Transcript A Hybrid Expert System-Neural Network for Capsule Formulation

A Hybrid Expert System-Neural Network (“Expert Network”) for Capsule Formulation Support
1Gunjan
2Mintong
1Yun
2Larry
Kalra,
Guo,
Peng,
L. Augsburger
University of Maryland, Department of Computer Science and Electrical Engineer, Baltimore County; 2 University of Maryland, School of Pharmacy, Baltimore
Introduction
GUI: interface
The objective was to construct a prototype
intelligent hybrid Prototype Expert Network
(PEN) for capsule formulation, which may yield
formulations meeting specific running and drug
delivery performance design criteria for BCS II
drugs. To that end, a rule-based expert system
(MES) was developed to specifically address
BCS Class II drugs and integrated with a neural
network (NN). These two components, which
comprise the decision module and the prediction
module, respectively, are connected together by
two information exchange paths to form a loop.
The system is believed to have the power to
design a suitable capsule formulation to meet
both requirements of quality control and
dissolution.
C functions
Prolog Engine
BCS II
N
CAPEX
Y
CU: calculate PS to meet contentuniformity limit
OM: if PS is small, add
diluent and use blend style
Conclusion
Training Data Set
INDEPENDENT VARIABLES
%Lactose
Wetting
Filler Particle Disintegrant Lubricant
in Lactose/ Disintegrant Agent
Run No. Size(um)
Type
Level (%) SSF/MS MCC Blend Level(%) SLS (%)
1
100
Explotab
1
0
50
5.0
0.5
7
100
Explotab
0.5
0
100
5.0
1
8
100
Explotab
0.5
0
0
5.0
1
33
60
Ac-Di-Sol
0.3
0
100
4.0
0.2
34
60
Ac-Di-Sol
0.6
50
0
8.0
0.6
37
60
Explotab
0.6
100
100
12.0
0.2
These are representative of a total of 62 runs that were used in training.
13
100
Explotab
1.5
0
100
5.0
0
14
100
Explotab
1.5
0
0
5.0
0
Final Formulation: calculate capsule size,
% excipients, and final formulation
18.0
18.0
63.76
40.25
77.02
60.82
79.49
65.48
Preliminary results indicate that the PEN is a working system.
Good predictive power of the NN module requires sufficient
training samples and a cross validation process.
Further research will be directed toward:
• Validation and refinement of PEN
• Automation of the parameter adjustment as a process of
optimization.
• Generalization of PEN to other drugs in BCS Class II.
Acknowledgement
Input Package
Prediction Engine
N
N
User:
Acceptable?
Parameter
Adjustment
Y
Y
Final formulation
Permeability
> 0.0004?
BCS -I or III
Predicted dissolution rate
for the current formulation
Materials and Method
MES
Prediction
Control
This work is being supported by Capsugel. We also gratefully
acknowledge Pfizer Central Research for the gift of piroxicam.
Dose/Sol
>250?
BCS -II
CAPEX
A rule-based expert system was developed in
Prolog by followed the decision procedures
in the flow chat, and integrated with a neural
network (NN). These two components, which
comprise the decision module and the
prediction
module,
respectively,
are
connected together by two information
exchange paths to form a loop.
3.31
1.61
DF: choose excipients types
Reformulate
Microcrystalline cellulose (Avicel PH102
(FMC), Emcocel 90M (Penwest)), anhydrous
lactose (direct tableting grade, Quest
International), piroxicam (donated from
Pfizer), magnesium stearate, Explotab
(Penwest), Ac-Di-Sol ( FMC) and sodium
lauryl sulfate have been used in the study. An
instrumented Zanasi LZ-64 was used for the
encapsulation process, and the compression
force was maintained at 100 ~ 200N to
achieve the specific target weight. The plug
height was adjusted at 14mm. The
dissolution testing was conducted on a
Vankel 5000 dissolution station, and
followed the USP procedure. The percentage
dissolved in 10, 30 and 45 minutes were
recorded as the measurements for the
dissolution rate. Sixty-three batches have
been generated to train and validate the
system.
RESPONSE
Surface lubricant % dissolved % dissolved % dissolved
(m^2)/g BlendTime in 10 min
in 30 min
in 45 min
(min.)
10
30
45
2.46
10.0
66.09
79.86
84.08
1.61
2.0
68.04
81.85
84.03
3.31
2.0
53.52
78.81
87.79
2.77
3.0
67.50
96.19
99.60
2.77
3.0
48.85
85.28
96.95
2.77
3.0
73.33
95.32
101.39
N
BCS -IV
SSM
Y
Dose
compute
ANN
result
Low < 50mg
Mod 50-100mg
High 100-1000mg
V. high >1000mg
Results and Discussion
An expert system (MES) in the decision module
(based on a decision tree modeled after the Capsugel
Expert System1 [CAPEX]) was developed to provide
decision rules for formulation recommendation. The
NN in the prediction module (using backpropagation
learning) was developed to provide predictive
capability for the expected outcomes of the
recommended formulation. The NN was trained with
experiment data to capture the causal associations
between the formulation and the outcome. The
training was conducted with two experimental
datasets using piroxicam as a model drug. The
datasets represent two response surface designs for
the capsule formulation which were developed to
reflect the mapping from such variables as filler
type/ratio, lubrication systems, drug particle
size/specific surface area, disintegrants and
surfactants to dissolution of the model compound.
The capsules were filled using dosator-type
automatic filling machines.
1
S. Lai, F. Podcek, J.M. Newton, and R. Daumesnil. An expert system to aid the
development of capsule formulations. Pharm. Tech. Eur., 8:60-65 (1996).
CU Module
Eval_HalfDose?
Using the given equation to
calculate required PS
to achieve required tolerance
Y
N
CAPEX
Fair
Bad
DF Module
Y
Dose
 d
100
%CV 
 100
N
 6D
N
3



Choose Glidant
N
Large or
V. Large
Low or Medium
Granulate
Y
Compute Carr’ Index
Lubricant
CAPEX
Choose Lubricant
New PS
PS < 10 um
N
Wettable
?
Drug
User Input: Bulk density of dose
Y
OM Module
Ask Mixing Style
from user ?
Compute Capsule Size
Liquid Addition
Dose
Volume
Poorly Soluble
250-1000ml
4% Sodium starch glycolate
Croscarmellose
CAPEX
N
Wetting Agent
Sodium Lauryl Sulfate
Choose Disintegrant
Y
Interactive Physical Blending
Adhere to
metal?
1/ 2
Use New PS
Old PS
Good
V. Good
Flowability
Insoluble
>1000ml
8% Sodium starch glycolate
Croscarmellose
Drug
Diluent needed?
Dose Volume > 1000
Choose Diluent
>150μm
Ask User
for new
particle size
Particle
Size
>1050μm
>50150μm
Diluent M-PS
Computer Carr Index
Dose Volume
>1000mL
Ask User
for new
particle size
Diluent MPS for OM
Diluent MINsol for OM
User Input: tapped & bulk density of OM
>250-1000mL
Diluent F-PS
Particle
Size
>1050μm
Diluent F-Insol
Test for DF (Plug Formation)
>150μm
>50150μm
Diluent M-InSol