Data-quality-guided AI strategies for natural product drug discovery.

The component specification problem-the mismatch between bioactivity observed in chemically unresolved extracts and the constituent-level chemical definition required for reliable artificial intelligence (AI) inference-remains a key bottleneck in computational natural product (NP) discovery. Without operational standards linking analytical evidence to AI strategy, even sophisticated models risk overinterpretation when applied to incompletely resolved extracts. Here, we introduce a data-quality-guided framework that aligns AI approaches with Metabolomics Standards Initiative (MSI) identification tiers. Informed by emerging case studies across MSI levels, this strategy enables researchers to match computational tools to available analytical resolution rather than exceed it. By integrating ethnopharmacological knowledge and multi-omics validation, the framework positions AI as a calibrated decision-support instrument for translational NP research.