Objective To analyze the expression of tumor metastasis-related gene transcription factor 21 (TCF21), ANGPT1(ANGPT1) and somatostatin receptor 2 (SSTR2) in thoracoscopic radical resection samples of lung cancer, and to construct and verify the joint prediction model of postoperative recurrence and metastasis, so as to provide reference for early clinical prediction of postoperative recurrence and metastasis. Methods A total of 149 patients with lung cancer who underwent thoracoscopic radical surgery in our hospital from March 2020 to February 2022 were prospectively selected as research objects, and divided into occurrence group and non-occurrence group according to whether recurrence and metastasis occurred one year after surgery. The random forest algorithm was used to screen and reduce the dimension of the characteristic variables of postoperative recurrence and metastasis. Logistic regression was used to analyze the related influencing factors of postoperative recurrence and metastasis, and multiple variables were fitted to predict the postoperative recurrence and metastasis model. Receiver operating characteristic curve (ROC) was used to analyze the predictive value of original covariates and joint predictors new for postoperative recurrence and metastasis. Results The T stage and N stage of the group were higher than those of the group without the disease, the number of lymph node stations, the number of lymph node stations, the number of lymph node stations and the expression levels of TCF21, ANGPT1 and SSTR2 were lower than those of the group without the disease (P＜0.05). Random forest algorithm showed that the top 3 variables in importance ranking were TCF21, SSTR2 and ANGPT1 expression. Logistic regression analysis showed that the expression levels of TCF21, SSTR2 and ANGPT1 were related factors for postoperative recurrence and metastasis (P＜0.05). Combined predictive factor New predicted that the AUC of postoperative recurrence and metastasis was greater than that of the original covariates (P＜0.05). Individual value prediction showed that, under the condition of diagnostic accuracy of 95.97%, no recurrence and metastasis occurred in this patient, and no recurrence and metastasis occurred in this case was confirmed by the joint predictor New. Conclusion The expression levels of TCF21, ANGPT1 and SSTR2 decreased in patients with lung cancer recurrence and metastasis after thoracoscopic radical surgery. The combined predictors of TCF21, ANGPT1 and SSTR2 fitted by Logistic regression model have certain predictive value for postoperative recurrence and metastasis