Challenges
Datasets
Workspaces
Papers
Discussions
Leaderboard
Log inSign up
Challenges
Datasets
Workspaces
Papers
Discussions
Leaderboard
Blog
Job Board
Q3AS

© 2026 Aqora Quantum S.A.S.

TermsPrivacyLegal Notice

International Conference on Automation, Computing and Renewable Systems (ICACRS)

Explainable Hybrid Quantum–Classical Models for Student and Employee Dropout Prediction: A Comparative Analysis

Rajesh Malick, Murugasamy P, Archana Jenis M R

Published Dec 10, 2025• 1 citation
Publisher

Abstract

High dropout and attrition rates among students and employees present critical challenges for educational institutions and corporations, with significant financial and social implications. Traditional machine learning approaches often struggle with imbalanced, high-dimensional data and frequently lack the interpretability necessary for actionable decision-making. This paper presents a comprehensive empirical comparison of hybrid quantum–classical machine learning models against state-of-the-art classical methods for dropout prediction across two distinct real-world domains: academic student performance and corporate employee attrition. We implement a Variational Quantum Classifier (VQC) in PennyLane and benchmark it against Random Forest, XGBoost, and BiLSTM-CNN using identical preprocessing protocols on academic (480 samples) and corporate (1,470 samples) datasets. Results demonstrate problem-dependent quantum advantage: BiLSTM-CNN achieves highest performance on student data (F1: 0.81 vs. VQC’s 0.75), while the tuned VQC outperforms classical methods on employee data (F1: 0.51 vs. BiLSTM-CNN’s 0.48). Integration of SHAP-based explainability enables transparent feature attribution and personalized intervention generation. This work contributes the first systematic empirical benchmark of quantum machine learning for dropout prediction and demonstrates that quantum advantage in the NISQ era is highly problem-dependent rather than universal.

DOI
10.1109/icacrs67045.2025.11324437

Related papers

Comparative Analysis of Classical Machine Learning and Quantum Machine Learning on the NASA C-MAPSS Turbofan Engine Dataset: A Comprehensive Experimental Evaluation Focused on Fault Detection (FD001–FD004)

Sevdanur Genç, ŞİRİN GÜLTEN ÇELİKASLAN

This study systematically compares classical machine learning (ML) and quantum machine learning (QML) methods in the context of fault detection across all subsets of the NASA Commercial Modular Aero-Propulsion System Simulation (C-MAPSS) dataset (FD001–FD004). Within a binary classification framework (RUL ≤ 30 cycles threshold), classical models such as Random Forest (RF), SMOTE+RF, Gradient Boosting (GBM), XGBoost, and LSTM are compared against quantum/hybrid models such as the Variational Quantum Classifier (VQC), Quantum Support Vector Machine (QSVM), Quantum LSTM (QLSTM), and Hybrid Quantum-Classical CNN (HQCNN). Experiments conducted using 4 qubits, 3 circuit layers, and a Noisy Intermediate-Scale Quantum (NISQ) simulator (PennyLane default.qubit) reveal that classical ML models—particularly XGBoost (F1 ≥ 0.940, AUC ≥ 0.990)—substantially outperform QML models in the current NISQ era (QLSTM: F1 ≈ 0.903, HQCNN: F1 ≈ 0.877). The ablation study demonstrates that this gap is attributable primarily to model architecture rather than to data volume. The study experimentally characterizes the current limitations of NISQ-era QML applications and offers practical guidance for the design of future hybrid pipelines.

International Conference on Computer Science and Engineering (UBMK)•Sep 1, 2026•0 citations

Comparative Evaluation of Quantum-Classical Hybrid Machine Learning Architectures in Healthcare: A Case Study on Heart Disease Prediction

Vikas Garg, Tapsi Nagpal, Svav Prasad

Abstract Heart disease remains to be a primary cause of death globally, requiring precise and timely predictive methodologies. Traditional machine learning methods that depend exclusively on classical clinical variables frequently have restricted predictive accuracy and do not adequately use the intricate nonlinear interactions present in patient data. This work examines the effectiveness of hybrid quantum-classical models for predicting heart disease by combining classical clinical data with quantum-derived representations via Variational Quantum Classifier (VQC), Estimator, and Sampler models. Twelve hybrid configurations, integrating three quantum feature maps (Z, ZZ, Pauli) with four ansatz architectures (RealAmplitudes, TwoLocal, EfficientSU2, PauliTwoDesign), were assessed inside the PennyLane and Qiskit frameworks. The results indicate that the VQC model in PennyLane reached optimal performance with Z–TwoLocal and ZZ–PauliTwoDesign configurations, achieving accuracy and F1-scores of up to 0.95. The Estimator model consistently produced reliable predictions, with Pauli–EfficientSU2 and Z–TwoLocal attaining accuracies of up to 0.91 and F1-scores of 0.92 across several frameworks. The Sampler model demonstrated 1reasonable performance, with the Pauli–EfficientSU2 configuration achieving an accuracy of 0.65 and an F1-score of 0.73, underscoring the supplementary role of stochastic quantum sampling. The analysis of feature significance indicated that the quantum-derived feature provided the most robust predictive signal, exceeding classical variables like Age, Blood Pressure, and Cholesterol. These findings demonstrate that meticulous alignment of feature maps and ansatz circuits is essential for optimizing model expressivity, convergence, and predictive effectiveness. In summary, hybrid quantum-classical models, especially the Estimator and VQC architectures, significantly improve heart disease prediction, providing a feasible method for precision clinical decision support and potential incorporation into privacy-preserving, multi-institutional healthcare systems.

•May 31, 2026•0 citations

Heart Disease Prediction Using Hybrid Quantum-Classical Machine Learning Models

Pavan Reddy T, D Vishwachetan, S. Rajarajeswari

Cardiovascular diseases remain the leading cause of mortality globally, necessitating accurate diagnostic tools for early detection and intervention. This paper presents a comprehensive comparative and hybrid analysis of quantum and classical machine learning models for heart disease prediction using the UCI Cleveland dataset. We implement classical baseline models including Logistic Regression, Random Forest, Support Vector Machine (SVM), and XGBoost, alongside quantum models: Quantum Support Vector Machine (QSVM) with statevector fidelity kernels, Variational Quantum Classifier (VQC) with data-reuploading ansätze, and Quantum Convolutional Neural Networks (QCNN) with parametric pooling. The primary contribution is the development of a hybrid quantum classical ensemble meta-classifier that fuses predictions from all classical and quantum learners using XGBoost as a meta learner. Experiments conducted in Google Colab using Qiskit 1.2+ and Penny Lane 0.36+ demonstrate that the proposed hybrid ensemble achieves superior performance with 98.33accuracy, 100perfect AUC-ROC of 1.0, substantially outperforming standalone classical models (best: 88.33(best: 80hybrid model’s superiority. These results underscore the potential of quantum feature spaces combined with classical decision boundaries in healthcare prediction systems, and demonstrate the immediate utility of hybrid quantum-classical integration in medical diagnostics on near-term quantum hardware.

International Conference on Knowledge Engineering and Communication Systems (ICKECS)•Apr 24, 2026•0 citations

A Fair Comparison of Quantum and Classical Machine Learning for Predictive Maintenance on the AI4I 2020 Dataset

Sevdanur Genç, Şirin Gϋlten ÇElİKaslan

This study presents a comparative analysis of classical machine learning (ML) and quantum machine learning (QML) algorithms for industrial equipment fault detection using the AI4I 2020 Predictive Maintenance dataset. A total of nine models were evaluated through 5-fold stratified cross-validation: three classical models - Logistic Regression, Random Forest, and XGBoost - alongside six quantum models, namely Quantum Support Vector Machine (QSVM), Quantum K-Nearest Neighbors (QKNN), Quantum Kernel-based hybrid classifiers (QK+RF, QK+GB), Pegasos Quantum Support Vector Classifier (PegasosQSVC), and Variational Quantum Classifier (VQC). All quantum experiments were conducted using the IBM Qiskit framework and AerSimulator backend. Under a controlled comparison using an identical 300-sample budget and the same 5-fold cross-validation protocol, the best classical model (Random Forest, F1=0.902) outperformed the best quantum kernel method (QKNN, F1=0.820) by approximately 8.2 points. These results indicate that quantum kernel methods do not provide a dataefficiency advantage over classical methods on this dataset, although they achieve reasonably competitive results (F1 $\geq 0.75$ for kernel-based methods) at substantially higher computational cost (quantum: ~124-137 s/fold for kernel methods, up to ~257 s/fold for VQC, vs. classical: 0.01-0.26 s/fold). We analyze the sources of this gap and attribute it primarily to a mismatch between the unitarily-invariant geometry of the fidelity kernel and the axisaligned threshold structure of the AI4I failure modes, rather than to information loss in dimensionality reduction, which the faircomparison design rules out. We further show that this apparent quantum competitiveness stems partly from an unequal-data-volume comparison common in prior QML benchmarking studies, which our fair-comparison protocol corrects.

International Artificial Intelligence and Data Processing Symposium (IDAP)•Sep 1, 2026•0 citations

Towards Explainable Quantum Machine Learning: A Hybrid Quantum SHAP-based Framework

M. Ramu, Donapati Deepika, Nikhila Baddula +3 more

Quantum machine learning is becoming popular as a method to enhance computational intelligence using concepts of quantum mechanics. However, quantum models tend to be complex and probabilistic, making them difficult to interpret and limiting their application in real-world scenarios where transparent decision-making is important. This work introduces a quantum-aware feature attribution framework that combines hybrid quantum-classical models with Quantum SHAP-based explainability. The proposed methodology enables prediction interpretation by analyzing the sensitivity of classical input features within quantum circuits. Experiments conducted on benchmark datasets demonstrate that the proposed method achieves competitive predictive performance while improving the stability and consistency of explanations. The framework also provides better understanding of feature contributions even in the presence of quantum noise. Overall, the findings indicate that explainability methods can improve transparency and trust in quantum machine learning models. This work contributes toward the development of more reliable and interpretable quantum-based learning systems by bridging quantum computing and explainable artificial intelligence.

International Conference on Computing Methodologies and Communication (ICCMC)•Jul 1, 2026•0 citations

Hybrid Quantum-Classical Machine Learning For Scalable AI Applications

Manish Patil, Shraddha Sanjay Mhetre

Background: The emergent need for scalable artificial intelligence (AI) models emphasizes the shortcomings of traditional machine learning in dealing with big, complicated datasets owing to computation and energy limitations. Objective: The architecture takes advantage of quantum encoding methods like amplitude and angle encoding for compact feature representation, with classical preprocessing providing compatibility and stability. Methods: To overcome these issues, this research introduces a Hybrid Quantum-Classical Machine Learning (HQML) framework that combines Quantum Variational Classifiers (QVCs) and Quantum Convolutional Neural Networks (QCNNs) with classical deep learning layers. Results: Experiments were performed on pre-curated datasets taken from the Scikit Machine Learning Repository, with hybrid models compared to classical Convolutional Neural Networks (CNNs). The implementation was done with PennyLane using TensorFlow Quantum for compatibility with noisy intermediate-scale quantum (NISQ) machines. The comparative analysis identified that the suggested HQML model achieves significant performance gains, as classification accuracy increased from 73 percent in the baseline classical CNN to 98% in the hybrid model. Precision also went up from 77 to 96 percent, recall from 81 to 100 percent, and F1-score from 79 to 98 percent. These findings confirm the ability of the model to leverage quantum benefits for extracting complicated data distributions without losing stability through classical components. The findings confirm the viability of HQML to real-world finance, healthcare, and intelligent automation applications where scalability and precision matter. Conclusion: This research establishes that quantum-classical integration not only minimizes training overhead but also attains near-perfect classification efficiency with constrained qubit resources. The research aids in filling the gap between theoretical quantum potential and deployable practical AI, opening the door for next-generation quantum-boosted intelligence.

International Journal of Advances in Signal and Image Sciences•Sep 25, 2025•0 citations
High dropout and attrition rates among students and employees present critical challenges for educational institutions and corporations, with significant financial and social implications. Traditional machine learning approaches often struggle with imbalanced, high-dimensional data and frequently lack the interpretability necessary for actionable decision-making. This paper presents a comprehensive empirical comparison of hybrid quantum–classical machine learning models against state-of-the-art classical methods for dropout prediction across two distinct real-world domains: academic student performance and corporate employee attrition. We implement a Variational Quantum Classifier (VQC) in PennyLane and benchmark it against Random Forest, XGBoost, and BiLSTM-CNN using identical preprocessing protocols on academic (480 samples) and corporate (1,470 samples) datasets. Results demonstrate problem-dependent quantum advantage: BiLSTM-CNN achieves highest performance on student data (F1: 0.81 vs. VQC’s 0.75), while the tuned VQC outperforms classical methods on employee data (F1: 0.51 vs. BiLSTM-CNN’s 0.48). Integration of SHAP-based explainability enables transparent feature attribution and personalized intervention generation. This work contributes the first systematic empirical benchmark of quantum machine learning for dropout prediction and demonstrates that quantum advantage in the NISQ era is highly problem-dependent rather than universal.
This study systematically compares classical machine learning (ML) and quantum machine learning (QML) methods in the context of fault detection across all subsets of the NASA Commercial Modular Aero-Propulsion System Simulation (C-MAPSS) dataset (FD001–FD004). Within a binary classification framework (RUL ≤ 30 cycles threshold), classical models such as Random Forest (RF), SMOTE+RF, Gradient Boosting (GBM), XGBoost, and LSTM are compared against quantum/hybrid models such as the Variational Quantum Classifier (VQC), Quantum Support Vector Machine (QSVM), Quantum LSTM (QLSTM), and Hybrid Quantum-Classical CNN (HQCNN). Experiments conducted using 4 qubits, 3 circuit layers, and a Noisy Intermediate-Scale Quantum (NISQ) simulator (PennyLane default.qubit) reveal that classical ML models—particularly XGBoost (F1 ≥ 0.940, AUC ≥ 0.990)—substantially outperform QML models in the current NISQ era (QLSTM: F1 ≈ 0.903, HQCNN: F1 ≈ 0.877). The ablation study demonstrates that this gap is attributable primarily to model architecture rather than to data volume. The study experimentally characterizes the current limitations of NISQ-era QML applications and offers practical guidance for the design of future hybrid pipelines.
Abstract Heart disease remains to be a primary cause of death globally, requiring precise and timely predictive methodologies. Traditional machine learning methods that depend exclusively on classical clinical variables frequently have restricted predictive accuracy and do not adequately use the intricate nonlinear interactions present in patient data. This work examines the effectiveness of hybrid quantum-classical models for predicting heart disease by combining classical clinical data with quantum-derived representations via Variational Quantum Classifier (VQC), Estimator, and Sampler models. Twelve hybrid configurations, integrating three quantum feature maps (Z, ZZ, Pauli) with four ansatz architectures (RealAmplitudes, TwoLocal, EfficientSU2, PauliTwoDesign), were assessed inside the PennyLane and Qiskit frameworks. The results indicate that the VQC model in PennyLane reached optimal performance with Z–TwoLocal and ZZ–PauliTwoDesign configurations, achieving accuracy and F1-scores of up to 0.95. The Estimator model consistently produced reliable predictions, with Pauli–EfficientSU2 and Z–TwoLocal attaining accuracies of up to 0.91 and F1-scores of 0.92 across several frameworks. The Sampler model demonstrated 1reasonable performance, with the Pauli–EfficientSU2 configuration achieving an accuracy of 0.65 and an F1-score of 0.73, underscoring the supplementary role of stochastic quantum sampling. The analysis of feature significance indicated that the quantum-derived feature provided the most robust predictive signal, exceeding classical variables like Age, Blood Pressure, and Cholesterol. These findings demonstrate that meticulous alignment of feature maps and ansatz circuits is essential for optimizing model expressivity, convergence, and predictive effectiveness. In summary, hybrid quantum-classical models, especially the Estimator and VQC architectures, significantly improve heart disease prediction, providing a feasible method for precision clinical decision support and potential incorporation into privacy-preserving, multi-institutional healthcare systems.
Cardiovascular diseases remain the leading cause of mortality globally, necessitating accurate diagnostic tools for early detection and intervention. This paper presents a comprehensive comparative and hybrid analysis of quantum and classical machine learning models for heart disease prediction using the UCI Cleveland dataset. We implement classical baseline models including Logistic Regression, Random Forest, Support Vector Machine (SVM), and XGBoost, alongside quantum models: Quantum Support Vector Machine (QSVM) with statevector fidelity kernels, Variational Quantum Classifier (VQC) with data-reuploading ansätze, and Quantum Convolutional Neural Networks (QCNN) with parametric pooling. The primary contribution is the development of a hybrid quantum classical ensemble meta-classifier that fuses predictions from all classical and quantum learners using XGBoost as a meta learner. Experiments conducted in Google Colab using Qiskit 1.2+ and Penny Lane 0.36+ demonstrate that the proposed hybrid ensemble achieves superior performance with 98.33accuracy, 100perfect AUC-ROC of 1.0, substantially outperforming standalone classical models (best: 88.33(best: 80hybrid model’s superiority. These results underscore the potential of quantum feature spaces combined with classical decision boundaries in healthcare prediction systems, and demonstrate the immediate utility of hybrid quantum-classical integration in medical diagnostics on near-term quantum hardware.
This study presents a comparative analysis of classical machine learning (ML) and quantum machine learning (QML) algorithms for industrial equipment fault detection using the AI4I 2020 Predictive Maintenance dataset. A total of nine models were evaluated through 5-fold stratified cross-validation: three classical models - Logistic Regression, Random Forest, and XGBoost - alongside six quantum models, namely Quantum Support Vector Machine (QSVM), Quantum K-Nearest Neighbors (QKNN), Quantum Kernel-based hybrid classifiers (QK+RF, QK+GB), Pegasos Quantum Support Vector Classifier (PegasosQSVC), and Variational Quantum Classifier (VQC). All quantum experiments were conducted using the IBM Qiskit framework and AerSimulator backend. Under a controlled comparison using an identical 300-sample budget and the same 5-fold cross-validation protocol, the best classical model (Random Forest, F1=0.902) outperformed the best quantum kernel method (QKNN, F1=0.820) by approximately 8.2 points. These results indicate that quantum kernel methods do not provide a dataefficiency advantage over classical methods on this dataset, although they achieve reasonably competitive results (F1 ≥0.75\geq 0.75≥0.75 for kernel-based methods) at substantially higher computational cost (quantum: ~124-137 s/fold for kernel methods, up to ~257 s/fold for VQC, vs. classical: 0.01-0.26 s/fold). We analyze the sources of this gap and attribute it primarily to a mismatch between the unitarily-invariant geometry of the fidelity kernel and the axisaligned threshold structure of the AI4I failure modes, rather than to information loss in dimensionality reduction, which the faircomparison design rules out. We further show that this apparent quantum competitiveness stems partly from an unequal-data-volume comparison common in prior QML benchmarking studies, which our fair-comparison protocol corrects.
Quantum machine learning is becoming popular as a method to enhance computational intelligence using concepts of quantum mechanics. However, quantum models tend to be complex and probabilistic, making them difficult to interpret and limiting their application in real-world scenarios where transparent decision-making is important. This work introduces a quantum-aware feature attribution framework that combines hybrid quantum-classical models with Quantum SHAP-based explainability. The proposed methodology enables prediction interpretation by analyzing the sensitivity of classical input features within quantum circuits. Experiments conducted on benchmark datasets demonstrate that the proposed method achieves competitive predictive performance while improving the stability and consistency of explanations. The framework also provides better understanding of feature contributions even in the presence of quantum noise. Overall, the findings indicate that explainability methods can improve transparency and trust in quantum machine learning models. This work contributes toward the development of more reliable and interpretable quantum-based learning systems by bridging quantum computing and explainable artificial intelligence.
Background: The emergent need for scalable artificial intelligence (AI) models emphasizes the shortcomings of traditional machine learning in dealing with big, complicated datasets owing to computation and energy limitations. Objective: The architecture takes advantage of quantum encoding methods like amplitude and angle encoding for compact feature representation, with classical preprocessing providing compatibility and stability. Methods: To overcome these issues, this research introduces a Hybrid Quantum-Classical Machine Learning (HQML) framework that combines Quantum Variational Classifiers (QVCs) and Quantum Convolutional Neural Networks (QCNNs) with classical deep learning layers. Results: Experiments were performed on pre-curated datasets taken from the Scikit Machine Learning Repository, with hybrid models compared to classical Convolutional Neural Networks (CNNs). The implementation was done with PennyLane using TensorFlow Quantum for compatibility with noisy intermediate-scale quantum (NISQ) machines. The comparative analysis identified that the suggested HQML model achieves significant performance gains, as classification accuracy increased from 73 percent in the baseline classical CNN to 98% in the hybrid model. Precision also went up from 77 to 96 percent, recall from 81 to 100 percent, and F1-score from 79 to 98 percent. These findings confirm the ability of the model to leverage quantum benefits for extracting complicated data distributions without losing stability through classical components. The findings confirm the viability of HQML to real-world finance, healthcare, and intelligent automation applications where scalability and precision matter. Conclusion: This research establishes that quantum-classical integration not only minimizes training overhead but also attains near-perfect classification efficiency with constrained qubit resources. The research aids in filling the gap between theoretical quantum potential and deployable practical AI, opening the door for next-generation quantum-boosted intelligence.