Hallym University Medical Center, medical data utilization competition fee...AI-based anti-cancer side effects prediction model won the grand prize
Oct 03, 2025
On the 17th, Hallim University Medical Center (Director Kim Yong-sun) held an award ceremony for the '2025 Medical Data Exploration & Analysis' at Ilsong Culture Hall on the 5th floor of Hallim University Sacred Heart Hospital.
As part of the medical data-oriented hospital support project, this competition was designed to strengthen creative problem-solving capabilities using medical big data and increase the possibility of clinical use of analysis results. Participants performed various analytical tasks based on anonymized clinical data provided by the Hallym University Medical Center's Next Generation Data Lake Cloud Platform 'Hero'.
Experts, including professors from Hallym University's School of Information Science and clinical professors from Hallym University Medical Center, participated as judges for the evaluation of expertise in each field, and one team, two teams with the best prize, and four teams with the best prize were selected through the review of the results report.
The grand prize was awarded to the 'Hanlim Big Data 2 Team' (nurses Hwang Hye-jin, Kim Yoo-jin, and Kim Ji-sun) who submitted the 'Development of a high-risk group prediction model for neutropenia reduction of patients receiving anticancer drugs' project.
The team devised an algorithm to predict the risk of Neutropenia in chemotherapy patients early based on multimodal data such as blood tests and anticancer drug administration history. Using multimodal (multimodal) data such as blood test results and history of anticancer drug administration, an AI model was conceived to warn early about the risk of neutrophil reduction in chemotherapy patients.
The Hanlim Big Data Team 2 has used the medical data of cancer patients at Hallym University Medical Center over the past five years to analyze the data of about 4,000 chemotherapy patients and implement a predictive model using Python, a computer programming language. As a result, this predictive model recorded a high value of 0.93 on ROC-AUC, a score that indicates the degree to which the disease is accurately hit.
Head nurse Hwang Hye-jin, who oversaw the task, said "We identified the possibility of substantially solving problems in the clinical field through medical data analysis."I hope the predictive model will be the basis for patient-specific treatment and the design of decision-making support systems" he said.
In addition, not only disease prediction models, but also various task studies that can proactively identify and respond to operational problems arising in real-world clinical settings, such as 'data-driven ER bottleneck segment prediction models' or 'pre-nursing workload time series prediction models'. These models have shown the potential to contribute to improving the overall quality of medical services, such as reducing patient waiting times, optimizing nursing staffing, and improving hospital operation efficiency.
Seo Young-kyun, director of the Big Data Center (Professor of Family Medicine at Hallym University Sacred Heart Hospital), said "There were many works that could be published in thesis or applied in clinical practice through this competition."We will not only continue to conduct future competitions, but also continue to manage papers and clinical applications, while expanding the scope of competitions in terms of data and participants to further develop a data-driven innovation ecosystem centered on HERO."
Hallym University Medical Center has been operating a big data center in the Doheon Digital Medical Innovation Research Institute (DIDIM) since 2021, and has developed and built a data lake cloud platform 'HERO'. In addition, it is in charge of various national projects such as medical data-oriented hospital support projects and K-CURE clinical data network construction, and is taking the lead in advancing the medical data utilization system by simultaneously obtaining 'Medical Data Content Certification' and 'Management System Certification' for the first time in a medical institution.
As part of the medical data-oriented hospital support project, this competition was designed to strengthen creative problem-solving capabilities using medical big data and increase the possibility of clinical use of analysis results. Participants performed various analytical tasks based on anonymized clinical data provided by the Hallym University Medical Center's Next Generation Data Lake Cloud Platform 'Hero'.
Experts, including professors from Hallym University's School of Information Science and clinical professors from Hallym University Medical Center, participated as judges for the evaluation of expertise in each field, and one team, two teams with the best prize, and four teams with the best prize were selected through the review of the results report.
The grand prize was awarded to the 'Hanlim Big Data 2 Team' (nurses Hwang Hye-jin, Kim Yoo-jin, and Kim Ji-sun) who submitted the 'Development of a high-risk group prediction model for neutropenia reduction of patients receiving anticancer drugs' project.
The team devised an algorithm to predict the risk of Neutropenia in chemotherapy patients early based on multimodal data such as blood tests and anticancer drug administration history. Using multimodal (multimodal) data such as blood test results and history of anticancer drug administration, an AI model was conceived to warn early about the risk of neutrophil reduction in chemotherapy patients.
The Hanlim Big Data Team 2 has used the medical data of cancer patients at Hallym University Medical Center over the past five years to analyze the data of about 4,000 chemotherapy patients and implement a predictive model using Python, a computer programming language. As a result, this predictive model recorded a high value of 0.93 on ROC-AUC, a score that indicates the degree to which the disease is accurately hit.
Head nurse Hwang Hye-jin, who oversaw the task, said "We identified the possibility of substantially solving problems in the clinical field through medical data analysis."I hope the predictive model will be the basis for patient-specific treatment and the design of decision-making support systems" he said.
In addition, not only disease prediction models, but also various task studies that can proactively identify and respond to operational problems arising in real-world clinical settings, such as 'data-driven ER bottleneck segment prediction models' or 'pre-nursing workload time series prediction models'. These models have shown the potential to contribute to improving the overall quality of medical services, such as reducing patient waiting times, optimizing nursing staffing, and improving hospital operation efficiency.
Seo Young-kyun, director of the Big Data Center (Professor of Family Medicine at Hallym University Sacred Heart Hospital), said "There were many works that could be published in thesis or applied in clinical practice through this competition."We will not only continue to conduct future competitions, but also continue to manage papers and clinical applications, while expanding the scope of competitions in terms of data and participants to further develop a data-driven innovation ecosystem centered on HERO."
Hallym University Medical Center has been operating a big data center in the Doheon Digital Medical Innovation Research Institute (DIDIM) since 2021, and has developed and built a data lake cloud platform 'HERO'. In addition, it is in charge of various national projects such as medical data-oriented hospital support projects and K-CURE clinical data network construction, and is taking the lead in advancing the medical data utilization system by simultaneously obtaining 'Medical Data Content Certification' and 'Management System Certification' for the first time in a medical institution.
|
This article was translated by Naver AI translator.
