Morbidity prediction and risk analysis of second primary malignancies of lung cancer patients

Advances in medical care have increased second primary malignancies (SPMs) among lung cancer survivors, necessitating targeted research to guide prevention, screening, and treatment. Using SEER data from 2004-2015, we analyzed 16 clinical factors with eight machine learning models to predict SPMs occurrence and prognosis among 99,901 lung cancer patients, of whom 13.2% developed SPMs. The Light GBM model achieved the highest AUC (0.7140 ± 0.0053 and 0.7182 ± 0.0121), with SHapley Additive exPlanation values indicating bilobectomy or pneumonectomy, lower staging, and age 54-85 as key factors for SPMs risk. Poor prognosis was linked to smaller resection, advanced staging, male gender, and age >70. The Kaplan-Meier survival curves indicate that patients with SPMs have a better prognosis and a longer median survival time. The Cox proportional hazards model highlighted that male, square cell carcinoma, cancer staging (N1-N3, M1), and single (included divorced, widowed and single) SPMs patients had worser prognosis, while Asian or Pacific Islander patients, black patients, certain surgical resections, beam radiation and chemotherapy performed SPMs patients had better prognosis. Increased monitoring of lung cancer patients aged 54-85 with bilobectomy or pneumonectomy and advanced staging may help prevent SPMs. Appropriate surgical resection and radiotherapy can significantly improve SPMs patient survival.