Manuka Biologicals have found NIR plus machine learning models to be the ideal quality control tool. It doesn't completely replace more time-consuming and expensive GC-FID analyses but rather complements them. NIR is used for quality control, while GC-FID is used for final Certificates of Analysis.
While an increasing proportion of the Manuka Biologicals oil is extracted from plantation-grown manuka (Leptospermum scoparium), some wild leaf material is also harvested. One potential issue with wild-sourced leaf material is the potential for contamination between manuka and kanuka (Kunzea ericoides). While these two plants are botanically very different, they look very similar and it is easy for harvesting teams to accidentally include kanuka leaf material with manuka leaf. Sagitto has used AI to build an accurate model to detect even small levels of accidental mixtures of these two different oils, and this check is included automatically with every report.
Manuka Biologicals are able to test every batch of oil, and immediately grade it. This means that there is no risk of down-blending : taking a high-value batch and mixing it with a lower-value batch.
"In conjunction with Sagitto we have been able to develop a model based on machine learning to rapidly determine MBTK levels during production or as part of our R&D, on very small sample quantities. This allows us to rapidly and accurately grade production oils and direct them to appropriate stock grade allocations within a few minutes, to maximise our sales revenues."
Dr Wayne Campbell, Manuka Biologicals Ltd
