Analysis of Relationship between Tumor Markers and Tumors Detected by Chemiluminescence Immunoassay via Artificial Neural Network

Abstract

The health of people is seriously threatened by the class of disorders known as tumors because of their high risk and high mortality. Early detection and treatment are the most efficient approaches to stop cancers and lower mortality, according to a substantial body of research and preventative data. Therefore, the search for early-diagnosis tumor indicators has taken center stage. The existence of tumors may be detected by a family of molecules known as tumor markers, which have a high sensitivity and specificity. Tumor markers are mainly used for diagnosing tumors, judging efficacy and evaluating prognosis. At present, the detection methods of tumor markers include enzyme-labeled immunoassay technology and chemiluminescence immunoassay (CLIA) technology. In recent years, with the continuous development of in vitro diagnostic technologies, CLIA has the advantages of high sensitivity, wide detection range, simple operation, good repeatability and specificity, high degree of automation, and no radioactive staining of reagents. It has a good application prospect in clinical medical diagnosis. In this paper, the relationship between tumor markers and tumors is detected by CLIA, and then the tumor markers are analyzed by artificial neural network, and the classification and screening of cancer detection is completed. Finally completed the following work: 1) Introduced the research status of CLIA at home and abroad, and provided a theoretical basis for the analysis method proposed later. 2) The technical principle of ANN is introduced and the SAE neural network model is proposed. 3) Select the model evaluation index, and select the SAE parameters through experiments to construct the optimal SAE model. Input the sample data and then get the accuracy, recall and F1 score of the model. Compared with other models, it can be found that the SAE model proposed in this paper has the best detection performance.