Application
Quantify components that overlap
Chromatography separates components physically and takes as long as it takes. A multivariate model separates them mathematically, in 1 second.
The measurement problem
In a real mixture, absorption bands overlap, the baseline drifts with temperature and particulates, and other components vary independently. Picking one peak and measuring its height fails for exactly that reason — it cannot tell a rise in your analyte from a rise in something that overlaps it.
Partial least squares regression uses the whole spectral region, finding the directions in the data that covary with the reference concentrations. That is what makes it possible to report several species at once from a single acquisition.
What a live reading changes
- Multiple analytes reported simultaneously from one measurement.
- Overlapping bands separated by the model rather than by a column.
- The model built against your own HPLC or UV-VIS reference values.
- Preprocessing — baseline correction, derivatives, normalisation, region selection — documented as part of the model so it is transferable and auditable.
Chromatography stays the reference method and the source of the calibration values. What it cannot be is continuous.
Is this your measurement?
Send the stream, the concentration range and the decision it blocks. Where it makes sense we test on your own process samples first.