Have a personal or library account? Click to login
Classifying Obsessive-Compulsive Disorder from Resting-State EEG Using Convolutional Neural Networks: A Pilot Study Cover

Classifying Obsessive-Compulsive Disorder from Resting-State EEG Using Convolutional Neural Networks: A Pilot Study

Open Access
|Jan 2026

Figures & Tables

Table 1

Descriptive Statistics for Demographics and Clinical Characteristics.

CONTINUOUS VARIABLEOVERALL (M, SD)OCD (M, SD)HC (M, SD)
Age, years35.42 (13.46)34.89 (14.41)35.90 (13.31)
DOCS Total17.00 (16.98)20.80 (17.99)13.20 (15.91)
MIS Score7.00 (6.33)5.90 (5.92)8.38 (6.97)
BAI Total13.55 (13.34)13.30 (11.87)13.80 (15.32)
BDI-II Total11.50 (13.34)16.00 (14.50)7.00 (10.98)
CATEGORICAL VARIABLESN (%)n(%)n (%)
Sex
    Male13 (65.0%)6 (60.0%)7 (70.0%)
    Female7 (35.0%)4 (40.0%)3 (30.0%)
Education Level
    Some College2 (10.0%)2 (20.0%)0 (0.0%)
    Bachelor Degree2 (10.0%)2 (20.0%)0 (0.0%)
    Master’s Degree1 (5.0%)0 (0.0%)1 (10.0%)
    Doctoral Degree6 (30.0%)4 (40.0%)2 (20.0%)
    Other/Professional9 (45.0%)2 (20.0%)7 (70.0%)

[i] Note: N = 20 total; n = 10 participants in each group. HC = Healthy Control; OCD = Obsessive-Compulsive Disorder; DOCS: Dimensional Obsessive-Compulsive Scale; MIS: Magical Ideation Scale; BAI: Beck Anxiety Inventory; BDI-II: Beck Depression Inventory-II.

Table 2

Subject-Level Classification Performance of the CNN.

CLASSPRECISIONRECALLF1-SCORESUPPORT
HC (0)0.890.800.8410
OCD (1)0.820.900.8610
Accuracy0.8520
Macro Avg0.850.850.8520
Wtd Avg0.850.850.8520
cpsy-10-1-149-g1.png
Figure 1

Subject-Level CNN Performance. (Left) The confusion matrix for subject-level predictions (N = 20). (Right) The Receiver Operator Curve (ROC) for the subject-level EEG scores, with an Area Under the Curve (AUC) of 0.88.

Table 3

Subject-Level Classification Performance of the SVM.

CLASSPRECISIONRECALLF1-SCORESUPPORT
HC (0)0.450.500.4710
OCD (1)0.440.400.4210
Accuracy0.4520
Macro Avg0.450.450.4520
Wtd Avg0.450.450.4520

[i] Notes: Performance metrics are calculated based on one prediction per subject (N = 20).

cpsy-10-1-149-g2.png
Figure 2

Subject-Level SVM Performance. (Left) The confusion matrix for subject-level predictions (N = 20). (Right) The receiver operator curve for the subject-level EEG scores, with an area under the curve (AUC) of 0.47, indicating chance-level performance.

Table 4

Predictive Performance for CNN + Clinical and Demographic Features.

FEATURENAUCACCURACYPRECISIONRECALLF1CHANGE (AUC)
Baseline EEG200.860.800.800.800.80
DOCS Total200.860.800.800.800.80None
MIS Score180.810.780.750.750.75Worse
BAI Total200.860.800.800.800.80None
BDI-II Total200.840.800.800.800.80Worse
Age190.760.790.800.800.80Worse
Sex200.810.800.800.800.80Worse
Education200.840.800.800.800.80Worse

[i] Note: Baseline metrics include only the convolutional neural network’s subject-level EEG scores in the second stage. N = number of subjects remaining after handling missing values. DOCS: Dimensional Obsessive-Compulsive Scale; MIS: Magical Ideation Scale; BAI: Beck Anxiety Inventory; BDI-II: Beck Depression Inventory-II.

Table 5

Predictive Performance for Clinical and Demographic Features.

FEATURENAUCACCURACYPRECISIONRECALLF1CHANGE
Baseline EEG200.470.450.450.450.45Baseline
DOCS Total200.970.850.890.800.84Improved
MIS Score180.620.670.710.560.63Improved
BAI Total200.940.900.900.900.90Improved
BDI-II Total200.900.901.000.800.89Improved
Age190.280.370.360.440.40Worse
Sex200.180.350.390.500.44Worse
Education200.380.450.430.300.353Worse

[i] Note: Baseline metrics include only the support vector machine’s subject-level EEG scores in the second stage. N = number of subjects remaining after handling missing values. DOCS: Dimensional Obsessive-Compulsive Scale; MIS: Magical Ideation Scale; BAI: Beck Anxiety Inventory; BDI-II: Beck Depression Inventory-II.

DOI: https://doi.org/10.5334/cpsy.149 | Journal eISSN: 2379-6227
Language: English
Submitted on: May 15, 2025
|
Accepted on: Dec 4, 2025
|
Published on: Jan 16, 2026
Published by: Ubiquity Press
In partnership with: Paradigm Publishing Services
Publication frequency: 1 issue per year

© 2026 Brian Zaboski, Sarah Fineberg, Patrick Skosnik, Stephen Kichuk, Madison Fitzpatrick, Christopher Pittenger, published by Ubiquity Press
This work is licensed under the Creative Commons Attribution 4.0 License.

Volume 10 (2026): Issue 1