An artificial intelligence (AI) that predicts the recurrence risk of non-small cell lung cancer patients one year in advance has been developed. Non-small cell lung cancer refers to the large type of lung cancer, accounting for 70-80% of all lung cancers. Given that the five-year survival rate for non-small cell lung cancer patients in Korea is about 37%, faster recurrence predictions are expected to help in formulating treatment strategies.
Professor Kim Hong-gwan of the Department of Thoracic Surgery at Samsung Medical Center and Professor Jeong Hyun-ae of the Department of Hematology and Oncology noted on the 8th, "We developed an AI called 'Radar Care' that predicts the recurrence risk by analyzing 14,177 non-small cell lung cancer patients." The findings of this research were published in the 'Journal of Clinical Oncology (JCO) Precision Oncology' on July 23.
The research team analyzed data from non-small cell lung cancer patients who underwent surgery from 2008 to 2022. By combining patient clinical information, pathology, and CT (computed tomography) results, the AI was designed to score the likelihood of recurrence one year later. The higher the possibility of recurrence, the higher the score.
The AI classifies patients into low-risk, intermediate-risk, and high-risk groups. The high-risk group showed a recurrence rate of 10% within one year. The intermediate-risk group had 5%, while the low-risk group had 1%. For example, high-risk patients categorized under stage 1 lung cancer had a recurrence and mortality risk 5.83 times greater than that of low-risk patients.
The research team explains that they can identify more dangerous patients even if they are all non-small cell lung cancer cases. Patients with a high recurrence risk score will undergo aggressive treatment, while those with a low score may consider a reduction in treatment duration. Professor Jeong Hyun-ae stated, "Early prediction of recurrence risk using AI will help determine a treatment direction favorable for patients."
References
JCO Precision Oncology (2025), DOI : https://doi.org/10.1200/PO-25-00172