Mass spectrometry-based proteomics technologies are the prime methods for the high-throughput identification of proteins expressed in complex biological samples. Nevertheless, mass spectrometry’s technical limitations still hinder its ability to identify low abundance proteins in complex samples. Characterizing such proteins is essential to provide a comprehensive understanding of the biological processes taking place in a sample. Still today, a large part of the mass spectrometry-based proteomics performed use a data-dependent approach that favors the acquisition of mass spectra and detection of proteins of higher abundance. Combined to the fact that the computational identification of proteins from mass spectrometry data is typically performed after mass spectrometry data acquisition, large numbers of mass spectra are redundantly collected from the same abundant proteins and little to no mass spectra are acquired for proteins of lower abundance. To address this problem, we propose a novel supervised learning algorithm that identifies proteins in real-time as mass spectrometry data is acquired and prevents the further data acquisition related to confidently identified proteins to improve the identification sensitivity of low abundance proteins. We show in real-time simulations of a previously performed mass spectrometry analysis of a HEK293 cell lysates that our approach can identify 92.1% of the proteins using 66.2% of the MS2 spectra acquired in the experiment. We also demonstrate that our approach is fast enough for real-time mass spectrometry analysis, is flexible and that it outperforms previously proposed methods. Our method efficient usage of mass spectrometry resources will provide a more comprehensive characterization of proteomes in complex samples.