Biostatistical analysis of principle component of salivary Raman spectra for NS1 infection

Abstract

Non-structural protein 1 (NS1) has been clinically accepted as an alternative biomarker for diseases related to flavivirus infection such as Dengue fever, Japanese encephalitis, Murray Valley encephalitis, Tick-borne encephalitis, West Nile encephalitis and Yellow fever. It is detectable in the blood serum and saliva of an infected patient during the febrile phase using the Enzyme-linked Immunosorbent Assay (ELISA) technique. However, low sensitivity is observed for detection of NS1 in saliva than in blood. To improve the detection of NS1 in saliva, Surface Enhanced Raman Spectroscopy (SERS), a highly sensitive molecular analysis technique has been proposed as an alternative detection technique. In this study, salivary SERS spectra of dengue suspected patients and control groups have been analyzed using Principal Component Analysis (PCA). Using PCA, the SERS spectra has been reduced to a lower dimension of data known as principal components (PC)s. The produce PCs are then analyzed with normality test to identify the distribution of each PC. Finally, Independent Sample T-test (ISTT) and Mann Whitney U-test (MWUT) are conducted on normal distribution and non-normal distribution PCs, respectively to investigate the statistical significance differences between the positive and negative samples. SPSS software is used for the statistical analysis. Two datasets; (i) NS1 ELISA and (ii) NS1 Rapid datasets are analyzed. Using ISTT and MWUT, PC1 and PC2 are identified to be statistically significance difference between the positive and negative samples of NS1 ELISA dataset. Meanwhile, for NS1 Rapid dataset, the statistically significant PCs are PC1, PC3 and PC5. These PCs are potential to be used as inputs for classification of positive and negative samples.