Skip to main content
PMC home page
AAPS PharmSci logoLink to AAPS PharmSci
. 1999 Dec 17;1(4):5–13. doi: 10.1208/ps010417

Empirical versus mechanistic modelling: Comparison of an artificial neural network to a mechanistically based model for quantitative structure pharmacokinetic relationships of a homologous series of barbiturates

Ivan Nestorov 1,✉, Malcolm Rowland 1, S T Hadjitodorov 2, I Petrov 2
PMCID: PMC2751347  PMID: 11741213

Abstract

The aim of the current study was to compare the predictive performance of a mechanistically based model and an empirical artificial neural network (ANN) model to describe the relationship between the tissue-to-unbound plasma concentration ratios (Kpu's) of 14 rat tissues and the lipophilicity (LogP) of a series of nine 5-n-alkyl-5-ethyl barbituric acids. The mechanistic model comprised the water content, binding capacity, number of the binding sites, and binding association constant of each tissue. A backpropagation ANN with 2 hidden layers (33 neurons in the first layer, 9 neurons in the second) was used for the comparison. The network was trained by an algorithm with adaptive momentum and learning rate, programmed using the ANN Toolbox of MATLAB. The predictive performance of both models was evaluated using a leave-one-out procedure and computation of both the mean prediction error (ME, showing the prediction bias) and the mean squared prediction error (MSE, showing the prediction accuracy). The ME of the mechanistic model was 18% (range, 20 to 57%), indicating a tendency for overprediction; the MSE is 32% (range, 6 to 104%). The ANN had almost no bias: the ME was 2% (range, 36 to 64%) and had greater precision than the mechanistic model, MSE 18% (range, 4 to 70%). Generally, neither model appeared to be a significantly better predictor of the Kpu's in the rat.

References

  • 1.Thakur AK. Model mechanistic vs empirical. In: Rescigno A, Thakur AK, editors. New Trends in Pharmacokineties. New York: Plenum Press; 1991. pp. 41–51. [Google Scholar]
  • 2.Rescigno A, Beck JS. The use and abuse of models. I Pharmacokinet Biopharm. 1987;15:327–327. doi: 10.1007/BF01066325. [DOI] [PubMed] [Google Scholar]
  • 3.Veng-Pedersen P, Modi NB. Neural networks in pharmacodynamic modeling: is current modeling practice of complex kinetic systems at a dead end? J Pharmacokinet Biopharm. 1992;20:397–412. doi: 10.1007/BF01062465. [DOI] [PubMed] [Google Scholar]
  • 4.Siegel RA. Commentary on “Neural networks in pharmacodynamic modeling: is current modeling practice of complex kinetic systems at a dead end?”. J Pharmacokinet Biopharm. 1992;20:413–416. doi: 10.1007/BF01062466. [DOI] [PubMed] [Google Scholar]
  • 5.Veng-Pedersen P. Response to Siegel's commentary. J Pharmacokinet Biopharm. 1992;20:417–418. doi: 10.1007/BF01062467. [DOI] [Google Scholar]
  • 6.Seydel JK, Schaper K-J. Quantitative structure-pharmacokinetic relationships and drug design. In: Rowland M, Tucker G, editors. Pharmacokinetics: Theory and Methodology International Encyclopedia of Pharmacology and Therapeutics, Section 122. Oxford: Pergamon Press; 1986. pp. 311–368. [Google Scholar]
  • 7.Wasserman PD. Neural Computing, Theory and Practice. New York: Van Nostrand Reinhold; 1989. [Google Scholar]
  • 8.Haykin S. Neural Networks. A Comprehensive Foundation. Toronto: Maxwell Macmillan Canada; 1994. [Google Scholar]
  • 9.Heicht-Nielsen R. Neurocomputing. Reading, MA: Addison Wesley; 1990. [Google Scholar]
  • 10.Jain AK, Mao J, Mohiuddin KM. Artificial neural networks: a tutorial. Computer. 1996;29:31–44. doi: 10.1109/2.485891. [DOI] [Google Scholar]
  • 11.Shang Y, Wah BW. Global optimisation for neural network training. Computer. 1996;29:45–55. doi: 10.1109/2.485892. [DOI] [Google Scholar]
  • 12.Erb RJ. Introduction to backpropagation neural network computation. Pharm Res. 1993;10:165–170. doi: 10.1023/A:1018966222807. [DOI] [PubMed] [Google Scholar]
  • 13.Erb RJ. The backpropagation neural network-;a Bayesian classifier. Introduction and applicability to pharmacokinetics. Clin Pharmacokinet. 1995;29:69–79. doi: 10.2165/00003088-199529020-00002. [DOI] [PubMed] [Google Scholar]
  • 14.Veng-Pedersen P, Modi NB. Neural networks in pharmacodynamic modeling. J Pharm Sci. 1993;82:918–926. doi: 10.1002/jps.2600820910. [DOI] [PubMed] [Google Scholar]
  • 15.Brier ME. Neural network gentamicin peak and trough predictions from prior dosing data. Clin Pharm Ther. 1995;57:157–157. [Google Scholar]
  • 16.Brier ME, Zurada JM, Aronoff GR. Neural network predicted peak and trough gentamicin concentrations. Pharm Res. 1995;12:406–412. doi: 10.1023/A:1016260720218. [DOI] [PubMed] [Google Scholar]
  • 17.Brier ME, Aronoff GR. Neural network predictions of cyclosporine A blood concentration. Clin Pharm Ther. 1995;57:157–157. [Google Scholar]
  • 18.Smith BP, Brier ME. Statistical approach to network model building for gentamicin peak predictions. J Pharm Sci. 1996;85:65–69. doi: 10.1021/js950271l. [DOI] [PubMed] [Google Scholar]
  • 19.Gobburu L, Kumar K, Shelver WH. Prediction of alfentanil plasma concentrations using artificial neural networks (ANN) Pharm Res. 1995;12:329–329. doi: 10.1023/A:1016283930696. [DOI] [PubMed] [Google Scholar]
  • 20.Brier ME. Empirical pharmacokinetic predictions for cyclosporine using a time series neural network. Pharm Res. 1995;12:363–363. doi: 10.1023/A:1016260720218. [DOI] [Google Scholar]
  • 21.Brier ME. Empirical pharmacokinetic predictions using neural network, generalized linear, and generalized additive models. Pharm Res. 1995;12:363–363. doi: 10.1023/A:1016260720218. [DOI] [Google Scholar]
  • 22.Brier ME, Aronoff GR. Application of artificial neural networks to clinical pharmacology. Int J Clin Pharm Ther. 1996;34:510–514. [PubMed] [Google Scholar]
  • 23.Smith BP, Brier ME. Empirical pharmacodynamic predictions using neural networks: Comparison to NONMEM. Pharm Res. 1995;12:328–328. [Google Scholar]
  • 24.Gobburu J, Chen EP. Artificial neural networks as a novel approach to integrated pharmacokinetic-; pharmacodynamic analysis. J Pharm Sci. 1996;85:505–510. doi: 10.1021/js950433d. [DOI] [PubMed] [Google Scholar]
  • 25.Kumar K, Gobburu JVS, Ballantyne JA, Shelver WH. Population pharmacokinetic parameter prediction with artificial neural networks (ANN) Pharm Res. 1995;12:329–329. doi: 10.1023/A:1016213721861. [DOI] [PubMed] [Google Scholar]
  • 26.Chow H-H, Tolle KM, Row DJ, Elsberry V, Chen H. Application of neural networks to population pharmacokinetic data analysis. J Pharm Sci. 1997;86:840–845. doi: 10.1021/js9604016. [DOI] [PubMed] [Google Scholar]
  • 27.Gobburu J, Shelver WH. Feasibility of artificial neural network application to derive in-vivo characteristics in a congeneric series of compounds. Pharm Res. 1995;12:364–364. [Google Scholar]
  • 28.Hussain AS. Artificial neural networks based in vitro-in vivo correlations. In: Young D, editor. In Vitro-;In Vivo Correlations. New York: Plenum Press; 1997. pp. 149–158. [Google Scholar]
  • 29.Dowell JA, Hussain AS, Stark P, Devane J, Young D. Development of in vitro-;in vivo correlations using various artificial neural network configurations. In: Young D, editor. In vitro-;in vivo correlations. New York: Plenum Press; 1997. pp. 225–239. [DOI] [PubMed] [Google Scholar]
  • 30.Hussain AS, Johnson RD, Vachharajani NN, Ritschel WA. Feasibility of developing a neural network for prediction of human pharmacokinetic parameters from animal data. Pharm Res. 1993;10:466–469. doi: 10.1023/A:1018917128684. [DOI] [PubMed] [Google Scholar]
  • 31.Ritschel WA, Akileswaran R, Hussain AS. Application of neural networks for the prediction of human pharmacokinetic parameters. Meth Find Exp Clin Pharmacol. 1995;17:629–643. [PubMed] [Google Scholar]
  • 32.Gobburu J, Shelver WH. Quantitative structure-pharmacokinetic relationship (QSPR) of beta blockers derived using neural networks. J Pharm Sci. 1995;84:862–865. doi: 10.1002/jps.2600840715. [DOI] [PubMed] [Google Scholar]
  • 33.Blakey GE, Nestorov IA, Arundel PA, Aarons LJ, Rowland M. Quantitative structure-pharmacokinetics relationships: I. Development of a whole-body physiologically based model to characterize changes in pharmacokinetics across a homologous series of barbiturates in the rat. J Pharmacokin Biopharm. 1997;25:277–312. doi: 10.1023/A:1025771608474. [DOI] [PubMed] [Google Scholar]
  • 34.Nestorov IA, Aarons LJ, Rowland M. Quantitative structure-pharmacokinetics relationships: II. A mechanistically based model to evaluate the relationship between tissue distribution parameters and compound lipophilicity. J Pharmacokin Biopharm. 1998;26:521–545. doi: 10.1023/a:1023221116200. [DOI] [PubMed] [Google Scholar]
  • 35.Sheiner LB, Beal S. Some suggestions for measuring predictive performance. J Pharmacokin Biopharm. 1981;9:503–512. doi: 10.1007/BF01060893. [DOI] [PubMed] [Google Scholar]
  • 36.ACSL Reference Manual, Edition 11. MGA Software, Concord MA 01742, USA, 1995.
  • 37.Nestorov I, Aarons L, Rowland M. Physiologically based pharmacokinetic modelling of a homologous series of barbiturates: a sensitivity analysis. J Pharmacokin Biopharm. 1997;25:413–447. doi: 10.1023/A:1025740909016. [DOI] [PubMed] [Google Scholar]
  • 38.Hadjitodorov S, Boyanov B, Dalakchieva N. A two-level classifier for text-independent speaker identification. Speech Communication. 1997;21:209–217. doi: 10.1016/S0167-6393(97)00004-6. [DOI] [Google Scholar]
  • 39.Neural Network Toolbox for use with MATLAB. The Mathworks Inc., Natick MA 01760-1500, USA, 1998.A.

Articles from AAPS PharmSci are provided here courtesy of American Association of Pharmaceutical Scientists

RESOURCES