SARS-CoV-2 infection poses a global health crisis. In parallel with the ongoing world effort to identify therapeutic solutions, there is a critical need for improvement in the prognosis of COVID-19. Here, we report plasma proteome finger print that predict high (hospitalized)and low risk(outpatients) cases of COVID-19 identified by a platform that combines machine learning with matrix-assisted laser desorption ionization mass spectrometry (MALDI-TOF MS) analysis. Sample preparation, MS and data analysis parameters were optimized to achieve an overall accuracy of 92%, sensitivity of 93%, and specificity of 92% in dataset without feature selection. Further on, we identified two distinct regions in the MALDI-TOF profile belonging to the same proteoforms. Unbiased discrimination of high and low-risk COVID-19patients employing a technology that is currently in clinical use may have a prompt application in the noninvasive prognosis of COVID-19. Further validation will consolidate its clinical utility.