Normative Model-Based Quantile Rank EEG Features Enhance Machine Learning for Psychiatric Diagnosis
Poster No:
487
Submission Type:
Abstract Submission
Authors:
Qiwei Dong1,2,3, Yuxi Zhou4,5, Zongwen Feng4,5, Mingjun Duan2, Yongxiu Lai2, jianfu li4, Li Dong4,5, Dezhong Yao3,4,5
Institutions:
1Institute of Basic Medical Sciences (IBMS), CAMS & PUMC, Beijing,China, 2The Clinical Hospital of Chengdu Brain Science Institute, MOE Key Lab for Neuroinformation,UESTC, Chengdu,Sichuan,China, 3Research Unit of NeuroInformation, Chinese Academy of Medical Sciences, Chengdu,Sichuan,China, 4School of Life Science and Technology,University of Electronic Science and Technology of China, Chengdu,Sichuan,China, 5Sichuan Institute for Brain Science and Brain-Inspired Intelligence, Chengdu,Sichuan,China
First Author:
Qiwei Dong
Institute of Basic Medical Sciences (IBMS), CAMS & PUMC|The Clinical Hospital of Chengdu Brain Science Institute, MOE Key Lab for Neuroinformation,UESTC|Research Unit of NeuroInformation, Chinese Academy of Medical Sciences
Beijing,China|Chengdu,Sichuan,China|Chengdu,Sichuan,China
Institute of Basic Medical Sciences (IBMS), CAMS & PUMC|The Clinical Hospital of Chengdu Brain Science Institute, MOE Key Lab for Neuroinformation,UESTC|Research Unit of NeuroInformation, Chinese Academy of Medical Sciences
Beijing,China|Chengdu,Sichuan,China|Chengdu,Sichuan,China
Co-Author(s):
Yuxi Zhou
School of Life Science and Technology,University of Electronic Science and Technology of China|Sichuan Institute for Brain Science and Brain-Inspired Intelligence
Chengdu,Sichuan,China|Chengdu,Sichuan,China
School of Life Science and Technology,University of Electronic Science and Technology of China|Sichuan Institute for Brain Science and Brain-Inspired Intelligence
Chengdu,Sichuan,China|Chengdu,Sichuan,China
Zongwen Feng
School of Life Science and Technology,University of Electronic Science and Technology of China|Sichuan Institute for Brain Science and Brain-Inspired Intelligence
Chengdu,Sichuan,China|Chengdu,Sichuan,China
School of Life Science and Technology,University of Electronic Science and Technology of China|Sichuan Institute for Brain Science and Brain-Inspired Intelligence
Chengdu,Sichuan,China|Chengdu,Sichuan,China
Mingjun Duan
The Clinical Hospital of Chengdu Brain Science Institute, MOE Key Lab for Neuroinformation,UESTC
Chengdu,Sichuan,China
The Clinical Hospital of Chengdu Brain Science Institute, MOE Key Lab for Neuroinformation,UESTC
Chengdu,Sichuan,China
Yongxiu Lai
The Clinical Hospital of Chengdu Brain Science Institute, MOE Key Lab for Neuroinformation,UESTC
Chengdu,Sichuan,China
The Clinical Hospital of Chengdu Brain Science Institute, MOE Key Lab for Neuroinformation,UESTC
Chengdu,Sichuan,China
jianfu li
School of Life Science and Technology,University of Electronic Science and Technology of China
Chengdu,Sichuan,China
School of Life Science and Technology,University of Electronic Science and Technology of China
Chengdu,Sichuan,China
Li Dong
School of Life Science and Technology,University of Electronic Science and Technology of China|Sichuan Institute for Brain Science and Brain-Inspired Intelligence
Chengdu,Sichuan,China|Chengdu,Sichuan,China
School of Life Science and Technology,University of Electronic Science and Technology of China|Sichuan Institute for Brain Science and Brain-Inspired Intelligence
Chengdu,Sichuan,China|Chengdu,Sichuan,China
Dezhong Yao
Research Unit of NeuroInformation, Chinese Academy of Medical Sciences|School of Life Science and Technology,University of Electronic Science and Technology of China|Sichuan Institute for Brain Science and Brain-Inspired Intelligence
Chengdu,Sichuan,China|Chengdu,Sichuan,China|Chengdu,Sichuan,China
Research Unit of NeuroInformation, Chinese Academy of Medical Sciences|School of Life Science and Technology,University of Electronic Science and Technology of China|Sichuan Institute for Brain Science and Brain-Inspired Intelligence
Chengdu,Sichuan,China|Chengdu,Sichuan,China|Chengdu,Sichuan,China
Introduction:
Mental and behavioral disorders contribute to approximately 25% of the global disease burden, posing challenges for diagnosis due to reliance on subjective clinical assessments influenced by disease heterogeneity and comorbidity(Singh, 2023). Electroencephalography (EEG), as a non-invasive, cost-effective, and high-temporal resolution tool, has demonstrated its utility in objectively diagnosing neurological disorders such as epilepsy, traumatic brain injury, and attention deficit hyperactivity disorder (ADHD). With the advancement of artificial intelligence, machine learning (ML) has shown promise for classifying mental disorders. However, due to the fact of imbalanced data distribution, heterogeneity, and extreme values, the classification performance of ML should be further improved using features derived from normative models(Verdonck, 2024).
Methods:
Resting-state EEG data for mental disorders were collected from 7,752 participants. The dataset was preprocessed using the WeBrain toolbox (Dong, 2021), and phase synchronization indices (PSI) of brain networks were calculated. According to the ICD-10 (2019 version), participants were categorized into three groups: organic mental disorders (OMD, n = 555, age range 8–96 years), schizophrenia spectrum and delusional disorders (SSD, n = 5945, age range 5–98 years), and mood affective disorders (MAD, n = 1252, age range 12–89 years). Subsequently, 30 participants from each group were randomly selected for training machine learning models, while the remaining data were used to establish normative models based on Generalized Additive Model for Location, Scale, and Shape (GAMLSS)(Bozek, 2023). Quantile rank features (QRN) were derived from the normative models, and classification models were trained separately using QRN and PSI features with a one-vs-one approach. Support vector machine (SVM) and k-nearest neighbor (KNN) algorithms were employed, and their performance was evaluated using 5-fold cross-validation. Finally, classification accuracy, receiver operating characteristic (ROC) curves, and other performance metrics were systematically compared across models.
Results:
As shown in Figure 1, the classification accuracies of both SVM and KNN classifiers were significantly higher when using QRN features (78.43% for SVM and 79.24% for KNN) compared to PSI features (71.16% for SVM and 74.71% for KNN). Similarly, the results from the ROC curves and confusion matrices also demonstrated superior performance of classifiers based on QRN features (QRN-based vs. PSI-based AUC: 0.83 vs. 0.79 for SVM and 0.84 vs. 0.72 for KNN).
·Workflow (a), normative model (b); SVM: accuracy (c), ROC (d), PSI- (e) QRN- (f)based confusion matrices , QRN; KNN: accuracy (g), ROC (h), PSI- (i), QRN- (j) based confusion matrices. (* = p < 0.05)
Conclusions:
This study pioneers the use of EEG normative model-based QRN features in ML, demonstrating their advantages in multiclass classification for mental disorders, offering a novel approach to improving ML performance in diagnosing mental disorders.
Disorders of the Nervous System:
Psychiatric (eg. Depression, Anxiety, Schizophrenia) 1
Modeling and Analysis Methods:
Classification and Predictive Modeling 2
Keywords:
Electroencephaolography (EEG)
Machine Learning
Modeling
Other - Normative modeling
1|2Indicates the priority used for review
Abstract Information
By submitting your proposal, you grant permission for the Organization for Human Brain Mapping (OHBM) to distribute your work in any format, including video, audio print and electronic text through OHBM OnDemand, social media channels, the OHBM website, or other electronic publications and media.
The Open Science Special Interest Group (OSSIG) is introducing a reproducibility challenge for OHBM 2025. This new initiative aims to enhance the reproducibility of scientific results and foster collaborations between labs. Teams will consist of a “source” party and a “reproducing” party, and will be evaluated on the success of their replication, the openness of the source work, and additional deliverables. Click here for more information. Propose your OHBM abstract(s) as source work for future OHBM meetings by selecting one of the following options:
Please indicate below if your study was a "resting state" or "task-activation” study.
Healthy subjects only or patients (note that patient studies may also involve healthy subjects):
Was this research conducted in the United States?
Were any human subjects research approved by the relevant Institutional Review Board or ethics panel? NOTE: Any human subjects studies without IRB approval will be automatically rejected.
Were any animal research approved by the relevant IACUC or other animal research panel? NOTE: Any animal studies without IACUC approval will be automatically rejected.
Please indicate which methods were used in your research:
Provide references using APA citation style.
Verdonck, T. (2024). Special issue on feature engineering editorial. Machine Learning, 113(7), 3917-3928.
Dong, L. (2021). WeBrain: A web-based brainformatics platform of computational ecosystem for EEG big data analysis. Neuroimage, 245, 118713.
Bozek, J. (2023). Normative models for neuroimaging markers: Impact of model selection, sample size and evaluation criteria. Neuroimage, 268, 119864.