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

Discover Quantum Science

Motor insurance data analysis by quantum machine learning

Muhsin Tamturk, Eran Ginossar, Marco Carenzo, Claude Perret

Published Feb 10, 2026• 0 citations
Publisher

Abstract

Abstract In this paper, we analyse motor insurance claim data using a quantum machine learning approach. The objective of this study is to demonstrate how insurance claims can be analysed by leveraging the properties of quantum computing and to show that quantum-based approaches can improve prediction accuracy in motor insurance claim analysis, a task of considerable importance in the highly competitive insurance market. We employ a hybrid quantum-classical algorithm based on quantum reservoir computing (QRC) to improve predictive modelling through resampling of the original insurance data. The algorithm is implemented on IBM’s noise-free Qiskit simulator running on classical hardware, as the method does not require a large number of qubits. To assess its effectiveness, we benchmark the QRC-based approach against established classical machine learning techniques, including linear regression, XGBoost, and CatBoost.

DOI
10.1007/s44464-026-00007-x

Related papers

Leveraging Quantum Machine Learning for Actuarial Predictions in Health Insurance

Jessica Taylor

Health insurance actuarial science has traditionally relied on classical statistical models to predict claims, assess risk, and set premiums. However, the growing complexity and high dimensionality of healthcare datasets challenge conventional techniques. Quantum Machine Learning (QML), integrating quantum computing and artificial intelligence, presents a promising paradigm for accelerating computations and capturing intricate correlations in insurance datasets. This paper develops a hybrid quantum-classical framework for actuarial predictions, encompassing claims forecasting, mortality and morbidity risk modeling, and dynamic premium optimization. Extensive simulation studies demonstrate QML’s ability to improve predictive accuracy, handle high-dimensional data, and enable risk-adjusted pricing strategies. The paper concludes with discussions on operational integration, ethical considerations, and future research directions.

•Dec 30, 2023•0 citations

Bridging Actuarial Science and Quantum Machine Learning: Applications in Health Insurance

Michael Brown

This paper presents a comprehensive framework for integrating Quantum Machine situating the problem: rising healthcare costs and the increasing complexity of health-claims data demand more powerful predictive models than standard generalized linear models (GLMs) alone can reliably deliver. We review classical actuarial techniques and contemporary Learning (QML) into actuarial prediction pipelines for health insurance. We begin by machine learning (ML) practices in healthcare-cost and claims prediction and summarize the current state of quantum computing and QML with special attention to algorithms and methods applicable in the near-term (NISQ) era. We then introduce a hybrid quantum–classical actuarial pipeline (HQCAP) that maps actuarial targets (claim frequency, severity, high-cost utilizers, readmission risk) to QML model families (quantum kernel methods, variational quantum circuits, and quantum-enhanced optimization), together with detailed data-encoding strategies, loss functions and calibration techniques appropriate for insurance. We present an experimental plan (datasets, baselines, metrics) for evaluating HQCAP on representative healthcare cost and claims problems (using MEPS/MIMIC public datasets and synthetic insurance-claims data), and discuss implementation considerations (noise, qubit counts, simulators, software stacks), regulatory and auditability constraints, interpretability and fairness, and an economic analysis of when QML could be preferable to classical approaches. We close with limitations, open research directions, and practical recommendations for actuaries and insurers exploring QML today. Key contributions: (1) a detailed, implementable hybrid QML/actuarial workflow; (2) explicit mapping between actuarial targets and QML primitives; and (3) reproducible experimental protocols for academic and industry evaluation.

•Dec 30, 2023•0 citations

Comparison of Classical ML algorithms and Quantum Machine Learning Algorithms in Identifying Insurance Claim Fraud

Venkateswara Rao P, Sucharita V, Srinivasa Rao D

Fraud detection in insurance system is a significant challenge because of the data imbalance and complex claim patterns. For these kinds of tasks classical machine learning models like support vector Machines have demonstrated very good performance. A promising paradigm that has surface recently for improving learning through quantum features is Quantum Machine Learning. In this research paper for the purpose of home insurance fraud detection a comparative study of the classical SVM with quantum support vector machines is done. A dataset with different features was used to evaluate the accuracy, false positives and recall. Qiskit ZZ Feature map and fidelity-based quantum kernels are used to implement the quantum support vector machines. The experimental results show that the QSVM achieves competitive accuracy but under class imbalance it struggles with fraud recall because of the limitations of quantum kernel methods. This paper gives the practical insights of Quantum Machine learning for the insurance fraud detection. Also, Explainable AI(XAI) technique was incorporated to analyze the fraud detections.

•Mar 28, 2026•0 citations

Quantum Machine Learning Technique for Insurance Claim Fraud Detection with Quantum Feature Selection

Subodh Nath Pushpak

This paper demonstrates a novel use of quantum machine learning (QML) algorithms for detecting fraudulent activities in the home insurance sector. Utilizing actual data and IBM Quantum processors through the Qiskit software stack, the study introduces an innovative method for selecting quantum features that are specifically designed to accommodate the limitations of Near Intermediate Scale Quantum (NISQ) technology by using the Quantum Support Vector Machine (QSVM) in conjunction with traditional machine learning techniques. A comprehensive comparison was conducted to evaluate their effectiveness in detecting fraud. The indicators such as accuracy, recall, and false positive rate are carefully analyzed. Despite the constraints of current quantum technology, QSVM shows excellent accuracy, especially on limited datasets, indicating its potential to enhance insurance fraud detection. The research emphasizes the crucial feature selection in optimizing QML algorithms for fraud detection tasks. It investigates the capacities of hybrid quantum/classical machine learning ensembles. Future research directions include expanding this study to actual hardware implementations to verify its practical feasibility. The work enhances financial security in the insurance business by using quantum computing technology in fraud detection approaches. It establishes the feasibility and efficacy of using quantum resources to solve difficult real-world issues, setting a foundation for or the broader application of QML in fraud detection and other fields.

Journal of Information Systems Engineering & Management•Jan 31, 2025•2 citations

Hardware-Aware Design and Layout Selection in Quantum Reservoir Computing for Insurance Claims Forecasting

Renáta Rusnáková, Eugen Šlapak, Gaurav K. Pandey +1 more

Quantum reservoir computing (QRC) processes temporal information through quantum dynamics, but it remains unclear whether small QRC models can forecast real data competitively with classical models and whether their predictions survive the transfer from ideal simulation to noisy hardware. We address both questions for a six-qubit gate-based reservoir that forecasts next-day motor-insurance claim counts. Five reservoir candidates were selected and frozen using hardware-native interaction graphs, Trotterized Ising dynamics, memory and measurement-robustness criteria, circuit cost, and calibrated-noise screening. On 232 held-out test days, the ideal candidates reached root-mean-square errors (RMSEs) of 5.66-6.12 claims/day, compared with 5.36 for ridge regression. For four of the five, the difference from ridge was not significant under a 95% block bootstrap, and their accuracy was comparable with classical recurrent and reservoir baselines. On the 156-qubit IBM Kingston processor, we ran 63 evaluations in 21 paired rounds, comparing hardware-aware, random-native, and Qiskit optimization level 1 layouts. Hardware-aware placement gave lower RMSE than Qiskit in every round, with a maximum transfer distortion of 7.59 claims/day (11.84 for Qiskit), and avoided the severe transfer failures seen for some random layouts. Most of the hardware distortion appeared as a systematic forecast shift, which a short hardware probe on training inputs detected before the test runs. Longer replay windows raised circuit cost without a consistent gain in forecasting accuracy or hardware transfer. Small QRC models can thus reach competitive accuracy on this task, and hardware-aware layout selection makes their transfer to IBM Kingston more consistent.

quant-pharXiv:2611.03891•Nov 4, 2026•0 citations

An Implementation of Quantum Machine Learning Technique to Determine Insurance Claim Fraud

Subodh Nath Pushpak, Sarika Jain

Quantum Machine Learning (QML) is an upcoming and active research field that uses quantum computing concepts for machine learning. This paper presents QML techniques for detecting fraud in housing insurance claims. Insurance fraud detection is challenging as the patterns may be complex. Therefore, various techniques and methods are employed to identify suspicious claims and prevent business losses due to these fraudulent claims. This paper explores Quantum Support Vector Machine (QSVM) techniques and feature engineering, feature selection, and parameter tweaking to identify fraudulent housing insurance claims. In addition, it compares Quantum SVM with classical SVM. Finally, this paper focuses on detecting property insurance claims fraud by using machine learning techniques.

International Conference on Reliability, Infocom Technologies and Optimization (Trends and Future Directions) (ICRITO)•Oct 13, 2022•13 citations
Abstract In this paper, we analyse motor insurance claim data using a quantum machine learning approach. The objective of this study is to demonstrate how insurance claims can be analysed by leveraging the properties of quantum computing and to show that quantum-based approaches can improve prediction accuracy in motor insurance claim analysis, a task of considerable importance in the highly competitive insurance market. We employ a hybrid quantum-classical algorithm based on quantum reservoir computing (QRC) to improve predictive modelling through resampling of the original insurance data. The algorithm is implemented on IBM’s noise-free Qiskit simulator running on classical hardware, as the method does not require a large number of qubits. To assess its effectiveness, we benchmark the QRC-based approach against established classical machine learning techniques, including linear regression, XGBoost, and CatBoost.
Health insurance actuarial science has traditionally relied on classical statistical models to predict claims, assess risk, and set premiums. However, the growing complexity and high dimensionality of healthcare datasets challenge conventional techniques. Quantum Machine Learning (QML), integrating quantum computing and artificial intelligence, presents a promising paradigm for accelerating computations and capturing intricate correlations in insurance datasets. This paper develops a hybrid quantum-classical framework for actuarial predictions, encompassing claims forecasting, mortality and morbidity risk modeling, and dynamic premium optimization. Extensive simulation studies demonstrate QML’s ability to improve predictive accuracy, handle high-dimensional data, and enable risk-adjusted pricing strategies. The paper concludes with discussions on operational integration, ethical considerations, and future research directions.
This paper presents a comprehensive framework for integrating Quantum Machine situating the problem: rising healthcare costs and the increasing complexity of health-claims data demand more powerful predictive models than standard generalized linear models (GLMs) alone can reliably deliver. We review classical actuarial techniques and contemporary Learning (QML) into actuarial prediction pipelines for health insurance. We begin by machine learning (ML) practices in healthcare-cost and claims prediction and summarize the current state of quantum computing and QML with special attention to algorithms and methods applicable in the near-term (NISQ) era. We then introduce a hybrid quantum–classical actuarial pipeline (HQCAP) that maps actuarial targets (claim frequency, severity, high-cost utilizers, readmission risk) to QML model families (quantum kernel methods, variational quantum circuits, and quantum-enhanced optimization), together with detailed data-encoding strategies, loss functions and calibration techniques appropriate for insurance. We present an experimental plan (datasets, baselines, metrics) for evaluating HQCAP on representative healthcare cost and claims problems (using MEPS/MIMIC public datasets and synthetic insurance-claims data), and discuss implementation considerations (noise, qubit counts, simulators, software stacks), regulatory and auditability constraints, interpretability and fairness, and an economic analysis of when QML could be preferable to classical approaches. We close with limitations, open research directions, and practical recommendations for actuaries and insurers exploring QML today. Key contributions: (1) a detailed, implementable hybrid QML/actuarial workflow; (2) explicit mapping between actuarial targets and QML primitives; and (3) reproducible experimental protocols for academic and industry evaluation.
Fraud detection in insurance system is a significant challenge because of the data imbalance and complex claim patterns. For these kinds of tasks classical machine learning models like support vector Machines have demonstrated very good performance. A promising paradigm that has surface recently for improving learning through quantum features is Quantum Machine Learning. In this research paper for the purpose of home insurance fraud detection a comparative study of the classical SVM with quantum support vector machines is done. A dataset with different features was used to evaluate the accuracy, false positives and recall. Qiskit ZZ Feature map and fidelity-based quantum kernels are used to implement the quantum support vector machines. The experimental results show that the QSVM achieves competitive accuracy but under class imbalance it struggles with fraud recall because of the limitations of quantum kernel methods. This paper gives the practical insights of Quantum Machine learning for the insurance fraud detection. Also, Explainable AI(XAI) technique was incorporated to analyze the fraud detections.
This paper demonstrates a novel use of quantum machine learning (QML) algorithms for detecting fraudulent activities in the home insurance sector. Utilizing actual data and IBM Quantum processors through the Qiskit software stack, the study introduces an innovative method for selecting quantum features that are specifically designed to accommodate the limitations of Near Intermediate Scale Quantum (NISQ) technology by using the Quantum Support Vector Machine (QSVM) in conjunction with traditional machine learning techniques. A comprehensive comparison was conducted to evaluate their effectiveness in detecting fraud. The indicators such as accuracy, recall, and false positive rate are carefully analyzed. Despite the constraints of current quantum technology, QSVM shows excellent accuracy, especially on limited datasets, indicating its potential to enhance insurance fraud detection. The research emphasizes the crucial feature selection in optimizing QML algorithms for fraud detection tasks. It investigates the capacities of hybrid quantum/classical machine learning ensembles. Future research directions include expanding this study to actual hardware implementations to verify its practical feasibility. The work enhances financial security in the insurance business by using quantum computing technology in fraud detection approaches. It establishes the feasibility and efficacy of using quantum resources to solve difficult real-world issues, setting a foundation for or the broader application of QML in fraud detection and other fields.
Quantum reservoir computing (QRC) processes temporal information through quantum dynamics, but it remains unclear whether small QRC models can forecast real data competitively with classical models and whether their predictions survive the transfer from ideal simulation to noisy hardware. We address both questions for a six-qubit gate-based reservoir that forecasts next-day motor-insurance claim counts. Five reservoir candidates were selected and frozen using hardware-native interaction graphs, Trotterized Ising dynamics, memory and measurement-robustness criteria, circuit cost, and calibrated-noise screening. On 232 held-out test days, the ideal candidates reached root-mean-square errors (RMSEs) of 5.66-6.12 claims/day, compared with 5.36 for ridge regression. For four of the five, the difference from ridge was not significant under a 95% block bootstrap, and their accuracy was comparable with classical recurrent and reservoir baselines. On the 156-qubit IBM Kingston processor, we ran 63 evaluations in 21 paired rounds, comparing hardware-aware, random-native, and Qiskit optimization level 1 layouts. Hardware-aware placement gave lower RMSE than Qiskit in every round, with a maximum transfer distortion of 7.59 claims/day (11.84 for Qiskit), and avoided the severe transfer failures seen for some random layouts. Most of the hardware distortion appeared as a systematic forecast shift, which a short hardware probe on training inputs detected before the test runs. Longer replay windows raised circuit cost without a consistent gain in forecasting accuracy or hardware transfer. Small QRC models can thus reach competitive accuracy on this task, and hardware-aware layout selection makes their transfer to IBM Kingston more consistent.
Quantum Machine Learning (QML) is an upcoming and active research field that uses quantum computing concepts for machine learning. This paper presents QML techniques for detecting fraud in housing insurance claims. Insurance fraud detection is challenging as the patterns may be complex. Therefore, various techniques and methods are employed to identify suspicious claims and prevent business losses due to these fraudulent claims. This paper explores Quantum Support Vector Machine (QSVM) techniques and feature engineering, feature selection, and parameter tweaking to identify fraudulent housing insurance claims. In addition, it compares Quantum SVM with classical SVM. Finally, this paper focuses on detecting property insurance claims fraud by using machine learning techniques.