#AI reads Urine# Rapid, non-invasive diagnosis of tuberculosis using desorption separation ionization mass spectrometry (DSI-MS) and urinary carbonyl metabolite fingerprints
Published 23 March, 2026
This study proposes a rapid, non-invasive tuberculosis diagnostic method based on Desorption Separation Ionization Mass Spectrometry (DSI-MS) combined with N-(2-Aminoethyl) piperidine (AED) carbonyl derivatization, using urine as the test sample. Only simple pretreatment (reagent mixing, heating, filtration) is required, and the analysis can be completed in 3 minutes. The study tested urine samples from 151 tuberculosis patients and 151 healthy controls, identifying 65 metabolites (32 derivatized and 33 underivatized). It was found that tuberculosis patients exhibit significant perturbations in pathways such as energy metabolism and nitrogen metabolism (especially arginine and proline metabolism), with elevated ornithine levels. A neural network classifier constructed through machine learning showed excellent performance, achieving an AUC of 0.927 on the training set and 0.922 on the test set. Notably, the diagnostic effect combining derivatized and underivatized features was the best. In addition, the study compared metabolic differences between the drug-resistant tuberculosis group and the disease control group, revealing that perturbations in pathways such as purine metabolism, nicotinate and nicotinamide metabolism may be associated with drug resistance. This method provides a new approach for rapid tuberculosis screening but still has limitations, including limited derivatization coverage and a small sample size in the drug-resistant group. Future studies need to expand the sample size and optimize the method to promote clinical application.
Talanta. 2026 Jan 13:302:129340. doi: 10.1016/j.talanta.2025.129340.
Youhe Gao
Statement: During the preparation of this work the author(s) used Doubao / AI reading for summarizing the content. After using this tool/service, the author(s) reviewed and edited the content as needed and take(s) full responsibility for the content of the published article.
For earlier AI Reads Urine articles：