Abstract The predictive performance of an artificial neural network (NN) was compared with the first-order multiple linear regression (MLR) using mean dissolution data of 28 diltiazem immediate release tablet formulations. The performance was evaluated using “Weibull” function parameters alpha and beta. Weibull parameters were used as dissolution markers of the eight principal, mainly compositional, variables. The parameters were obtained by fitting the Weibull function to the mean (n = 12) dissolution profiles of 28 diltiazem hydrochloride tablet formulations. The generated set of 28 pairs of Weibull function parameters was evaluated for internal and external predictability using both the MLR and the artificial NN. A three-layered 8-5-2 feedforward NN was found to be an adequate descriptor of the dissolution data. Internal predictions were based on the data of 24 products. External predictions used the 24 product data to test four products not used in the training phase. The predictive performances of the two techniques were evaluated using bias (mean prediction error; MPE) and precision (mean absolute error; MAE). The study results suggested that, for the studied data set, NN is a superior internal and external predictor to MLR. The artificial NN predicted order of the formulation composition variables, influencing the dissolution parameters as follows: hydrogenated oil > microcrystallinecellulose > ethyl cellulose > eudragit > hydroxypropylcellulose > coat > hydroxypropylmethyl-cellulose > Speed. Internal (alpha, beta)External (alpha, beta)MPE (%)MAE (%)MPE (%)MAE (%)MLR0, 012.5, 5.711.2, −7.118.1, 7.1NN0.3, 0.13.3, 1.7−0.7, −0.43.3, 0.9
[1]
Chee Peng Lim,et al.
Use of Artificial Neural Networks to Predict Drug Dissolution Profiles and Evaluation of Network Performance Using Similarity Factor
,
2000,
Pharmaceutical Research.
[2]
Lewis B. Sheiner,et al.
Some suggestions for measuring predictive performance
,
1981,
Journal of Pharmacokinetics and Biopharmaceutics.
[3]
Ajaz S. Hussain,et al.
Feasibility of Developing a Neural Network for Prediction of Human Pharmacokinetic Parameters from Animal Data
,
1993,
Pharmaceutical Research.
[4]
Kozo Takayama,et al.
Artificial Neural Network as a Novel Method to Optimize Pharmaceutical Formulations
,
2004,
Pharmaceutical Research.
[5]
Patrick Henry Winston,et al.
Artificial intelligence (2nd ed.)
,
1984
.
[6]
H. Penny Nii,et al.
The Handbook of Artificial Intelligence
,
1982
.
[7]
Nils J. Nilsson,et al.
Artificial Intelligence
,
1974,
IFIP Congress.
[8]
Sharon C. Salveter.
Review of The handbook of artificial intelligence, vol. 1 by Avron Barr and Edward A. Feigenbaum. William Kaufman, Inc. 1981.
,
1982
.
[9]
Robert D. Johnson,et al.
Application of Neural Computing in Pharmaceutical Product Development
,
1991,
Pharmaceutical Research.