A Wrapper-Filter Algorithms For Identification Of Serum Biomarkers To Detect Early Silicosis Patients

Silicosis, an occupational disease with the characteristics of external cell matrix deposition in lung interstitial, is due to the long-time inhalation of silica dust particles. Early health care is the key to silicosis prevention, but until now no effective indicators of early health care. Discussing the pathogenesis of silicosis and searching for serum early diagnosis (screening) biomarkers for prevention, treatment and ultimately the elimination of silicosis has important economic and social significance. In recent years, the development of proteomics technology, in particular, Mass spectrometry technology for the diagnosis of silicosis has opened up new areas. The ClinProt system, composed by the beads, matrix-assisted laser desorption ionization time of flight mass spectrometry (MALDI-TOF-MS) and ClinProTools, with its high sensitivity and repeatability can be used to study the serum protein Spectrum. The protein was purified with magnetic bead separation method to remove high abundance proteins and other impurities, while enrichment of low abundance protein. The peptide mass fingerprintings in the silica dust exposure group and control group were obtained. Which provide important clues for searching for the biomarkers of silicosis. Applying feature selection(FS) techniques play an important role in analyzing high dimensional data that is common in high-throughput screening such as microarray and mass spectrometry data. Spectrometry data analyses are mainly discussed in our work and the application of Genetic Algorithm algorithm combined with SVM in biomarker selection is also studied in the work. Development of feature selection techniques and the discovery of biomarkers are of much value in clinical diagnosis. The main contents of this dissertation are as follows:1) The fundamental principle of feature selection techniques was studied. Based on the analysis of the advantage and disadvantage of the advantages and disadvantages of different algorithm, a appropriate feature selection method is received.2) New method is raised using Relief algorithm to filter the original data set, and then using Genetic Algorithm as searching process Combined with support vector machines(SVM),it was applied to screening for biomarker or serum protein for early diagnosis of silicosis.3) The experiments show that a optimal feature subset were selected by Relief-GA-SVM to establish a diagnostic model. We establish the diagnostic pattern to distinguish each stage of silica-exposed population, the diagnostic pattern worked excellently with 96.30%, 96.85% and 97.44% of classifying accurate rate for phase 0,phase 0+ and phaseâ… of silicosis respectively.The main contributions of this paper are summarized and the further researches on work are suggested at the end of this dissertation.