Our method is based on the prediction of Sepsis label with the help of all the parameters (vitals, lab and demographics) except EtCO2. Research has shown that EtCO2 has low but significant, correlation with SOFA score . Therefore, we formulated a hypothesis which is consideration of EtCO2 might produce biased results due to very less correlation with SOFA score, so, we took out the EtCO2 from the total feature set. Then, we fit a support vec-tor machine (SVM) regression model to the physiological parameters (vitals and lab parameters without EtCO2) and demographic data. The perfor-mance of the model without EtCO2 increased considerably than considering EtCO2. In our second approach, when we used SOFA and qSOFA scores computed according to the third international consensus definitions for sep-sis, with equal weight as the SVM model, the performance was better. In this approach, the NAN values of vital signs, were replaced by an estimation of that parameter in that time instant. The performance of our both approaches on PhysioNet/CinC Challenge 2019 public data set, which are computed by 2019 Challenge scoring algorithm, are shown in Table 1. Also, we found changes in cross-correlation of vital signs, especially heat rate and oxygen saturation (XCorr-HR-SpO2), after happening of sepsis, which is recently shown for prediction of sepsis in low birth weight infants. In future, we would like to implement the cross-correlation parameters as discriminatory features and to investigate more on the association between EtCO2 with the qSOFA and SOFA scores. We believe these changes would be helpful in developing the model more rather than not considering the parameters.