A quantum machine learning model has been developed to predict the risk of death in patients with early-onset colorectal cancer, achieving an accuracy rate of 90%.
Led by Professor Park Yoo Rang from the Department of Medical Life System Information at Yonsei University Medical School, the team developed this model using clinical data from patients with early-onset colorectal cancer. They revealed it on the 12th, highlighting the model’s high predictive accuracy.
Early-onset colorectal cancer, affecting individuals under 50, has a significantly higher incidence in South Korea, especially among those in their 20s and 40s, with a rate of 12.9 per 100,000 population — the highest in the world. Early-onset colorectal cancer is more aggressive and has a lower survival rate compared to colorectal cancer diagnosed in other age groups, underscoring the importance of early detection and accurate prognosis.
The research team utilized data from 1,253 early-onset colorectal cancer patients who visited Severance Hospital from 2008 to 2020 to develop a quantum machine learning model. This model considers 93 variables, including patient information data, clinical data related to the stage of the disease, and treatment information.
The research team evaluated the model’s effectiveness by comparing and analyzing the accuracy based on the optimal number of variables, sample size, and proportion of outcome variables against an existing machine learning model. The Receiver Operating Characteristic Curve (AUROC) index was used to analyze prediction accuracy. The AUROC, the area under the ROC curve, measures the prediction accuracy of a test tool for a specific prognosis. It is commonly used to assess the performance of AI models. Generally, an AUROC close to 1 indicates excellent performance, and a value above 0.8 indicates a high-performance model.
As a result of the analysis, the prediction accuracy of the conventional machine learning model (Conventional SVM) was 70%. In comparison, the quantum machine learning model recorded a 90% prediction accuracy for the risk of death in patients with early-onset colorectal cancer.
To verify the robustness of quantum computing, the research team conducted performance verification by adjusting the ratio of death to survival. As a result, the conventional machine learning model showed a prediction performance of 80% when the death rate was adjusted unevenly.
Furthermore, the team tested the robustness of the quantum machine learning model by adjusting the rai.
On the other hand, the quantum machine learning model maintained a high prediction accuracy of 88% even in situations where the death rate was uneven. It was confirmed that the quantum machine learning model maintained a higher prediction accuracy than the conventional machine learning model even when the ratio of death to survival was uneven.
Professor Park said, “Through this research, we have built a quantum machine learning model that can accurately predict the risk of death in patients with early-onset colorectal cancer. This model can potentially expand to various healthcare areas using the quantum machine learning model in the future.”
“This research exemplifies the integration of digital healthcare with quantum computing and medical artificial intelligence in oncology. The adoption of digital healthcare technology in cancer diagnosis, treatment, and management revolutionize the field of cancer treatment in the future,” Professor Kim explained.
This study also involved Dr. Yoo Jae Yong, Researcher Shim Woo Sub, and Professor Kim Han Sang of Yonsei Cancer Hospital’s Department of Oncology.
