Development Of The Diagnostic System Of Lung Cancer

OBJECTIVES:1.Lung Cancer is the most common human cancer,each year in worldwide, there were about 100 million people die of lung cancer.To improve lung cancer treatment and survival rates,more effective measures is as early as diagnosis,as early as the inspection and early treatment.2.At rural secondary and tertiary medical institutions,for single simple means of clinical diagnosis and doctors of low technological level,the development of lung cancer early diagnosis system based on the evidence-based patient cases to judge the basic situation,then,make out the conclutions and the follow-up recommendations for doctors' physicians provide.It can save expenses.METHODS:1.The use of questionnaires collected in the form of primary lung cancer(as the case group) benign lung diseases(pneumonia,tuberculosis,pneumonia, inflammatory pseudotumor and so on,as the control group)in patients with a total of 1883 cases of clinical data;2.SPSS11.5 statistical software were used for retrospective analysis of data processing,base on the relevant variables(the impact of factors or indicators), single-factor(t-test,chi-square test) and multi-factor non-conditional Logistic regression analysis(forward method,enter method) were used to build up the establishment of regression model; 3.Reference the models,by using of Microsoft Visual Studio 2005 System Development platform and VB.Net language to develop the lung cancer diagnosis system.RESULTS:1.Single-factor analysisVariable on the classification of a single factor analysis(chi-square test)showed that:"gender","marital status","professional","residence","smoking" in the case of the basic group and the control group,there was a significant difference(P＜0.05)." blood Sputum","chest pain","Fever","shortness of breath","weight loss,fatigue,loss of appetite","superior vena cava syndrome","hoarseness","the proliferation of other local","distant lymph node metastasis","liver","Bone transfer","brain metastases" in the case of signs,such as case group and control group,there was a significant difference(P＜0.05).In addition,the "sputum smears","pleural effusion smears", "pleural effusion sediment biopsy Solidification","brushing bronchofibroscope", "bronchoalveolar lavage check","Superficial lymph node fine needle aspiration cytology","by fiberoptic bronchoscopy needle aspiration cytology","Percutaneous needle aspiration cytology","B-ultrasound","Tblb check","CT biopsy","surgical removal of superficial lymph nodes histopathological examination","fiberoptic bronchoscopy","X-ray examination","CT examination","exploratory thoracotomy", "PET examination","MRI examination","SPECT examination","looking for the transfer"the results of such examinations in the case group and control group,there was a significant difference(P＜0.05).Numerical model of a single factor analysis of variables(t-test) showed that: "Age","alanine aminotransferase(ALT)","aspartate aminotransferase(AST)", "albumin","Mg","Albumin/Globulin","Ca","P","neuron-specific enolase","Tissue peptide antigen(TPS)","tissue polypeptide specific antigen(TPS)","CEA","Cancer antigen 199(CA199)","Cancer antigen242(CA242)","Prostate Specific Antigen(PSA)",et al.These variables between the case group and control group, there was a significant difference(P＜0.05).2.Multi-factor non-conditional logistic regression analysis Reference to the results of the single-factor analysis,then proceed to the second classification of non-conditional logistic regression(forward method) analysis,a total of 21 operations come into the election.Finally,for lung cancer,significantly affected variables(P＜0.05)is:"smoking","bone metastases","brain metastases","blood Sputum ","chest pain","fever","sputum smears","pleural effusion cytology","brush biopsy bronchofibroscope","tblb-histological type","CT-targeted percutaneous lung biopsy -histological type","superficial lymph node fine needle aspiration cytology","X-ray -lesion density","CT inspection-the edge of lesions","exploratory thoracotomy-the number of lesions","PET-histological type","PET-lesion edge","whole body bone imaging(SPECT)","looking for transfer","age",such as 20 variables.3.The diagnosis modelCome to consider the case when they were on the basic inspection,first of all, the basic situation of category(age,smoking history) and out-patient check-type (chest pain,fever,blood Sputum,bone metastases,brain metastases)were set into the multi-factor non-conditional logistic model(Model one),its goodness-of-fit is 0.459, accuracy rate of 73.8%.Base on the model one,increase in CT examination and Sputum smear,a Model two,its goodness-of-fit is 0.535,accuracy rate of 77.0%. Base on the model one,increase in X-ray and Sputum smears,a Model three,its goodness-of-fit is 0.509,accuracy rate of 76.3%.4.The usage of the diagnosis system180 cases of lung cancer and non-lung cancer cases were used to verify the extrapolation.It shows that the accuracy rates of the diagnosis system were 70.5% (Model one),72.8%(Model two),and 71.7%(Model three).CONCLUSIONS:1.To the software system,although the model can be good used for diagnosis to existing the cases.However,the core place——mathematical model is still in the foundation stage,the reliability of using data mining technology has yet to be verified in practice.2.Considering the data was sourced by a single unit,so the extrapolation was not to be known,along with the increase in multi-center data,the model's ability to enhance will extrapolation.