Multiobjectives optimization to optimal Moroccan diet

using genetic algorithm

International Conference on Intelligent Systems and Computer Vision
(ISCV 2022)

Ahourag Abdellah,Karim El Moutaouakil, Saliha Chellak, Hicham

Baizri, Mouna Cheggour

Engineering Science Laboratory, PFT Taza University Sidi Mohamed Ben Abdellah

May 19, 2022

Ahourag Abdellah (Engineering Science Laboratory)

Multiobjectives optimization to optimal Moroccan diet using
Maygenetic
19, 2022
algorithm 1 / 26
 Outline

1 Introduction

2 Multi-objective optimization for optimal diet problem

3 Resolution of the optimal diet model by genetic algorithm

4 Conclusion and Outlook

Ahourag Abdellah (Engineering Science Laboratory)

Multiobjectives optimization to optimal Moroccan diet using
Maygenetic
19, 2022
algorithm 2 / 26
 Introduction

Outline

1 Introduction

2 Multi-objective optimization for optimal diet problem

3 Resolution of the optimal diet model by genetic algorithm

4 Conclusion and Outlook

Ahourag Abdellah (Engineering Science Laboratory)

Multiobjectives optimization to optimal Moroccan diet using
Maygenetic
19, 2022
algorithm 3 / 26
 Introduction

Introduction

A healthy and balanced diet is essential for quality of life.

Because of the increasing number of patients suffering from lifestyle
diseases, it is important to control and design a balanced diet.
It is clear that foods with a high glycemic load are likely to unbalance
blood glucose levels.
In the long term, high consumption is associated with an increased
risk of type 2 diabetes and cardiovascular disease.
Good blood sugar control aims to prevent or delay the complications
of diabetes.

Ahourag Abdellah (Engineering Science Laboratory)

Multiobjectives optimization to optimal Moroccan diet using
Maygenetic
19, 2022
algorithm 4 / 26
 Introduction

The first diet problem was proposed by Stigler and solved by Danzig
as the first test problem for the simplex algorithm .
The first researcher to introduce a linear programme with glycaemic
load as an objective function was Esra Bas (2014).
In this work, we determine diets based on constrained multi-objective
linear programming where the first objective function is the glycemic
load of the diet while the second objective function is the cost of the
diet.

Ahourag Abdellah (Engineering Science Laboratory)

Multiobjectives optimization to optimal Moroccan diet using
Maygenetic
19, 2022
algorithm 5 / 26
 Multi-objective optimization for optimal diet problem

Outline

1 Introduction

2 Multi-objective optimization for optimal diet problem

3 Resolution of the optimal diet model by genetic algorithm

4 Conclusion and Outlook

Ahourag Abdellah (Engineering Science Laboratory)

Multiobjectives optimization to optimal Moroccan diet using
Maygenetic
19, 2022
algorithm 6 / 26
 Multi-objective optimization for optimal diet problem

Multi-objective optimization for optimal diet problem

Data :
In this work, we consider fifteen favorable nutrients and four unfavorable
nutrients.
Favorable: Calories(c), Protein(p), Carbohydrate(car),
Potassium(po), Magnesium(mg), Dietary fibre (tdf), Calcium(ca),
Iron (ir), Phosphorus(ph), Zinc(z), Vitamin b6(vb6), b12(vb12),
C(vc), A(va), and E(ve).
Unfavorble nutrients: saturated fatty acids (sf), Sodium,
Cholesterol, and Fat (tf).

Ahourag Abdellah (Engineering Science Laboratory)

Multiobjectives optimization to optimal Moroccan diet using
Maygenetic
19, 2022
algorithm 7 / 26
 Multi-objective optimization for optimal diet problem

Requirements:
1 The daily positive nutrient requirements:
bc = 2.000kcal, bp = 91g, bcar = 271g, bpo = 4044mg, bmg =
380mg, btdf = 31g, bca = 1316mg, bir = 18mg, bph = 1740mg, bz =
14mg, bvb6 = 2.4mg, bvb12 = 8.3µg, bvc = 155mg, bva =
1052µg, bve = 9.5mg,
2 the minimum tolerable negative nutrient requirement:
fsf = 17g, fs = 1.779mg, fch = 230mg, andftf = 65g.
Unknown:
Optimal quantities of each food that meets all requirements and with
minimal total glycemic load and cost.

Ahourag Abdellah (Engineering Science Laboratory)

Multiobjectives optimization to optimal Moroccan diet using
Maygenetic
19, 2022
algorithm 8 / 26
 Multi-objective optimization for optimal diet problem

Modeling the optimal regime problem

Symbols:
From now on, we adopt the following symbols:
A: matrix of the values of foods, positive nutrients; in our case,
(number of positive nutrients, number of foods)= (4, 177);
E : matrix of the values of foods negative nutrients; in our case,
(number of negative nutrients, number of foods)= (4, 177);
b: minimum required positive nutrients (number of positive nutrients,
1);
f : maximum tolerable negative nutrients (number of negative
nutrients,1);
Ac : line vector corresponding to the calories coming from the
positive nutrient;

Ahourag Abdellah (Engineering Science Laboratory)

Multiobjectives optimization to optimal Moroccan diet using
Maygenetic
19, 2022
algorithm 9 / 26
 Multi-objective optimization for optimal diet problem

ci : vector of the calories coming from the nutrient i of

{car , p, tf , sf };
ri : percentage of total calories of i nutrients from
rp = 18%, rcar = 55%,
rsf = 7.8% and rtf = 29%;
xj : the quantity of food j adopted in our diet. Thus, we define the
vector x = (xj )1≤j≤177 .
g: vector of glycemic loads generated by each food;
C = (C1 , ..., C177 )T : is a vector whose components represent the cost
of 100 gram of the 177 foods.

Ahourag Abdellah (Engineering Science Laboratory)

Multiobjectives optimization to optimal Moroccan diet using
Maygenetic
19, 2022
algorithm 10 / 26
 Multi-objective optimization for optimal diet problem

Constraints:
Constraints on favorable nutrients:
1 If we adopt the decision x = (xj )1≤j≤177 , Ax are the vector of 15 rows
each of which represents the total amount of each positive nutrient
offere by our diet, then we have the following family of constraints
Ax ≥ b.
2 The total calories from carbohydrate must make up at least 55% off
the calories from all positive diet,nutrients;and hence we have the
T x ≥ 0.55 AT x .


constraint ccar c
3 The total calories from Protein must be at least 18% of the calories
from all positive diet

nutrients; hence we have the constraint
cpT x ≥ 0.18 AT c x .

Ahourag Abdellah (Engineering Science Laboratory)

Multiobjectives optimization to optimal Moroccan diet using
Maygenetic
19, 2022
algorithm 11 / 26
 Multi-objective optimization for optimal diet problem

Constraints on unfavorable nutrients:

1 Ex is the vector of 4 rows each of which represents the total amount
of each negative nutrient offered by our diet. Since f represents the
total maximum of each negative nutrient, then we have the following
family of constraints Ex ≤ f .
2 The total calories from total fat must be at most 29% off the calories
from all positive diet,

nutrients; hence the constraint
ctfT x ≤ 0.29 ATc x .
3 The total calories from saturated fat must be at most 7.8% of the
calories from all positive diet nutrients; hence the constraint
csfT x ≤ 0.078 AT


c x .

Ahourag Abdellah (Engineering Science Laboratory)

Multiobjectives optimization to optimal Moroccan diet using
Maygenetic
19, 2022
algorithm 12 / 26
 Multi-objective optimization for optimal diet problem

Multi-objectives diet model

The least expensive mathematical model that minimizes the total glycemic
load and ensures the minimum necessary positive nutrients and limits the
consumption of negative nutrients is given by :

min g T x , min C T x (1)

A.x > b (2)
(P) s.t : E .x 6 f (3)
 
ciT x > pi AT
c x i ∈ {car , p} (4)
 
ciT x 6 qi AT
c x i ∈ {tf , sf } (5)
x ≥0 (6)

Where pcar = 0.55, pp = 0.29, qtf = 0.55, and qsf = 0.18

Ahourag Abdellah (Engineering Science Laboratory)

Multiobjectives optimization to optimal Moroccan diet using
Maygenetic
19, 2022
algorithm 13 / 26
 Resolution of the optimal diet model by genetic algorithm

Outline

1 Introduction

2 Multi-objective optimization for optimal diet problem

3 Resolution of the optimal diet model by genetic algorithm

4 Conclusion and Outlook

Ahourag Abdellah (Engineering Science Laboratory)

Multiobjectives optimization to optimal Moroccan diet using
Maygenetic
19, 2022
algorithm 14 / 26
 Resolution of the optimal diet model by genetic algorithm

Resolution of the optimal diet model by genetic algorithm

To calculate the different parameters (A, E, b, f, g, C) of the diet

multi-objectives model (P), the project team, collected almost 177
foods (available on the Moroccan marketwhile considering the
nutrients with positive (respectively negative) effects on the human
body (respectively negative) on the human body.

Table: Basic nutrients

Positive Nutrients Negative Nutrients
Calories, Protein, Carbohydrate Satured fat
Potassium, Phosphorus, Magnesium Sodium
Calcium, Iron, Phosphorus, Zinc Cholesterol
Vb6, Vb12 , VC, VA, VE Total fat

Ahourag Abdellah (Engineering Science Laboratory)

Multiobjectives optimization to optimal Moroccan diet using
Maygenetic
19, 2022
algorithm 15 / 26
 Resolution of the optimal diet model by genetic algorithm

For each 100g of each of the 177 foods, we estimate the content of
each of the nutrients, thus considered.
For example, the table shows the content in 100g of apricot of the
different nutrients (on the left the positive nutrients and on the right
the negative nutrients).

Figure: The content in 100g of apricot of the different nutrients

Ahourag Abdellah (Engineering Science Laboratory)

Multiobjectives optimization to optimal Moroccan diet using
Maygenetic
19, 2022
algorithm 16 / 26
 Resolution of the optimal diet model by genetic algorithm

In addition, we consulted several sites and documents to estimate

positive daily nutrient requirements .

Table: The positive nutrient requirements

Favorable Nutrient Minimum required
Calories 2000 kcal
Protein 91g
Carbohydrate 271g
Potassium 4044mg
Magnesium 380mg
Calcium 1316 mg
Iron 18mg
Phosphorus 1740mg
Zinc 14mg
Vb6 2.4mg
Vb12 8.3µg

Ahourag Abdellah (Engineering Science Laboratory)

Multiobjectives optimization to optimal Moroccan diet using
Maygenetic
19, 2022
algorithm 17 / 26
 Resolution of the optimal diet model by genetic algorithm

Table: The negative nutrient requirements

Nutrient negatives Maximum tolerable
Satured fat 17 g
Sodium 1779mg
Cholesterol 230mg
Total fat 65 g

The first objective function of the optimal diet problem is formed by

the blood glucose vector g.

Ahourag Abdellah (Engineering Science Laboratory)

Multiobjectives optimization to optimal Moroccan diet using
Maygenetic
19, 2022
algorithm 18 / 26
 Resolution of the optimal diet model by genetic algorithm

The table gives the glycemic load contents in 100g of apricot, apricot
Dry, Garlic, pineapple, bananas conserve, Artichaut, Asparagus,
Eggplant, Veal brain, Chestnut, Shrimp chips, Cabbage white, and
Sauerkraut.

Table: The content in 100g of some foods in Glycemic load

Name of the food Glycemic
low medium high
Apricot 5.13 5.13 5.13
Garlic 3.225 3.225 3.225
Pineaple 3.57 3.753 3.936
Ananas conserve 0 0.313 0.626
Artichaut 0.735 0.735 0.735
Asparagus 0.48 0.48 0.48
Eggplant 0.945 0.945 0.945
Cherry 29.88 29.88 29.88
Veal brain 0 0 0
Ahourag Abdellah (Engineering Science Laboratory)
Multiobjectives optimization to optimal Moroccan diet using
Maygenetic
19, 2022
algorithm 19 / 26
 Resolution of the optimal diet model by genetic algorithm

Concerning the second objective function, we collected the

information from the different Moroccan market; the following table
shows the price range for 15 foods.

Table: the price range of 16 foods

Name of the food Price
low medium high
Apricot 0.9 1,1 1,3
Garlic 4 6 8
Almond 6.9 20.4 33,9
Pineapple 3 3.4 3,8
Artichoke 0.8 1 1,2
Asparagus 0.5 0.64 0,78
Aubergine 0.2 0.5 0,8
Avocado 5 6.5 8
Baguette 0.26 0.67 1,08

Ahourag Abdellah (Engineering Science Laboratory)

Multiobjectives optimization to optimal Moroccan diet using
Maygenetic
19, 2022
algorithm 20 / 26
 Resolution of the optimal diet model by genetic algorithm

we use the genetic algorithm to solve the optimal diet problem

presented via our model (P).
In this context, different models are developed using multi-objective
techniques.

Figure: Pareto front of model glycemic minimum and minimum cost regime

Ahourag Abdellah (Engineering Science Laboratory)

Multiobjectives optimization to optimal Moroccan diet using
Maygenetic
19, 2022
algorithm 21 / 26
 Resolution of the optimal diet model by genetic algorithm

Table: Glycemic minimum and minimum cost regime

Name of the food regim
Eggplant 5.048
Cooked egg white 1.006
sauerkraut 1.867
zucchini 6.000
Soy milk 6.000
Lentil 6.000
chickpea 4.218
Salsify 2.120
tea 4.986
Lemon zest 7.926
Glycemic load of diet 82.9127
cost of diet 47.1626

Ahourag Abdellah (Engineering Science Laboratory)

Multiobjectives optimization to optimal Moroccan diet using
Maygenetic
19, 2022
algorithm 22 / 26
 Resolution of the optimal diet model by genetic algorithm

Consider different values of the glycemic load and the cost of different
foods, the estimated diets are almost equivalent.
Indeed, some foods are always present (eggplant, soya milk, lentils,
chickpeas and tea) and rare are the foods that are absent, such as
cooked zucchini.

Ahourag Abdellah (Engineering Science Laboratory)

Multiobjectives optimization to optimal Moroccan diet using
Maygenetic
19, 2022
algorithm 23 / 26
 Conclusion and Outlook

Outline

1 Introduction

2 Multi-objective optimization for optimal diet problem

3 Resolution of the optimal diet model by genetic algorithm

4 Conclusion and Outlook

Ahourag Abdellah (Engineering Science Laboratory)

Multiobjectives optimization to optimal Moroccan diet using
Maygenetic
19, 2022
algorithm 24 / 26
 Conclusion and Outlook

Conclusion and Outlook

1 The diets obtained generate acceptable glycemic loads and respect

the purchasing power of Moroccan citizens.
2 In many cases, minimum cost linear programming models do not take
into account the variability of nutrients in foods.
3 In mathematical programming, this variability of the model
parameters is dealt with in the framework of stochastic programming.
In this technique, the objective function or certain constraints are
treated in probabilistic terms.

Ahourag Abdellah (Engineering Science Laboratory)

Multiobjectives optimization to optimal Moroccan diet using
Maygenetic
19, 2022
algorithm 25 / 26
 Conclusion and Outlook

Ahourag Abdellah (Engineering Science Laboratory)

Multiobjectives optimization to optimal Moroccan diet using
Maygenetic
19, 2022
algorithm 26 / 26