The Classification Of Multipara Afterpains Based On Xgboost Algorithm

Objective:This study introduces XGBoost algorithm machine learning,which may affect the harem production each factor in afterpains as the production of a harem shrinkage pain patients,the level of pain in patients with the classified recognition,in an objective perspective,to develop a new method of pain assessment,for clinical medical staff to provide a complete set of pain assessment system,improve the past do not take the harem pain reduction production situation,to provide evidence for production harem shrinkage of pain treatment.Method:1.600 multipara were randomly selected and questioned to collect the datas.The questionnaire included multiparas' basic information,delivery information,the degree of afterpains,etc.SPSS20.0 software is used to collect data for statistical analysis.2.Using XGBoost machine learning methods,each multipara is regarded as a sample,the data as the sample of attribute values,will be 70% before 600 samples as training input XGBoost computer algorithm program for black-box operation,form the evaluation model of machine,after will be 30% of the sample to verify the accuracy of the model as test group,in the end of machine learning to produce harem shrinkage degree of pain assessment.3.Will be subjective pain assessment compare with the result of objective pain assessment,with ROC curve and AUC value indicates that the effect of the two methods of assessment.Results:1.The multipara afterpains general information questionnaire consists of 15 items,incuding The vaginal birth information contains seven items and cesarean delivery operation information contains five items,in addition,to design supplementary delivery information questionnaire,increase seven vaginal birth infor-mation and seven cesarean section surgery information to improve the delivery information.2.The afterpains assessment,by evaluating seven fixed situation(the following Numbers for(1)(2)…(7))to assess pain score.The pain score results as follows:(1)Press uterine pain scored an average of 8.48±1.34;(2)ID of oxytocin pain scored an average of 6.93±1.51;(3)IM of oxytocin when pain scored an average of 5.25±1.54;(4)breastfeeding pain scored an average of 4.51±1.46;(5)when ambulation the pain scored an average of 5.67±1.32;(6)Change the position pain scored an average of 4.40±1.70,above(2)(3)(4)(5)(6)is moderate pain;The(7)without any stimulus scored an average of 3.48±1.75,for mild pain.3.The exploratory factor analysis,general information questionnaire,vaginal birth information questionnaire,cesarean delivery information questionnaire KMO value was 0.725,0.771and0.725,respectively,show strong correlation between the content of the questionnaire,there is a common factor(P < 0.001).4.There were significant difference(P<0.05)among multiparas' degree,the hospital diagnosis,delivery way,number of vaginal birth and number of cesarean section surgery,time and whether the maternal separation and breastfeeding;Whether or not to use in the labor epidural multipara produce the afterpains score difference obvious(P<0.05);Cesarean section surgery whether intraoperative hemostatic operation,postoperative micturition is obstructed,whether abdominal distension and uterine contractions are in good condition of multipara is produced harem contraction pain pain score had significant difference(P < 0.05).5.According to the multiple linear regression analysis,the main impact factors of multiparas' afterpains score was degree.In addition,whether or not to use in the labor epidural multipara produce was the main impact factor in the vaginal delivery multiparas' afterpains score.Whether intraoperative hemostatic operation was the main impact factor in the cesarean section multiparas' afterpains score.6.There are positive correlations among the afterpains in seven different time points(P<0.01).7.The show of classification result of machine learning,machine can according to the impact factor of single factor t test to determine the level of pain accurately,th-us to guide the clinical workers implement the corresponding nursing measures.8.Machine learning objective to evaluate the degree of multipara harem pain reduction production results and the results are compared,and the effect of subjective evaluation that the Kappa coefficient is 0.735,shows two kinds of evaluation results is highly consistency;Moreover,machine learning classification effect of AUC value = 0.869,shows that machine learning classification assessment results have a certain accuracy in statistics.Conclusion:1.The self-desinged questionnaire has good pertinence,concentration,which can be objective and accurate response multiparas' afterpains level.The questionnaire is good for investigate and collection the status of multiparas' afterpains.2.In the multiparas,most of their tolerance of afterpains is lower,there are many factors which impact on the afterpains about multipara.It's suggest the Obstetrics and Gynecology hospital that to pay attention to afterpains phenomenon,establish a standardized postpartum pain assessment process,and produce the prevention and control measures of afterpains,further reducing to lying-in physical labor pain for women and their families,so as to improve the quality of hospital services and maternal puerperal life.3.Using the machine learning classification to assess degree of afterpains,extract the impact factors of pain score automatically from the assessed,it is concluded that the classification of the degree of pain as a result.Not only greatly improve the efficiency of the clinical workers,but also opened up a new method for pain assessment,created a new way of thought,the future research is expected to apply this technology to cancer pain,other surgical pain,for the clinical of promotion.