas, Y.; Duarte-Jurado, A.P.; Diaz-Perez, R.N.; Saucedo-Cardenas, O.; Loera-Arias, M.J.; Montes-De-Oca-Luna, R.; Rodriguez-Rocha, H.; Garcia-Garcia, A. Pursuing Multiple Biomarkers for Early Idiopathic Parkinson’s Disease Diagnosis. Mol. Neurobiol. 2021, 58, 5517–5532. [Google Scholar] [CrossRef] [PubMed]
  45. Mitchell, T.; Lehéricy, S.; Chiu, S.Y.; Strafella, A.P.; Stoessl, A.J.; Vaillancourt, D.E. Emerging Neuroimaging Biomarkers across Disease Stage in Parkinson Disease. JAMA Neurol. 2021, 78, 1262–1272. [Google Scholar] [CrossRef]
  46. Li, X.; Fan, X.; Yang, H.; Liu, Y. Review of Metabolomics-Based Biomarker Research for Parkinson’s Disease. Mol. Neurobiol. 2022, 59, 1041–1057. [Google Scholar] [CrossRef] [PubMed]
  47. Reichmann, H. Premotor Diagnosis of Parkinson’s Disease. Neurosci. Bull. 2017, 33, 526–534. [Google Scholar] [CrossRef] [PubMed]
  48. Goldman, J.G.; Postuma, R. Premotor and nonmotor features of Parkinson’s disease. Curr. Opin. Neurol. 2014, 27, 434–441. [Google Scholar] [CrossRef]
  49. Fullard, M.E.; Morley, J.F.; Duda, J.E. Olfactory Dysfunction as an Early Biomarker in Parkinson’s Disease. Neurosci. Bull. 2017, 33, 515–525. [Google Scholar] [CrossRef]
  50. Haehner, A.; Boesveldt, S.; Berendse, H.; Mackay-Sim, A.; Fleischmann, J.; Silburn, P.; Johnston, A.; Mellick, G.; Herting, B.; Reichmann, H.; et al. Prevalence of smell loss in Parkinson’s disease—A multicenter study. Park. Relat. Disord. 2009, 15, 490–494. [Google Scholar] [CrossRef]
  51. Ponsen, M.M.; Stoffers, D.; Booij, J.; van Eck-Smit, B.L.F.; Wolters, E.C.; Berendse, H.W. Idiopathic hyposmia as a preclinical sign of Parkinson’s disease. Ann. Neurol. 2004, 56, 173–181. [Google Scholar] [CrossRef] [PubMed]
  52. Braak, H.; Ghebremedhin, E.; Rüb, U.; Bratzke, H.; Del Tredici, K. Stages in the development of Parkinson’s disease-related pathology. Cell Tissue Res. 2004, 318, 121–134. [Google Scholar] [CrossRef] [PubMed]
  53. Abbott, R.D.; Petrovitch, H.; White, L.R.; Masaki, K.H.; Tanner, C.M.; Curb, J.D.; Grandinetti, A.; Blanchette, P.L.; Popper, J.S.; Ross, G.W. Frequency of bowel movements and the future risk of Parkinson’s disease. Neurology 2001, 57, 456–462. [Google Scholar] [CrossRef] [PubMed]
  54. Feng, F.; Cai, Y.; Hou, Y.; Ou, R.; Jiang, Z.; Shang, H. Excessive daytime sleepiness in Parkinson’s disease: A systematic review and meta-analysis. Park. Relat. Disord. 2021, 85, 133–140. [Google Scholar] [CrossRef]
  55. Abbott, R.D.; Ross, G.W.; White, L.R.; Tanner, C.M.; Masaki, K.H.; Nelson, J.S.; Curb, J.D.; Petrovitch, H. Excessive daytime sleepiness and subsequent development of Parkinson disease. Neurology 2005, 65, 1442–1446. [Google Scholar] [CrossRef] [PubMed]
  56. Williams-Gray, C.H.; Mason, S.L.; Evans, J.R.; Foltynie, T.; Brayne, C.; Robbins, T.W.; Barker, R.A. The CamPaIGN study of Parkinson’s disease: 10-year outlook in an incident population-based cohort. J. Neurol. Neurosurg. Psychiatry 2013, 84, 1258–1264. [Google Scholar] [CrossRef] [PubMed]
  57. Schuurman, A.G.; Akker, M.v.D.; Ensinck, K.T.; Metsemakers, J.F.; Knottnerus, J.A.; Leentjens, A.F.; Buntinx, F. Increased risk of Parkinson’s disease after depression: A retrospective cohort study. Neurology 2002, 58, 1501–1504. [Google Scholar] [CrossRef] [PubMed]
  58. Leentjens, A.F.; Akker, M.V.D.; Metsemakers, J.F.; Lousberg, R.; Verhey, F.R. Higher incidence of depression preceding the onset of Parkinson’s disease: A register study. Mov. Disord. Off. J. Mov. Disord. Soc. 2003, 18, 414–418. [Google Scholar] [CrossRef]
  59. Howell, M.J.; Schenck, C.H. Rapid Eye Movement Sleep Behavior Disorder and Neurodegenerative Disease. JAMA Neurol. 2015, 72, 707–712. [Google Scholar] [CrossRef]
  60. Postuma, R.B.; Iranzo, A.; Hu, M.; Högl, B.; Boeve, B.F.; Manni, R.; Oertel, W.H.; Arnulf, I.; Ferini-Strambi, L.; Puligheddu, M.; et al. Risk and predictors of dementia and parkinsonism in idiopathic REM sleep behaviour disorder: A multicentre study. Brain J. Neurol. 2019, 142, 744–759. [Google Scholar] [CrossRef]
Brainsci 13 01546 g001
Figure 1. Flow diagram of selection of patients.
Figure 1. Flow diagram of selection of patients.
Brainsci 13 01546 g001
Brainsci 13 01546 g002
Figure 2. Receiver operating characteristic (ROC) curve of LASSO-LR, decision tree, random forest, XGboost, SVM, and KNN for the training set and validation set. (A) Integration of ROC for the PD classification model based on the six models for the training set. (B) Integration of ROC for the PD classification model based on the six models for the validation set.
Figure 2. Receiver operating characteristic (ROC) curve of LASSO-LR, decision tree, random forest, XGboost, SVM, and KNN for the training set and validation set. (A) Integration of ROC for the PD classification model based on the six models for the training set. (B) Integration of ROC for the PD classification model based on the six models for the validation set.
Brainsci 13 01546 g002
Brainsci 13 01546 g003
Figure 3. Variable importance. Histograms (A–F) depict the proportion of factor importance for different predictors in the model. For each model, the relative importance is quantified by assigning a weight between 0 and 1 for each variable.
Figure 3. Variable importance. Histograms (A–F) depict the proportion of factor importance for different predictors in the model. For each model, the relative importance is quantified by assigning a weight between 0 and 1 for each variable.
Brainsci 13 01546 g003
Table 1. Characteristics of total participants.
Table 1. Characteristics of total participants.
Training Set (n = 1028)pValidation Set (n = 628)p
PDHCPDHC
(n = 524)(n = 504)(n = 305)(n = 323)
Demographics
Age, year66 (61, 71.25)67 (63, 70)0.22269 (61, 75)68 (65, 73)0.159
Male257 (49)227 (45)0.221173 (56.7)162 (50.2)0.117
Education, year9 (5, 11.12)9 (9, 12)<0.0018 (5, 11)9 (6, 11)0.001
Family History of PD44 (8.4)12 (2.4)<0.00127 (8.9)3 (0.9)<0.001
Lifestyle behaviors
Smoking124 (23.7)188 (37.3)<0.00182 (26.9)121 (37.5)0.006
Alcohol 175 (33.4)209 (41.5)0.00962 (20.3)145 (44.9)<0.001
Tea 203 (38.7)323 (64.1)<0.00157 (18.7)213 (65.9)<0.001
Coffee66 (12.6)93 (18.5)0.0128 (2.6)32 (9.9)<0.001
Environmental exposure
Pesticides157 (30)110 (21.8)0.00491 (29.8)80 (24.8)0.181
Organic solvent13 (2.5)18 (3.6)0.4015 (1.6)10 (3.1)0.351
Heavy metal12 (2.3)18 (3.6)0.3014 (1.3)6 (1.9)0.753
Head injury 21 (4)12 (2.4)0.1932 (0.7)6 (1.9)0.288
General anesthesia105 (20)117 (23.2)0.24544 (14.4)51 (15.8)0.715
Non-motor symptoms
pRBD172 (32.8)28 (5.6)<0.00175 (24.6)12 (3.7)<0.001
Olfactory dysfunction319 (60.9)63 (12.5)<0.001217 (71.1)38 (11.8)<0.001
Constipation347 (66.2)73 (14.5)<0.001169 (55.4)54 (16.7)<0.001
Global cognitive deficit273 (52.1)97 (19.2)<0.001185 (60.7)103 (31.9)<0.001
Depression242 (46.2)63 (12.5)<0.00144 (14.4)17 (5.3)<0.001
Daytime somnolence216 (41.2)37 (7.3)<0.001161 (52.8)47 (14.6)<0.001
Gene mutation
MMRN1 rs6532194208 (39.7)143 (28.4)<0.001106 (34.8)105 (32.5)0.609
RAB7L1 rs823144192 (36.6)144 (28.6)0.007127 (41.6)94 (29.1)0.001
SNCA rs356182405 (77.3)326 (64.7)<0.001248 (81.3)182 (56.3)<0.001
LRRK2 rs3477834865 (12.4)40 (7.9)0.02423 (7.5)23 (7.1)0.961
SNCA rs356219208 (39.7)134 (26.6)<0.001109 (35.7)98 (30.3)0.176
MCCC1 rs12637471352 (67.2)294 (58.3)0.004218 (71.5)182 (56.3)<0.001
GBA rs4210167 (1.3)0 (0)0.0155 (1.6)0 (0)0.027
Abbreviations: pRBD = possible REM sleep behavior disorder; HC = healthy control. Values are n (%) or medians and interquartile range (IQRs).
Table 2. A comparison of the model performances with the training and validation set.
Table 2. A comparison of the model performances with the training and validation set.
LASSO-LRDTRFXGboostSVMKNN
Training set
AUC0.930
(95% CI:
0.915–0.945) 		0.888
(95% CI:
0.868–0.909)		0.963
(95% CI:
0.952 −0.973)		0.946
(95% CI:
0.946–0.959)		0.941
(95% CI:
0.927–0.955)		0.941
(95% CI:
0.928–0.954)
Accuracy0.861
(95% CI:
0.840–0.881)		0.851
(95% CI:
0.829–0.873)		0.911
(95% CI:
0.892–0.928)		0.884
(95% CI:
0.865–0.904)		0.874
(95% CI:
0.854–0.896)		0.870
(95% CI:
0.849–0.890)
Sensitivity (Recall)0.826
(95% CI:
0.789–0.860)		0.834
(95% CI:
0.803–0.866)		0.910
(95% CI:
0.884–0.936)		0.880
(95% CI:
0.850–0.909)		0.838
(95% CI:
0.805–0.872)		0.891
(95% CI:
0.863–0.919)
Specificity0.897
(95% CI:
0.870–0.924)		0.869
(95% CI:
0.840–0.898)		0.911
(95% CI:
0.884–0.936)		0.889
(95% CI:
0.860–0.916)		0.911
(95% CI:
0.885–0.936)		0.847
(95% CI:
0.817–0.880)
Precision0.893
(95% CI:
0.866–0.919)		0.869
(95% CI:
0.839–0.897)		0.914
(95% CI:
0.890–0.937)		0.892
(95% CI:
0.867–0.918)		0.907
(95% CI:
0.882–0.933)		0.859
(95% CI:
0.831–0.888)
F1 score0.858
(95% CI:
0.834–0.879)		0.851
(95% CI:
0.828–0.875)		0.912
(95% CI:
0.894–0.930)		0.886
(95% CI:
0.866–0.905)		0.871
(95% CI:
0.849–0.895)		0.875
(95% CI:
0.854–0.896)
Validation set
AUC0.925
(95% CI:
0.905–0.945)		0.831
(95% CI:
0.80–0.862)		0.912
(95% CI:
0.890–0.934)		0.908
(95% CI:
0.886–0.931)		0.928
(95% CI:
0.908–0.947)		0.896
(95% CI:
0.871–0.921)
Accuracy0.842
(95% CI:
0.812–0.869)		0.787
(95% CI:
0.753–0.820)		0.834
(95% CI:
0.804–0.863)		0.819
(95% CI:
0.788–0.849)		0.844
(95% CI:
0.814–0.871)		0.811
(95% CI:
0.779–0.841)
Sensitivity0.810
(95% CI:
0.764–0.853)		0.751
(95% CI:
0.701–0.80)		0.866
(95% CI:
0.826–0.902)		0.889
(95% CI:
0.851–0.922)		0.826
(95% CI:
0.786–0.866)		0.830
(95% CI:
0.779–0.841)
Specificity0.873
(95% CI:
0.834–0.909)		0.820
(95% CI:
0.777–0.862)		0.805
(95% CI:
0.764–0.848)		0.752
(95% CI:
0.707–0.799)		0.861
(95% CI:
0.820–0.898)		0.793
(95% CI:
0.747–0.835)
Precision0.858
(95% CI:
0.816–0.90)		0.798
(95% CI:
0.751–0.845)		0.807
(95% CI:
0.765–0.851)		0.772
(95% CI:
0.727–0.819)		0.849
(95% CI:
0.807–0.891)		0.791
(95% CI:
0.746–0.838)
F1 score0.833
(95% CI:
0.799–0.864)		0.774
(95% CI:
0.731–0.811)		0.835
(95% CI:
0.802–0.867)		0.826
(95% CI:
0.794–0.858)		0.837
(95% CI:
0.803–0.868)		0.810
(95% CI:
0.774–0.842)
Abbreviations: LASSO-LR = least absolute shrinkage and selection operator–logistic regression; DT = decision tree; RF = random forest; XGboost = eXtreme gradient boosting; SVM = support vector machine; KNN = k-nearest neighbor; AUC = area under the curve of the receiver operating characteristic curve; CI = confidence interval.
Table 3. Comparative Analysis of Parkinson’s Disease Diagnosis Using Machine Learning Approaches.
Table 3. Comparative Analysis of Parkinson’s Disease Diagnosis Using Machine Learning Approaches.
ReferenceModeling
Approach		Features SelectionObjectiveSource of DataNo. of SubjectsEvaluation
R. Prashanth et al. [38] Naïve Bayes, SVM, Boosted Trees, RFNMS,
CSF,
dopaminergic imaging markers 		Classification
of PD from HC 		PPMI401 PD +
183 HC		Best performing:
SVM with
Accuracy = 96.40%, Sensitivity = 97.03%, Specificity = 95.01%,
AUC = 98.88%.
Shu et al. [39]Nomogram MRI,
clinical characteristics,
NMS		Identification of early-stage
PD		PPMI168 PD + 168HC
+ atypical
PD 58		AUC of training,
testing and verification sets were 0.937, 0.922, and 0.836, respectively; the specificity
were 83.8, 88.2, and 91.38%, respectively; and the sensitivity were 84.6, 82.4, and 70.69%.
Karabayir et al. [40]Light Gradient BoostingQuestionnaires;
simple non-invasive clinical tests		To predict a future diagnosis of PD HAAS 292 subjectsIndividuals who developed PD within 3 years: AUC = 0.82, (95% CI 0.76–0.89),
5 years:
AUC = 0.77 (95% CI 0.71–0.84).
Lin et al. [41] SVM
RF		Gathering gait and postural transition data using wearable sensors.To diferentiate early-stage PD from ET Ruijin Hospital, Shanghai Jiao Tong University School of Medicine 84 early-stage PD and 80 ET subjects Best performing:
weighted average ensemble classifcation models with
accuracy = 84%,
sensitivity = 85.9%,
Specifcity = 82.1%,
AUC = 0.912.
Govindu et al. [42] SVM,
RF,
KNN,
Logistic Regression models		MDVP audio data Early detection of PDPPMI23PD +
8 HC		Best performing:
RF with
accuracy = 91.83%,
sensitivity = 0.95.
Abbreviations: PD = Parkinson’s disease; HC = healthy control; CSF = Cerebrospinal fluid; NMS = nonmotor symptoms; SVM = support vector machine; RF = random forest; AUC = area under the curve; PPMI = Parkinson’s Progress Markers Initiative; HAAS = Honolulu–Asia Aging Study; ET = essential tremor; KNN = k-nearest neighbors.
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Lai, H.; Li, X.-Y.; Xu, F.; Zhu, J.; Li, X.; Song, Y.; Wang, X.; Wang, Z.; Wang, C. Applications of Machine Learning to Diagnosis of Parkinson’s Disease. Brain Sci. 2023, 13, 1546. https://doi.org/10.3390/brainsci13111546

AMA Style

Lai H, Li X-Y, Xu F, Zhu J, Li X, Song Y, Wang X, Wang Z, Wang C. Applications of Machine Learning to Diagnosis of Parkinson’s Disease. Brain Sciences. 2023; 13(11):1546. https://doi.org/10.3390/brainsci13111546

Chicago/Turabian Style

Lai, Hong, Xu-Ying Li, Fanxi Xu, Junge Zhu, Xian Li, Yang Song, Xianlin Wang, Zhanjun Wang, and Chaodong Wang. 2023. "Applications of Machine Learning to Diagnosis of Parkinson’s Disease" Brain Sciences 13, no. 11: 1546. https://doi.org/10.3390/brainsci13111546

APA Style

Lai, H., Li, X.-Y., Xu, F., Zhu, J., Li, X., Song, Y., Wang, X., Wang, Z., & Wang, C. (2023). Applications of Machine Learning to Diagnosis of Parkinson’s Disease. Brain Sciences, 13(11), 1546. https://doi.org/10.3390/brainsci13111546

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop