Quantum Machine Learning-Based Prediction of 10-Year Survival in Differentiated Thyroid Cancer

Abstract

Background: Studies have shown the gradual application of quantum machine learning (QML) in the medical field, but there is a lack of relevant applications in the study of cancer prognosis prediction. The aim of this study was to construct a QML model for predicting the 10-year survival rate of patients with differentiated thyroid cancer (DTC).

Methods: In this study, data were gathered from the SEER database encompassing patients with DTC between 2004 and 2007. We built a Quantum Support Vector Classifier (QSVC) to forecast the 10-year survival rate of patients, evaluating its efficacy against ten commonly used machine learning (ML) methods. The performance of QSVC, alongside other ten ML methods, is evaluated using metrics such as accuracy, precision, recall, F1 index, and AUROC. Besides that, an in-depth analysis and exploration are carried out using SHAP.

Results: A total of 27,027 eligible patients with a mean age of 48.9 years and a median age of 48 years were included. We constructed a quantum circuit containing 16 quantum bits. Compared with other models, the highest AUROC in the QML was 0.7903.

Conclusion: We successfully constructed QSVC for predicting the 10-year survival rate of DTC patients by taking advantage of quantum properties, and the models performed well, which provides great advantages in the case of complex data set features and large data volume. Technical support is provided for other predictive model construction.

Note:

Funding Information: This research received no specific grant from any funding agency in public, commercial or not-for-profit sectors.

Conflict of Interests: The authors have no conflicts of interest to declare.