Research Articles
Application of Neural Networks to Pharmacodynamics

https://doi.org/10.1002/jps.2600820910Get rights and content

Abstract

Neural networks (NN) are computational systems implemented in software or hardware that attempt to simulate the neurological processing abilities of biological systems. A synopsis is presented of the operational characteristics, structures, and applications of NN. The NN technology has primarily been aimed at recognition science (e.g., handwriting, voice, signal, picture, image, pattern, etc.). It is pointed out that NN may also be particularly suitable to deal with pharmacokinetic (PK) and pharmacodynamic (PD) systems, especially in cases such as multivariate PK/PD population kinetics when the systems are so complex that modeling by a conventional structured model building technique is very troublesome. The main practical advantage of NN is the intrinsic ability to closely emulate virtually any multivariate system, including nonlinear systems, independently of structural/physiologic relevance. Thus, NN are most suitable to model the behavior of complex kinetic systems and unsuitable to model the structure. In a practical sense, this structure limitation may be inconsequential because NN in its multivariate formulation may consider any physiologic, clinical, or population variable that may influence the kinetic behavior. The application of NN in PD is demonstrated in terms of the ability of an NN to predict, by extrapolation, the central nervous system (CNS) activity of alfentanil. The drug was infused by a complex computer-controlled infusion scheme over 180 min with simultaneous recording of the CNS effect quantified by a fast Fourier transform power spectrum analysis. The NN was trained to recognize (emulate) the drug input–drug effect behavior of the PD system with the input–effect data for the 180 min as a training set. The trained NN was then used to predict, by extrapolation, the CNS effect in the subsequent 120-min period that contained a complex infusion scheme to provide an extra challenge in testing how well the trained NN could predict the effect. The relative prediction performance of the NN was 66% (SD = 32, n = 6), which indicates an excellent prediction considering the intrinsic high degree of fluctuations in the effect variable. The NN was best in predicting the high intensity effects and also learned to recognize an overshoot phenomenon indicative of development of short-term tolerance.

References and Notes (16)

  • Veng-PedersenP. et al.

    J. Pharm. Sd.

    (1991)
  • Veng-PedersenP. et al.

    J. Pharm. Sci.

    (1992)
  • KohonenT.

    Neural Networks

    (1988)
  • MarenA.J.
  • CutlerD.J.

    J. Pharmacokinet. Biopharm.

    (1981)
  • Veng-PedersenP.

    J. Pharmacokinet. Biopharm.

    (1988)
  • Veng-PedersenP.

    J. Pharmacokinet. Biopharm.

    (1988)
  • VerottaD. et al.

    Am. J. Physiol.

    (1989)
There are more references available in the full text version of this article.

Cited by (60)

  • New application of bioelectrical impedance analysis by the back propagation artificial neural network mathematically predictive model of tissue composition in the lower limbs of elderly people

    2012, International Journal of Gerontology
    Citation Excerpt :

    Application of linear regression analysis on single independent variables and single dependant variables is suitable; however, multiple parameters, especially when interactions exist between the parameters, will result in errors in the predictive equation. Besides the most popular method for outcome prediction, artificial neural networks (ANNs)26–33, there are other mathematical methods such as logistic regression34, disscriminant analysis and recursive partitioning35. The ANN model has performed well, with greater precision and accuracy in the prediction of intercellular fluid and TBW in patients with chronic hemodialysis than that of the linear regression model36.

  • Using neural networks to determine the contribution of danshensu to its multiple cardiovascular activities in acute myocardial infarction rats

    2011, Journal of Ethnopharmacology
    Citation Excerpt :

    It has been shown that BP NN could approximate the PK and PD profiles generated from simulations of several structural PK/PD models (Gobburu and Chen, 1996). Furthermore, BP NN has been successfully applied in the bioequivalence study (Opara et al., 1999), population pharmacokinetics (Chow et al., 1997), clinical pharmacology (Brier and Aronoff, 1996; Urquidi-Macdonald et al., 2004; Mager et al., 2005) and PK/PD modeling (Veng-Pedersen and Modi, 1993; Haidar et al., 2002). Ischemic heart disease has been regarded as a complex disorder associated with numerous risk factors such as hypertension, diabetes and atherosclerosis (Ferdinandy et al., 2007).

  • Analysis of pellet properties with use of artificial neural networks

    2010, European Journal of Pharmaceutical Sciences
    Citation Excerpt :

    Superior abilities of ANNs as empirical modeling tools have attracted attention of scientists in the pharmaceutical field as well. ANNs were used as predictive models in pharmacokinetics and pharmacodynamics (Brier and Żurada, 1995; Brier and Aronoff, 1996; Chow et al., 1997; Hussain et al., 1993; Veng-Pedersen and Modi, 1992, 1993), in vitro–in vivo correlation (IVIVC) development (Dowell et al., 1999), and pharmaceutical technology as well (Hussain et al., 1994). Hussain et al. (1991) were the first to introduce ANNs to the field of pharmaceutical technology, pointing to the possible advantages of guided search for the optimal pharmaceutical formulation.

View all citing articles on Scopus
View full text