Resting-state functional connectivity is widely used in machine learning-based classification of neuropsychiatric disorders. Quantum machine learning models have been proposed for functional connectivity data, but few studies have evaluated them under leave-site-out validation or against matched classical models. We compared six quantum model families with classical ones using identical correlation-derived features across four public cohorts (4,308 subjects). Leave-site-out validation was used for the Autism Brain Imaging Data Exchange (ABIDE I), ADHD-200 (attention-deficit/hyperactivity disorder), and REST-meta-MDD (major depressive disorder) cohorts, while the UCLA Consortium for Neuropsychiatric Phenomics (UCLA-CNP) was evaluated using stratified 10-fold cross-validation. The quantum kernel was statistically indistinguishable from a matched classical kernel across all four cohorts, with the average accuracy difference within 1.4 percentage points at 95% confidence. On ABIDE I, a hybrid quantum convolutional network achieved similar accuracy to its classical counterpart (0.645 vs. 0.641), while a trainable quantum kernel exceeded its classical counterpart by 0.064 in area under the curve (AUC). Quantum convolutional networks also achieved higher AUC than their matched classical networks on this cohort and showed smaller train-test gaps on all four cohorts, without an accuracy gain. Overall, most quantum models matched their classical counterparts, but none consistently outperformed them. This study, among the first quantum evaluations of ABIDE I under leave-site-out validation, provides a feature-matched benchmark of quantum and classical models comprising 284 matched comparisons. We further analyse fidelity-based graph construction and account for the observed quantum-classical parity through effective-dimension and eigenspectrum analyses.
Resting-state functional connectivity is widely used in machine learning-based classification of neuropsychiatric disorders. Quantum machine learning models have been proposed for functional connectivity data, but few studies have evaluated them under leave-site-out validation or against matched classical models. We compared six quantum model families with classical ones using identical correlation-derived features across four public cohorts (4,308 subjects). Leave-site-out validation was used for the Autism Brain Imaging Data Exchange (ABIDE I), ADHD-200 (attention-deficit/hyperactivity disorder), and REST-meta-MDD (major depressive disorder) cohorts, while the UCLA Consortium for Neuropsychiatric Phenomics (UCLA-CNP) was evaluated using stratified 10-fold cross-validation. The quantum kernel was statistically indistinguishable from a matched classical kernel across all four cohorts, with the average accuracy difference within 1.4 percentage points at 95% confidence. On ABIDE I, a hybrid quantum convolutional network achieved similar accuracy to its classical counterpart (0.645 vs. 0.641), while a trainable quantum kernel exceeded its classical counterpart by 0.064 in area under the curve (AUC). Quantum convolutional networks also achieved higher AUC than their matched classical networks on this cohort and showed smaller train-test gaps on all four cohorts, without an accuracy gain. Overall, most quantum models matched their classical counterparts, but none consistently outperformed them. This study, among the first quantum evaluations of ABIDE I under leave-site-out validation, provides a feature-matched benchmark of quantum and classical models comprising 284 matched comparisons. We further analyse fidelity-based graph construction and account for the observed quantum-classical parity through effective-dimension and eigenspectrum analyses.
Abstract Psychiatric disorders such as attention deficit hyperactivity disorder, schizophrenia, and bipolar disorder are notoriously difficult to distinguish on neuroimaging alone, owing to overlapping symptom profiles and heterogeneous neurobiological substrates. Conventional machine learning classifiers applied to structural MRI and electroencephalography data have shown modest discriminative performance and, more critically, tend to produce poorly calibrated probability estimates that limit their utility in clinical decision-making. We present a dual-pipeline framework in which quantum kernel support vector machines are applied independently to MRI-derived morphometric features and to EEG spectral representations, followed by rigorous post-hoc calibration and interpretability analyses. On the UCLA Consortium for Neuropsychiatric Phenomics dataset, a 4-qubit quantum kernel SVM achieved an AUC of 0.957 and a Matthews correlation coefficient of 0.679 for ADHD versus control discrimination, while a tuned classical radial basis function SVM performed at chance (AUC=0.497). For EEG-based emotion and cognitive state classification, a hybrid quantum-classical kernel outperformed both its pure quantum and classical counterparts (AUC=0.841 versus 0.824 and 0.812, respectively). Temperature scaling reduced expected calibration error by 85% for the MRI pipeline and by 49% for the EEG pipeline. Multimodal fusion of the two quantum kernel Gram matrices yielded an ECE of 0.0075, the lowest observed across all experimental configurations. Gradient-weighted class activation mapping and Shapley additive explanations recovered neurobiologically established signatures of ADHD and psychotic disorder in prefrontal, cingulate, and basal ganglia regions, lending biological credibility to the learned representations. Taken together, these results indicate that quantum kernel methods can confer genuine advantages over classical baselines in small-sample neuroimaging settings, and that the resulting models can be made reliable enough for clinical deployment through standard calibration procedures.
Brain imaging is a critical tool in mental health diagnostics, particularly for detecting anomalies associated with behavioral disorders, anxiety, depression, and neuropsychiatric conditions. Traditional deep learning approaches, such as Convolutional Neural Networks (CNNs), have shown promise in analyzing functional Magnetic Resonance Imaging (fMRI), but they often suffer from computational limitations, feature extraction inefficiencies, and generalization issues. Quantum Machine Learning (QML) offers a transformative solution by enhancing feature representation, improving classification accuracy, and exploiting the power of quantum computation for large-scale medical imaging datasets. In this paper, we explored the potential of quantum models for detecting anomalies in brain imaging. Using the UCLA Consortium for Neuropsychiatric Phenomics dataset, we compared quantum models, including the Variational Quantum Classifier (VQC), Quantum Convolutional Neural Network (QCNN), and Quantum Support Vector Machine (QSVM), with classical models like CNNs and the Support Vector Machine (SVM). Our results revealed that QSVM and VQC outperformed classical models, both achieving 92% accuracy and perfect recall (100%), making them highly effective at identifying positive cases-a critical requirement in clinical diagnostics. QCNN, while less accurate at 74%, still showed strong precision (90%) and an F1score of 85%, highlighting its potential where avoiding false positives is key. Classical models, in comparison, performed moderately well: SVM reached 87% accuracy and CNN achieved 83%, but both had lower precision and F 1 -scores, suggesting less consistent performance. These findings suggest that quantum models, especially VQC and QSVM, offer a promising advantage in neuroimaging-based mental health diagnostics.
The rapid growth of computer vision and increasingly complex image recognition tasks has exposed fundamental computational limitations of classical machine learning models, motivating the exploration of quantum computing as an emerging new paradigm. This paper presents a comprehensive benchmarking study of classical and quantum machine learning models for image recognition on the MNIST handwritten digit dataset, evaluating both traditional models, a Classical Support Vector Machine (CSVM) and a Quantum Support Vector Machine (QSVM), and deep neural network models, a Classical Convolutional Neural Network (CCNN) and a Quantum Convolutional Neural Network (QCNN), across four performance dimensions: classification accuracy, computational runtime, parameter count, and memory requirements. Experiments are conducted as functions of both feature dimensionality and sample size, and across CPU and GPU execution environments, providing a controlled, multidimensional comparison to address gaps in prior work. For the SVM-based models, QSVM consistently outperforms CSVM in accuracy, reaching ∼ 0.90 versus ∼ 0.85 at 1,000 samples, with a higher computational cost. A feature count of 10 qubits and a sample size in the range of 200 -- 500 emerge as practical operating points that balance accuracy and runtime. For the neural network models, CCNN and QCNN achieve comparable classification accuracy, both exceeding 0.96 at 64 features and 60,000 samples, yet QCNN offers superior parameter and memory efficiency at higher feature counts, while incurring higher runtime. Across both model families, quantum models consistently outperform classical models by greater margins in accuracy as feature dimensionality or sample size increases.
Early and accurate identification of autism spectrum disorder (ASD) remains a central challenge in clinical neuroscience due to the subjective nature and high resource demands of traditional diagnostic methods. This paper presents a hybrid classical–quantum learning framework for ASD detection, combining the representational power of quantum circuits with the reliability of classical machine learning. The proposed model, a Variational Quantum Classifier (VQC) implemented using PennyLane and PyTorch, processes structured behavioural and demographic screening data from child, adolescent, and adult cohorts. Features are classically preprocessed, encoded into quantum states through angle embedding, and refined using strongly entangling layers, before being post-processed through a classical dense network. Experimental results show perfect classification performance with 100% accuracy, precision, recall, and F1-score, outperforming comparable classical approaches. The training phase exhibited smooth convergence across 70 epochs, confirming model stability and learning efficiency. Although evaluated in simulation, this study establishes a proof-of-concept for leveraging quantum machine learning in medical diagnostics, laying groundwork for future integration with real quantum hardware in clinical screening applications.
Quantum kernel methods are the most concrete near-term proposal for quantum machine learning on classical data, yet published benchmarks disagree on whether they outperform tuned classical kernels, and the disagreement cannot be resolved from the literature because the comparisons are not protocol-controlled. We argue that the obstacle is one of measurement, not theory: the usual setup fixes the quantum feature map's data scaling while grid-searching the classical kernel's hyperparameters, recomputes nothing per fold, and reports cost in simulator time. We therefore treat the comparison itself as the object of study and impose three controls. The bandwidth is tuned on the same grid budget as the classical γ, the kernel is recomputed inside every fold from fold-fitted preprocessing, and cost is counted in hardware circuit evaluations. On top of these we run a factorial design of 324 cells in which dimensionality, training size, class separability and data structure are explicit covariates, together with four standard datasets, a shots-versus-gate-noise ablation, and a cost sweep. The dominant effect we measure is not quantum: tuning the feature map's bandwidth is worth 0.120 accuracy on synthetic data and 0.339 on real data, exceeding every quantum-classical difference we observe. Tuned quantum kernels are then statistically indistinguishable from a tuned RBF on all four real datasets. The advantage on synthetic data is confined almost entirely to data generated by the feature map itself, and it shrinks with dimensionality (-0.0045 per qubit) while growing with sample size. Finite sampling costs an order of magnitude more accuracy than gate and readout noise, and a fidelity kernel needs 5×107 circuits at n = 10,000. Protocol variance is thus sufficient to explain the field's contradictory conclusions.