摘要： Background Escherichia coli is currently unable to be reliably differentiated from Shigella species by routine matrix-assisted laser desorption ionization-time of flight mass spectrometry (MALDI-TOF MS) analysis. In the present study, a reliable and rapid identification method was established for Escherichia coli and Shigella species based on a short-term high-lactose culture using MALDI-TOF MS and artificial neural networks (ANN). Materials and methods The Escherichia coli and Shigella species colonies, treated with (Condition 1)/without (Condition 2) a short-term culture with an in-house developed high-lactose fluid medium, were prepared for MALDI-TOF MS assays. The MS spectra were acquired in linear positive mode, with a mass range from 2000 to 12000 Da and were then compared to discover new biomarkers for identification. Finally, MS spectra data sets 1 and 2, extracted from the two conditions, were used for ANN training to investigate the benefit on bacterial classification produced by the new biomarkers. Results Twenty-seven characteristic MS peaks from the Escherichia coli and Shigella species were summarized. Seven unreported MS peaks, with m/z 2330.745, m/z 2341.299, m/z 2371.581, m/z 2401.038, m/z 3794.851, m/z 3824.839 and m/z 3852.548, were discovered in only the spectra from the E. coli strains after a short-term high-lactose culture and were identified as belonging to acid shock protein. The prediction accuracies of the ANN models, based on data set 1 and 2, were 97.71±0.16% and 74.39±0.34% (n = 5), with an extremely remarkable difference (p < 0.001), and the areas under the curve of the receiver operating characteristic curve were 0.72 and 0.99, respectively. Conclusions In summary, adding a short-term high-lactose culture approach before the analysis enabled a reliable and easy differentiation of Escherichia coli from the Shigella species using MALDI-TOF MS and ANN.