PHARMACEUTICAL TECHNOLOGY SEMINAR

Faculty of Pharmacy
UMP, HCMC, Vietnam

GEA-NUS PPRL
Department of Pharmacy
NUS, Singapore

Pharmaceutical product formulation

and extraction process development
assisted by intelligent software systems
Giap V. Dang
Department of Pharma IT, Faculty of Pharmacy, UMP, HCMC, Vietnam

Ho Chi Minh City

30-31 May, 2011

Problems of interest
General background
Pharmaceutical product formulation
Extraction process development
Future prospects

Problems of interest
Product innovation
All pharmaceutical products have a limited life:
Obsolescence
Decline
Maturity
Growth
Initial development

Stategies for product innovation:

- substitution of a new product
- extention of life cycle

Problems of interest
Formulating knowledge
Ingredients (APIs, excipients)

Production facilities

Formulation

Products

Product specifications

Commercial factors

Problems of interest
Areas of pharmaceutical R&D
RULES

A. Optimization & Prediction

B. Rules generation
C. Simultation

A
DATA

C
THEORY
Adapted from: http://www. intelligensys.co.uk/ARG/rulegen.htm

Problems of interest
Artificial intelligence (AI)
Computational
intelligence

Neural
networks

Evolutionary
algorithms

Evolutionary
programming
Evolution
strategies

Genetic
programming

Fuzzy
systems

Genetic
algorithms

Problems of interest
A powerful R&D toolkit
Design of experiments
e.g. FormData *

Statistics

Experimentation
Neuro-fuzzy
logic
Rules generation
e.g. FormRules *

Neural networks &

Gene expression
programming
Optimization & Prediction
e.g. INForm *

* Intelligensys Ltd., Springboard Business Centre, Stokesley TS9 5JZ, United Kingdom

Problems of interest
General background
Pharmaceutical product formulation
Extraction process development
Future prospects

General background
Experimental design
Formulation designs (mixture designs)
Formulation variables:
- Type of ingredients
- Amount of ingredients
Process designs (factorial designs)
Process variables:
- Equiment parameters
- Manufacturing methods
Combined designs

The factor space

The combination of a formulation design with a process design.

General background
Experimental design
x1

x2

x3

...

y1

1
2
3
...
n

For example:
D-optimal design: 3^2 + 2^1, n = 14 (n = 16).
Taguchi OA: 4^5, n = 16 (1024)

y2

y3

...

General background
Neurofuzzy logic

Abstraction

Knowledge discovery
D.
C.
B.
A.

D
C
B
A
Amount

Wisdom
Knowledge
Deep
Shallow (Declarative)
Information
Data

Adapted from: Rowe R. C. and Roberts R. J.

Intelligent Software for Product Formulation,
Taylor & Francis, UK (1998).

FormRules the strengths of neural networks with those of

neurofuzzy logic, has been applied to knowledge discovery or data
mining (automatic rule induction).

General background
Neurofuzzy logic

Input space

Weight vector

Associative Memory Networks

Multivariate
membership functions

Network
output

As a grey box

General background
Neurofuzzy logic
Structure of generated rules
Deep knowledge = full understanding (exactly BECAUSE)
Shallow knowledge = rule expression (perhaps WHY)
The antecedent
IF (Condition 1) AND (Condition 2) THEN (Observed behaviour)
The consequent
The complexity of rule statements depends on the fuzzy set inputs.

General background
NN - GEP combination
Modelling and optimization
Modeling: What if?
Product
properties

Formulation

Optimization: How to get?

INForm:

Adapted from: http://www. www.intelligensys.co.uk/info/industrial.htm

Uses neural networks (NN) to develop cause-effect models for

formulation optimization and what if prediction.

Combines gene expression programming (GEP) with fuzzy logic
to build its optimizer that is intuitive and easy to use.

General background
NN - GEP combination
Multilayer perceptron (MLP) networks
W1
NET
= XW

W2

X
W3

F
MLP

X = Input
Y = Output
W = Weight
F = Transfer functions: Linear, Sigmoid & Tanh

As a black box

General background
NN - GEP combination
Desirability functions

UP

TENT

FLAT

DOWN

FLAT TENT

Problems of interest
General background
Pharmaceutical product formulation
Extraction process development
Future prospects

Pharmaceutical product formulation

In co-operation with:
Prof. Dr. Quan-Nghiem Le and his co-workers
Department of Industrial Pharmacy
Assoc. Prof. Dr. Hoa V. Huynh and his co-workers
Department of Pharmaceutics

Pharmaceutical product formulation

A concept of quality

GMP

Research &
Development

Formulation and
process development

Clinical trials

...Quality assurance is a wide-ranging concept covering all matters

that influence the quality of a product. It therefore incorporates GMP
and other factors such as product design and development (WHO).

Pharmaceutical product formulation

Cause-and-effect relationships
Ingredients

Formulation

Product properties
Processing conditions

Independent variables (xi) = Causes

Ingredients: type, amount (or concentration), etc.
Processing conditions: methods, equipment, etc.
Dependent variables (yi) = Effects
Physico-chemical properties, in vitro profiles,
in vivo responses, costs, etc.

Multivariate

Multiresponse

Pharmaceutical product formulation

Artificial tears
Research objectives
Investigated ingredients:
Hydroxypropyl methylcellulose (HPMC)
Sodium chloride (NaCl)
Calcium chloride, anhydrous
Benzalkonium chloride
0.05 M Sodium borate solution
0.2 M Boric acid solution
qs.

x1 g
(5 15)
x2 g
(2 4)
0.067 g
0.1 g
x3 ml (50 150)
1000 ml

Desired properties:
y1: Viscosity (mPas): Max
y2: Reflectivity: Min
y3: Osmolality: 296 304 mOsm/ L
y4: pH value: 6.8 7.2

Up
Down
Flat Tent
Flat Tent

Pharmaceutical product formulation

Artificial tears
Experimental data
1
2
3
4
5
6
7
8
9
10
11
12
13
14

x1
15
5
5
10
15
5
15
15
5
15
10
10
5
10

x2
4
3
2
3
3
2
3
2
3
4
3
4
4
2

x3
150
150
50
150
50
50
50
150
150
50
50
50
150
150

y1
11.97
5.56
5.50
7.92
11.87
5.62
11.93
11.93
5.50
12.00
7.90
7.90
5.58
7.86

y2
1.3369
1.3352
1.3352
1.3359
1.3369
1.3353
1.3369
1.3369
1.3352
1.3369
1.3360
1.3359
1.3353
1.3359

y3
342.4
331.6
283.4
341.4
330.2
283.4
330.2
289.1
331.5
342.2
291.1
342.3
342.8
283.6

y4
7.56
7.12
6.78
7.55
6.79
6.78
6.80
7.52
7.03
6.83
6.81
6.82
7.54
7.52

Notes:
x1: HMPC (g)
x2: Sodium chloride (g)
x3: 0.05 M Borate soln (ml)
y1: Viscosity (mPas)
y2: Reflectivity
y3: Osmolality (mOsm/ L)
y4: pH value

Pharmaceutical product formulation

Artificial tears
Trends of relationships
Viscosity

x1

HMPC

x2
x3

x1
x2
x3

NaCl
Osmolality

y1
y2
y3
y4

x1

y1
y2
y3
y4

x1

HMPC

Reflectivity

x2
x3

x2
x3

Borate soln

y1
y2
y3
y4

y1
y2
y3
y4
pH value

Pharmaceutical product formulation

Artificial tears
Strenghts of relationships

Product
properties

Ingredient variables

Train R-square
values

x1

x2

x3

y1

97.9910

y2

98.8581

y3

75.9369

y4

80.2844

Structural Risk Minimisation

Pharmaceutical product formulation

Artificial tears
Generated rules
IF x1 is LOW THEN y1 is LOW (1.00)
IF x1 is HIGH THEN y1 is HIGH (0.95)
IF x1 is LOW THEN y2 is LOW (1.00)
IF x1 is HIGH THEN y2 is HIGH (0.98)
IF x2 is LOW THEN y3 is LOW (0.89)
IF x2 is HIGH THEN y3 is HIGH (1.00)
IF x3 is LOW THEN y4 is LOW (0.97)
IF x3 is HIGH THEN y4 is HIGH (0.80)

Pharmaceutical product formulation

Artificial tears
2-D graphs

12.0
11.0
10.0
9.0
y1

8.0
7.0
6.0

x1
5

10

11

12

13

14

15

1.3369
1.3367
1.3365
1.3363
y2 1.3361
1.3359
1.3357
1.3355
1.3353

x1
5

10

11

12

13

14

15

Pharmaceutical product formulation

Artificial tears
Optimized formulation
Hydroxypropyl methylcellulose (HPMC)
Sodium chloride (NaCl)
Calcium chloride, anhydrous
Benzalkonium chloride
0.05 M Sodium borate solution
0.2 M Boric acid solution
qs.

10.35 g
2.67 g
0.067 g
0.1 g
114.9 ml
1000 ml

Experimental verification
Viscosity
Reflectivity
Osmolality
pH value

OBS1
11.24
1.3357
302.2
6.90

OBS2
11.17
1.3360
301.4
6.92

OBS3
11.22
1.3359
301.7
6.94

Mean
11.21
1.3359
301.8
6.92

PRED
11.35
1.3358
303.7
6.98

Pharmaceutical product formulation

Some recently published articles
1. Dang-Thoai Nguyen, Minh-Chau Hoang; Giap V. Dang. Formulation of
Gilanka capsules containing Gingko biloba standardized extract for
single daily administration. HCM City Journal of Medical Sciences,
Supplement, 15 (1), 56-60 (2011).
2. Nhat-Hung Phan, Kha V. Nguyen; Giap V. Dang. Formulation of
Perindopril erbumine 4 mg tablets, using intelligent software as a frame
work. HCM City J. Med. Sci., Supplement, 15 (1), 61-65 (2011).
3. Viet-Phuong T. Nguyen, Quan-Nghiem Le; Giap V. Dang. Cause-effect
study on the formulation of Loratidine 10 mg fast-disolving tablets.
Vietnam Journal of Pharmaceutical Sciences, 49 (401), 11-13 (2009).
4. Cam-Vy T. Vo, Quan-Nghiem Le; Giap V. Dang. Design and
optimization of the formulation for Gliclazide 30 mg SR tablets. Vietnam
Journal of Pharmaceutical Sciences, 361, 21-23 (2006).
5. Cam-Vy T. Vo, Quan-Nghiem Le; Giap V. Dang.. Cause-and-effect
study on the formulation of Gliclazide 30 mg SR tablets. Vietnam
Journal of Pharmaceutical Sciences, 363 11-13 (2006).

Problems of interest
General background
Pharmaceutical product formulation
Extraction process development
Future prospects

Extraction process development

In co-operation with:
Prof. Dr. Minh-Duc Nguyen and his co-workers
Department of Pharmacognosy

Extraction process development

GxP for mecidinal plants

GACP

GEP
GPP/ GMP

GACP: Good Agricultural and Collection Practices

GEP: Good Extraction Practices
GPP: Good Processing Practices
GMP: Good Manufacturing Practices

Extraction process development

Cause-and-effect relationships
Materials

Process

Product properties
Extracting conditions

Independent variables (xi) = Causes

Materials: sources, parts of use, solvents, etc.
Extraction conditions: methods, equipment, etc.
Dependent variables (yi) = Effects
Extraction yield (%), active substance(s),
unwanted substance(s), costs, etc.

Multivariate

Multiresponse

Extraction process development

Phyllanthus amarus extraction process
Research objectives
Material
Extraction
Fluid extract

x1 = EtOH
(Low, Mid, High)
x2 = Solvent/ Material ratio (9, 12, 15)
x3 = Extraction times
(2, 3)

Evaporation
Concentrate
Spray drying
Dried extract
y1: Max

Purification
Residue

CHCl3

Phyllanthin
y2: Max

Extraction process development

P. amarus extraction process
Experimental data
1
2
3
4
5
6
7
8
9
10
11
12
13
14

x1
Mid
Mid
Mid
Low
Low
Low
Mid
High
High
Low
Mid
High
Low
High

x2
15
9
15
9
12
9
9
15
9
15
12
12
12
15

x3
3
2
2
2
3
3
3
2
3
2
2
3
2
3

y1
7.21
6.34
6.92
5.50
6.30
6.06
6.89
6.02
5.89
6.10
6.71
6.12
5.88
6.32

y2
1.55
1.36
1.28
0.67
0.55
0.63
1.07
4.01
3.04
0.47
1.91
3.95
0.52
2.66

Notes:
x1: Ethanol concentration
x2: Sovent/material ratio
x3: Extraction times
y1: Extraction yield (%)
y2: Phyllanthin content (%)

Extraction process development

P. amarus extraction process
Trends of relationships
Extraction yield

x1
All

x2

y2

x3

x1
x2
x3

y1

EtOH concentration

y1
y2
Phyllanthin content

Extraction process development

P. amarus extraction process
Strenghts of relationships
Process variables

Extract
properties

x1

x2

x3

y1

98.6346

y2

91.2376

Minimum Description Length

Train R-square
values

Extraction process development

P. amarus extraction process
Generated rules
Related to the extraction yield:
IF x1 is LOW THEN y2 is LOW (1.00)
IF x1 is MID THEN y2 is HIGH (1.00)
IF x1 is HIGH THEN y2 is LOW (1.00)
IF x3 is HIGH THEN y2 is HIGH (0.95)
IF x2 is 1:9 THEN y2 is LOW (1.00))
Related to the phyllanthin content:
IF x1 is LOW THEN y1 is LOW (0.97)
IF x1 is HIGH THEN y1 is HIGH (0.83)

Extraction process development

P. amarus extraction process
3-D visualization
y1
6.884
6.601
6.317
6.034
5.750
5.467
0.0

0.4

0.4
0.8

1.2

1.6

x1

2.0

2.0

1.6

1.8
x2

0.8

0.0

Extraction process development

P. amarus extraction process
Optimized parameters
x1 = Ethanol concentration = Mid
x2 = Solvent/ material ratio = 15
x3 = Extraction times = 2
Experimental verification
OBS

PRED

Mean

Extraction yield (%)

6.18

5.98

6.09

6.09

6.38

Phyllanthin (%)

2.73

2.65

2.40

2.40

2.34

Extraction process development

Some recently published articles
1. Linh-Tuyen T. Nguyen, Minh-Duc Nguyen; Giap V. Dang. Optimization
of the extraction procedure for Houtuynia cordata Thunb. HCM City
Journal of Medical Sciences, Supplement, 15 (1), 551-554 (2011).
2. Duc-Hanh Nguyen, Minh-Duc Nguyen; Giap V. Dang. Development of
the extraction procedure for Curcuma longa L. HCM City Journal of
Medical Sciences, Supplement, 14 (1), 140-144 (2010).
3. Hong-Cuc T. Le, Minh-Duc Nguyen; Giap V. Dang. Cause-effect study
and multivariate optimization for extraction of Noni fruits. Vienam
Journal of Medicinal materials, 16 (2), 85-89 (2010).
4. Duc-Hanh Nguyen, Minh-Duc Nguyen; Giap V. Dang. Development of
the extraction process for Phyllanthus amarus Schum & Thonn. HCM
City Journal of Medical Sciences, Supplement, 13 (1), 263-267 (2009).

Problems of interest
General background
Pharmaceutical product formulation
Extraction process development
Future prospects

Future prospects
Formulation strategic solutions
Solving formulation challenges
-

Which variables affect each product property?

What rules govern the drug delivery system?
How to optimize formulations to meet their desired goals?
How changing ingredients affects product properties?

Minimizing possible mistakes

- Measuring unnecessary variables, or missing critical ones.
- Using ingredients having little impact on the properties.
- Spending time for exploration of unfruitful ideas.

Future prospects
Formulation development
Ingredients

Formulation

Product properties
Processing conditions

Pharmaceutics

Neutraceutics/ Dietetics

Cosmetics

Future prospects
Process development
Materials

Process

Product properties
Processing conditions

GMP

Pharmaceutical
manufactuting

GEP
GPP
GMP
Extraction of
natural products

GLP
Pharmaceutical
analysis