ML and AI-Enhanced Metabolomics and Foodomics

LC-MS/MS data processing is currently carried out based on the software packages provided by the instrument’s manufacturer (In our laboratory raw files are processed in Sciex OS using the MQ4 integration algorithm) through a semi-automated yet labor-intensive workflow. While the software packages provide a starting point, manual review and re-integration by analysts remain essential to correct for peak tailing, background noise, matrix interferences, and retention time shifts. Although this manual curation step ensures data quality, it substantially limits throughput. In a full batch of 100 samples, data processing alone can take up to three working days. This project aims to replace labor-intensive LC–MS/MS data processing with a robust, AI-assisted workflow that accelerates throughput while preserving the rigor required in accredited environments. The primary goals are to: (1) automate peak detection and integration for scheduled MRM data to reduce processing time from multiple days per batch to hours; (2) minimize human error and inconsistency by standardizing decisions across large datasets; and (3) maintain full analyst oversight through an intuitive, responsive GUI that enables rapid batch-level review, transparent adjustments, and efficient curation. Success will be measured by reductions in processing time and re-integration rates, reproducibility gains across batches and matrices, and maintenance of accuracy at or above manual curation benchmarks—delivering a trustworthy, high-throughput solution that supports expanding analytical demands.