lung cancer, adenocarcinoma, pathological prediction, high-risk subtypes, follow-up time, nomogram, machine learning