Korean medical researchers have developed an AI model capable of predicting the recurrence risk in early-stage non-small cell lung cancer (NSCLC) patients up to one year in advance.

NSCLC accounts for 85 percent of all lung cancers, with early-stage patients comprising 35 percent, for whom surgical treatment is generally considered.

Currently, patients undergoing surgery for early-stage NSCLC receive follow-up examinations every three to six months, based on their disease stage, regardless of their individual risk of recurrence. However, even within the same stage, differences in patient condition and tumor characteristics lead to varying recurrence risks, creating clear limitations in establishing personalized treatment strategies.

Samsung Medical Center announced on Monday that a research team led by Professor Kim Hong-kwan of the Department of Thoracic and Cardiovascular Surgery and Professor Jung Hyun-ae of the Department of Hematology and Oncology has developed the artificial intelligence model called “RADAR CARE (Real-time Risk-Adapted Surveillance Comprehensive Strategy AI Model for Early-Stage NSCLC).”

This model predicts the likelihood of recurrence within one year by integrating clinical, pathological, and test data from patients with early-stage NSCLC who underwent surgical treatment.

The research team analyzed data from 14,177 patients with early-stage NSCLC who underwent surgery at Samsung Medical Center between January 2008 and September 2022. RADAR CARE is a transformer-based deep learning model that simultaneously processes various types of data, including clinical information, pathology test results, and CT scan results. The research team incorporated patient-specific clinical information and test results into this model to quantify the risk of recurrence within one year.

Based on the RADAR CARE scores calculated by this model, the research team classified patients into three groups: low-risk (scores of 0.3 points or below), intermediate-risk (scores between 0.3 points and 0.6 points), and high-risk (scores above 0.6 points).

According to the study, regardless of lung cancer stage, the high-risk group with a high radar score showed a 10% recurrence rate within one year, compared to 5% in the intermediate-risk group and 1% in the low-risk group. This confirmed a trend where recurrence rates decrease as the radar score decreases.

Notably, even stage 1 patients with high RADAR CARE scores could have higher recurrence rates than stage 3 patients, while stage 3 patients with low RADAR CARE scores had a very low recurrence probability. Regardless of stage (1-3), the intermediate-risk group had a 3.59-fold higher risk of recurrence or death compared to the low-risk group, and the high-risk group had a 9.67-fold higher risk.

Furthermore, within the same stage, the high-risk group showed a significantly higher risk of recurrence and death compared to the low-risk group: 5.83 times higher for stage 1, 1.75 times higher for stage 2, and 1.84 times higher for stage 3.

Based on this, the research team classified patients according to their radar score immediately after surgery and changes over the follow-up period, proposing a tailored treatment strategy. Patients with persistently high RADAR CARE scores, indicating a high recurrence risk, should receive aggressive treatment. Those whose scores decrease over time may be considered for shorter treatment durations.

“Although 34.6 percent of lung cancer patients in Korea are diagnosed at an early stage, the five-year survival rate remains only 36.8 percent,” Professor Jung said. “This is because existing staging classifications alone make it difficult to accurately predict patient prognosis. We expect the model developed this time to help determine more favorable treatment directions for patients.”

Professor Kim remarked, “Early-stage non-small cell lung cancer patients generally have a relatively good prognosis if detected quickly and undergo surgery early. However, it's still difficult to feel completely reassured.”

Kim continued, “This model is the result of careful consideration, recognizing that treatment outcomes can vary even for the same stage, depending on the strategy chosen. I believe it will be invaluable in helping more patients recover their health with peace of mind.”