Construction And Validation Of A Predictive Model For Central Lymph Node Metastasis In Papillary Thyroid Carcinoma

Objective: To explore the correlation between preoperative fine needle aspiration tissue DNA(FNAT-DNA)BRAF V600 E gene mutation and cervical central lymph node metastasis of papillary thyroid carcinoma,and to establish a predictive model for cervical central lymph node metastasis.Methods: Retrospective analysis of clinicopathological data(gender,age,concomitant diseases,maximum diameter of cancer nodules,location of cancer nodules,aspect ratio of cancer nodules,calcification of cancer nodules,boundary of cancer nodules,whether it invades capsule,FNAT-DNA BRAF V600 E gene mutation results,number of lesions and neck Lymph node metastasis)of 488 patients with papillary thyroid carcinoma admitted to our hospital from January 2019 to December 2021.According to the BRAF V600 E gene mutation results,they were divided into positive group and negative group.According to postoperative pathological lymph node metastasis,they were divided into central lymph node non-metastasis group and central lymph node metastasis group.Univariate analysis and binary logistic regression were used to analyze the correlation between BRAF V600 E gene mutation,lymph node metastasis in central cervical region and clinicopathological characteristics.Construction of Nomogram Prediction Model by Logistic Regression Analysis,building a Convolutional Neural Network Prediction Model Based on Artificial Intelligence Deep Learning.The prediction performance of the two models was compared by ROC curve and prediction probability.Results:1.Among the 488 patients with papillary thyroid carcinoma,99 patients had no BRAF V600 E gene mutation,and 389 patients had BRAF V600 E gene mutation,with a mutation rate of 79.7%.2.The results of the relationship between BRAF V600 E gene mutation and clinicopathological characteristics show that: BRAF V600 E gene mutation is related to capsule invasion and lymph node metastasis(P<0.05).3.Univariate analysis of central cervical lymph node metastasis and clinicopathological data showed that central lymph node metastases were associated with age,maximum diameter of cancer nodules,capsule invasion,and BRAF V600 E gene mutation(P<0.05).4.Binary Logistic regression analysis showed: age(R=3.380,P=0.000,maximum diameter of cancer nodules(OR=2.228,P=0.000;OR=4.795,P=0.000),invasion of capsule(OR=1.507,P=0.000;OR=4.795,P=0.000)P=0.027)and BRAF V600 E gene mutation(OR=6.410,P=0.000)were independent risk factors for central cervical lymph node metastasis.5.The results of the nomogram prediction model and prediction efficiency show that: the area under the ROC curve of the nomogram prediction model is 0.778,which is higher than the age,the largest diameter of the cancer nodule,the invasion Capsule and BRAF V600 E mutations have great predictive value.6.The results of the convolutional neural network prediction model and prediction performance show: the area under the ROC curve of the training set = 0.89,and the area under the ROC curve of the test set =0.78.Conclusion:1.Age,maximum diameter of cancer nodules,capsule invasion,and BRAF V600 E gene mutation may be independent risk factors for lymph node metastasis in the central cervical region.2.Convolutional neural network model and nomogram model may have better predictive performance of central lymph node metastasis.