I started in bioanalysis in 1969 when gas chromatography was ‘the technique’ for drug analysis. My interest in calibration mathematics and Monte Carlo techniques started in the early 1980’s when computer based chromatographic data systems appeared with differing regression mathematics, which were hotly debated by my then colleagues, especially with regard to weighting. Thanks to a brilliant maths teacher, I began to understand weighted regressions, teach others, and investigate the limitations of obtaining precise calibration curves with small data populations of calibrators. Around this time I also became interested in non-linear behaviour of assays and developed the regression for y = a + bx +cxln(x), which I found described most non-linear chromatographic behaviour.
Much of my career was spent as a project manager/team leader in various bioanalytical laboratories, but later I moved into installation qualification of bioanalytical software and effectively retired about five years ago.
One aspect that always attracted me about science was that fellow scientists were encouraged to present data from their findings in a way that was readily understood and open to critical review by their peers. In chromatographic bioanalysis this openness seems to have been overlooked in the area of calibration curves. In the days before the ubiquitous computer (yes, there was such an era when I started) assay calibrations were often performed manually by plotting data on graph paper, fitting the best line by eye and reading off the resultant estimates from this curve. This process was fine for the technology of the day and for assays with a dynamic range of about ten. Modern assays now have a typical dynamic range of about 1000, quantification is carried out automatically using regression curves generated within the data system and the calibration graph has been relegated to being a piece of decoration within a study report, serving no useful purpose other than to ‘tick a box’ as a requirement to be included in the study. It is time to reinstate the usefulness of calibration curves by changing their format.