A Study On Predicting The Efficacy Of Administering Expansive Open-door Laminoplasty With Cervical OPLL Based On A Categorical And Regression Tree Model

Objective To predict the surgical effect of cervical OPLL patients after expansive open-door laminoplasty by using classification regression tree model,so as to provide scientific reference for perioperative management.Methods Retrospective analysis was made on 90 patients with cervical OPLL who were treated in Ningxia Medical University General Hospital from 2015 to2022.According to the JOA score efficacy standard,the improvement rate of JOA score after surgery ?50% was defined as the excellent efficacy group(n=58 cases),and the improvement rate of JOA score<50% was defined as the poor efficacy group(n=32 cases).The age,sex,follow-up time,symptom duration,spinal canal space occupying rate,C2-C7 Cobb angle,cervical spine range of motion,cervical spine segment range of motion,T1 S,K line((10)/-),whether the ossification focus involved the segment 3,OPLL classification,whether the intramedullary MRI T2 WI was significantly high signal,and the number of titanium plates were compared between the two groups.The classification regression tree was used to build a model of surgical efficacy and influencing factors,and the model was verified and evaluated.Results The analysis results showed that(1)90 patients JOA scores from10.31±2.60 preoperatively to 14.06±1.98 at the last postoperative follow-up,JOA score improvement rate of 58.07%±21.35%,excellent rate of 64.44% VAS score from 3.24±1.34 preoperatively to 1.66±1.06 at the last postoperative follow-up;(2)neurological function recovery The differences in preoperative JOA score,postoperative JOA score,JOA score improvement rate,and postoperative VAS score were statistically significant in the excellent group compared with the poor group(P<0.05),and the differences in preoperative VAS score in the excellent group compared with the poor group were not statistically significant(P>0.05);(3)In the analysis of the differences in the factors affecting surgical efficacy between the two groups,age and whether intramedullary MRI T2 WI significant high signal were compared The difference was statistically significant(P<0.05);(4)in the analysis of the categorical regression tree model,the top three factors affecting surgical efficacy based on the importance of features were age,C2-C7 Cobb angle,and cervical segmental mobility in that order;(5)the model was applied to the test set validation,and its accuracy rate was obtained as 80%.when the efficacy was excellent,the accuracy,recall,and F1 scores were 84%,80%,and 82%,respectively;When the efficacy was moderately poor,the accuracy,recall,and F1 scores were 75%,80%,and 77%,respectively;the performance of the model was evaluated using the confusion matrix,and its accuracy,sensitivity,and specificity were 80%;according to the ROC curve analysis,the area under the ROC curve AUC=0.81.Conclusion 1.Advanced age and significant high signal on intramedullary MRI T2 WI are the risk factors for posterior single-opening in cervical OPLL patients to affect surgical outcomes;2.The importance of model features in this study revealed that age,C2-C7 Cobb angle,and cervical segmental mobility are the top three predictors of surgical outcomes;3.The model constructed based on the categorical regression tree has better performance and is simple to use and operable,which can provide a basis for clinicians to formulate subsequent interventions for cervical OPLL patients,identify high-risk groups and improve neurological recovery.