Optimisation of HPLC gradient separations using artificial neural networks (ANNs): application to benzodiazepines in post-mortem samples

J Chromatogr B Analyt Technol Biomed Life Sci. 2009 Mar 1;877(7):615-20. doi: 10.1016/j.jchromb.2009.01.012. Epub 2009 Jan 21.

Abstract

Artificial neural networks (ANNs) were used in conjunction with an experimental design to optimise a gradient HPLC separation of nine benzodiazepines. Using the best performing ANN, the optimum conditions predicted were 25 mM formate buffer (pH 2.8), 10% MeOH, acetonitrile (ACN) gradient 0-15 min, 6.5-48.5%. The error associated with the prediction of retention times and peak widths under these conditions was less than 5% for six of the nine analytes. The optimised method, with limits of detection (LODs) in the range of 0.0057-0.023 microg/mL and recoveries between 58% and 92%, was successfully applied to authentic post-mortem samples. This method represents a more flexible and convenient means for optimising gradient elution separations using ANNs than has been previously reported.

Publication types

  • Evaluation Study

MeSH terms

  • Animals
  • Autopsy
  • Benzodiazepines / blood*
  • Chromatography, High Pressure Liquid / methods*
  • Chromatography, High Pressure Liquid / standards*
  • Humans
  • Neural Networks, Computer*
  • Sheep

Substances

  • Benzodiazepines