Development of a Qualitative Analysis Method Using Machine Learning in Simultaneous Analysis by GC-TOF-MS

Key Points

• The study aims to improve the analysis of unknown compounds using machine learning combined with GC-TOF-MS.
• Combined mass spectrometry with machine learning techniques.
• Optimized input data and model to enhance prediction accuracy.
• Utilized computational methods for chemical derivatization and resin pyrolysis.
• Increased prediction accuracy from 0.72 to 0.86 using cosine similarity.
• Achieved significant expansion of structural candidates for unknown compounds.
• Successfully estimated structures of compounds not present in the NIST database from metabolite samples.