Machine learning-based assessment of sustainable extraction methodologies tackling the biotechnological exploitation of Arnica montana extracts

Heat-assisted (HAE), ultrasound-assisted (UAE), microwave-assisted (MAE), and pressurized liquid extraction (PLE) represent diverse techniques with distinct physical principles that influence the efficiency and selectivity of bioactive compound recovery from Arnica montana (AM) flowers. These techniques are expected to generate unique metabolite profiles, affecting the composition and functionality of extracts. This study combined untargeted metabolomics with machine learning-based chemometrics and a comprehensive assessment of in vitro biological activities of hydroethanolic AM extracts obtained using the four techniques. PLE yielded the most distinctive phytochemical profile. All extracts contained key phenolics such as anthocyanins, lignans, and related compounds. MAE extract exhibited strong antioxidant and neuroprotective effects, associated with triterpenoid metabolites, while PLE extract showed anti-inflammatory, cytotoxic, and antioxidant activities, mainly influenced by anthocyanins and flavonols. These findings deepen our understanding of how extraction technologies shape the functional potential of AM extracts, facilitating their application as nutraceuticals or bioactive ingredients in the food sector.