Genetic Algorithm for Scheduling Diet

Case: Liver Patient

M. F. Syahputra

, R. F. Rahmat

, Dian Aria Ningsih

, U. Andayani

Department of Information Technology, Universitas Sumatera Utara , Jl. Alumni no.3, Medan, Indonesia

Keywords: genetic algorithm, diet schedule, liver patient.

Abstract: The main function of liver is to process the absorbed nutrients from food consumed. Several factors that affect

its functions are alcohol, infection, drugs or herbal medicine, cancer, autoimmune disease, etc. Balanced and

healthy diet can help the liver patients to heal the disease by considering the calorie needs. In this research,

we schedule diet for liver patients by using genetic algorithm, which starts from calculating patient’s calorie

needs and then initializing the population, calculating fitness score, performing crossover and mutation. Next,

7 individuals with the lowest score will be selected as a diet schedule in a week (7 days). From the 8 times of

testing results, we get 0.5 lowest fitness score with 20 initial population and 100 generations. It shows that

the greater the created generation, the more the opportunities to obtain best individual with fitness score

approaching 0 (zero) or equal to 0 (zero).

1 INTRODUCTION

The liver is the largest glandular organ that serves to

process the absorbed nutrients from food consumed

and also to keep the body pure of toxins and harmful

substances (Dixit, 2014). Several factors that affect its

functions are alcohol, infection, drugs or herbal

medicine, cancer, autoimmune disease, etc. The most

common diseases that causes of chronic liver failure

include hepatitis, cirrhosis, hemochromatosis, etc.

One of the treatment able to help the liver patients

to cure the disease is balanced and healthy diet, by

considering the calorie needs. It aims to prevent the

fatal liver failure, to repair the damaged liver tissue,

to reduce its workload, to meet nutrition of patients

and to prevent the complication of liver failure

(Kargulewicz, 2014). Food consumed by liver patient

needs specific nutrient levels, such as protein, energy

and fat in a certain amount. Therefore, diet scheduling

system is required for liver patients to help them meet

the nutrient level and cure the disease more rapidly,

and also to inform them various food menu able to be

consumed so that it can increase the appetite of

patients.

Several method had been proposed for dietary

recommender system. The latest research was done

by Norouzi, et al. in 2016 who narratively reviewed

food recommender system for a diabetic patients

(Norouzi, 2016). They recognized three common

methods for food recommender system, namely

Collaborative Filtering Recommender System

(CFRS), knowledge based recommender system

(KBRS), and content aware recommender system

(CARS). Artificial intelligence method and semantic

web techniques are also often used in food

recommender systems.

In (El-Dosuky, 2012) used ontology and

heuristics for building food recommender system.

They concluded that their proposed algorithm could

give better result of F-measure, recall, precision and

accuracy than the other existing semantic

recommenders (El-Dosuky, 2012). But, due to the

complexity of the ontology construction, many

researchers prefer to use artificial intelligence

techniques. In the next year, Chen, et al. constructed

a diet recommendation system using fuzzy rules and

knapsack method. They designed a diet

recommendation system which has the expert

knowledge of three high chronic diseases, namely

diabetes, hypertension and cholesterol. The system

recommended suitable diet based on six parameters,

namely height, weight, activity levels, kidney

function, hypertension and hyperlipidaemia. They

also stated that their system had been evaluated by the

nutritionist to prove that it is effective (Chen, 2013).

Recently, genetic algorithm (GA) is often used for

solving optimization problems, such as shortest path

Syahputra, M., Rahmat, R., Ningsih, D. and Andayani, U.

Genetic Algorithm for Scheduling Diet Case: Liver Patient.

DOI: 10.5220/0010091919051909

In Proceedings of the International Conference of Science, Technology, Engineering, Environmental and Ramiﬁcation Researches (ICOSTEERR 2018) - Research in Industry 4.0, pages

1905-1909

ISBN: 978-989-758-449-7

1905

and also scheduling system (Kumar, 2010). In 2013,

Skinner, et al. presented genetic algorithm based

optimization approach to improve container handling

operations at a container terminal in Australia. The

experimental result of their research showed that GA

can reduce the overall time for container transfers and

improve the overall performance (Skinner, 2013).

Another related research in 2014, Filho, et al. solved

scheduling problems on flexible manufacturing

systems (FMS) using genetic algorithm (Filho, 2014).

Next research, Xu, et al. also used GA for task

scheduling on heterogeneous computing systems

using multiple priority queues. They concluded that

GA can cover a larger search space than the

deterministic scheduling approaches and do not lead

to high computational cost (Xu, 2014).

In this research, we propose genetic algorithm to

schedule diet for liver patients. The rest of the paper

will be organized as follows: Section II describes the

proposed method. Results and discussions are

presented in Section III. Section IV provides

summary and suggestions for future research.

2 CONCEPT DEVELOPMENT

Here are the steps in implementing the GA to

calculate the calorie needs:

2.1 Data Collection

This data of the calorie needs consist of two

parameter, i.e., liver patient and food nutrition. The

liver patient parameters, i.e., age, sex, weight, height,

allergy, ability of oral intake and edema. The food

nutrition parameters are food name, type of food

(staple food, side dishes, vegetables, fruits, and

complement), energy (kcal), protein (gr), fat (gr) and

carbohydrate (gr) and the ingredients data in 100

grams obtained from Nutri Survey Indonesia in 2007.

2.2 Calculation of Calorie Needs

The calorie needs can be calculated using Harris-

Benedict equation which estimates resting energy

expenditure (REE). The total calorie needs for liver

patients is called by basal calorie that used to create a

chromosome that consists of food menu.

Here, the following steps to calculate the calorie

needs (basal calorie).

Step 1 : Calculating the ideal body weight (kg)

based on the formula rule that showing in Table 1.

Table 1: Rule of Ideal Body Weight Calculation

Sex Height

Formula

Male < 160 cm

(Height-100) x 1 kg

Female < 150 cm

Male ≥ 160 cm

((Height-100) x 1 kg) x

90%

Female ≥ 150 cm

Step 2 : Calculating Basal Calorie (kcal)

based on rule in Table 2.

Table 2: Rule of Basal Calorie Calculation

Sex

Formula

Male

Ideal Body Weight x 30 kcal

Female

Ideal Body Weight x 25 kcal

2.3 Gen and Chromosome Structure

In genetic algorithm, an individual consists of gen and

chromosome. In this subsection, we will discuss

about the structure of gen and chromosome.

2.3.1 Gen

It consists of food types that will be initialized to

create initial population in genetic algorithm. In this

research, there are 5 genes (kind of food), namely

staple food (st), side dishes (sd), vegetables (v), fruits

(f) and complement (c). Each of them has two values,

namely calorie and food weight. The value in each of

gene is called by allele. Fig. 1 shows the structure of

gen.

Figure 1: Gen structure

2.3.2 Chromosome

In this research, there are three chromosomes in each

of individual, namely breakfast, lunch and dinner.

Each of chromosome has 5 genes, so there are 15

genes in an individual. It also has two values, namely

fitness score and the total of calorie. The total of

calorie is the calorie of food menu in a chromosome

for scheduling diet in a day. The fitness score is used

to evaluate the generated food menu based on the

patient calorie needs. Fig. 2 shows the structure of

chromosome.

ICOSTEERR 2018 - International Conference of Science, Technology, Engineering, Environmental and Ramiﬁcation Researches

1906

Figure 2: Chromosome structure

2.4 Initialization of Population

In this research, population is the data of food menu.

The greater the size of population, the more the

generated diet variation. Several steps below are done

to initialize the population.

Step 1 : Normalize food menu which

aims to adjust the generated diet

2.4.1 Divide the Total of Calorie Needs

Total of calorie needs is divided into 3 chromosomes,

namely breakfast, lunch and dinner. It aims to make

the diet schedule consider the balance of total calorie

for breakfast, lunch and dinner. It is done as

formulated in equation (1)

𝑇𝑜𝑡𝑎𝑙 𝑐𝑎𝑙𝑜𝑟𝑖𝑒 𝑝𝑒𝑟 𝑐ℎ𝑟𝑜𝑚 

   



(1)

2.4.2 Calculate Total Calorie for Every Kind

of Food

Each of chromosome consists of 5 genes of food type,

namely staple food, side dishes, vegetables, fruits,

and complement, which respectively has a weight of

0.45, 0.2, 0.15, 0.15, and 0.2. It can be calculated

using equation (2) until equation (6).

𝑠𝑡𝑎𝑝𝑙𝑒 𝑓𝑜𝑜𝑑  0.45 𝑥 𝑡𝑜𝑡𝑎𝑙 𝑐𝑎𝑙𝑜𝑟𝑖𝑒 𝑝𝑒𝑟 𝑐ℎ𝑟𝑜𝑚 (2)

𝑠𝑖𝑑𝑒 𝑑𝑖𝑠ℎ𝑒𝑠  0.2 𝑥 𝑡𝑜𝑡𝑎𝑙 𝑐𝑎𝑙𝑜𝑟𝑖𝑒 𝑝𝑒𝑟 𝑐ℎ𝑟𝑜𝑚 (3)

𝑣𝑒𝑔𝑒𝑡𝑎𝑏𝑙𝑒𝑠  0.15 𝑥 𝑡𝑜𝑡𝑎𝑙 𝑐𝑎𝑙𝑜𝑟𝑖𝑒 𝑝𝑒𝑟 𝑐ℎ𝑟𝑜𝑚 (4)

𝑓𝑟𝑢𝑖𝑡𝑠  0.15 𝑥 𝑡𝑜𝑡𝑎𝑙 𝑐𝑎𝑙𝑜𝑟𝑖𝑒 𝑝𝑒𝑟 𝑐ℎ𝑟𝑜𝑚 (5)

𝑐𝑜𝑚𝑝𝑙𝑒𝑚𝑒𝑛𝑡  0.2 𝑥 𝑡𝑜𝑡𝑎𝑙 𝑐𝑎𝑙𝑜𝑟𝑖𝑒 𝑝𝑒𝑟 𝑐ℎ𝑟𝑜𝑚 (6)

2.4.3 Calculate Each Food Weight in 100

Gram

Each of food in database is expressed in units of 100

grams. In the database, there are a few of menu that

has calorie more than needed calorie for the food. To

generate the number of calories and weight that

corresponds to the total calorie needs of DM patient,

we do the calculation as formulated in equation (7)

𝐹𝑜𝑜𝑑 𝑤𝑒𝑖𝑔ℎ𝑡



𝑔𝑟





 

 

𝑥100 (7)

Step 2 : Random and generate 15 food

menu based on its type as an initial population.

Step 3 : Input the calorie of food into

chromosome. The generated food menu from

previous step will be divided into 3 chromosomes,

namely breakfast, lunch and dinner, so each of

chromosome has 5 genes which contain the calorie of

staple food, side dishes, vegetables, fruits, and

complement. The number of generated population is

based on the number of initial individual and

generation.

2.5 Calculation of Fitness Score

Fitness score is used to evaluate the generated

solution and compare among the individuals to

determine the better or the worse one. In this research,

the fitness score for each individual is based on the

total of patient calorie needs and the total of food

calorie in a day. An optimal solution for an individual

is if the difference between the total of patient calorie

needs and the total of food calorie in a day equal or

approach to 0 (zero). Therefore, the less the generated

fitness score, the more optimal the individual. Fitness

score can be calculated using equation (8), while the

average of fitness score for each generation can be

calculated using equation (9).

𝐹𝑖𝑡𝑛𝑒𝑠𝑠 𝑠𝑐𝑜𝑟𝑒  |𝑡𝑜𝑡𝑎𝑙 𝑐𝑎𝑙𝑜𝑟𝑖𝑒  𝑡𝑜𝑡𝑎𝑙 𝑓𝑜𝑜𝑑 𝑐𝑎𝑙𝑜𝑟𝑖𝑒| (8)

𝑓𝑖𝑡𝑛𝑒𝑠𝑠 𝑎𝑣𝑒𝑟𝑎𝑔𝑒 

∑















(9)

Where:

= fitness score for each individual

k = individual in a week (1-7)

2.6 Selection

In this research, we use Rank-Based fitness so that the

selection of individual is based on the lowest fitness

score. It aims to select the individual which becomes

a parent for the next phase, crossover. The number of

selected individual depends on the probability of

crossover.

2.7 Crossover

The used method of crossover in this research is one-

point crossover which generates the offspring from

every 2 parents. These following steps are done in the

process of crossover.

Step 1 : Calculate the number of

individual which becomes a parent for crossover by

using equation (10)

𝑁



 𝑁



𝑥 0.1 (10)

Where N

is the number of individual for

crossover and N

is the number of individual.

Step 2 : Determine the cut point of gen

by generating a random number from 1 until the

number of chromosome for each parent. The

crossover will start from the gen with the position

after the random number.

Step 3 : Performing crossover. Genes in

parent 1 will be copied to the offspring before the cut

Genetic Algorithm for Scheduling Diet Case: Liver Patient

1907

point, while genes in parent 2 will be copied to the

offspring after the cut point. This process will

generate one new offspring.

2.8 Mutation

Random mutation will be conducted after the

crossover by randomizing the position of

chromosome and gen to be mutated. These following

steps will clearly explain the process of random

mutation.

Step 1 : Calculate the number of gen to

be mutated. It depends on the Probability of mutation

) and the number of genes from all chromosomes.

In this research, we set 0.2 as probability value, while

the total of gen depends on the number of initial

population and generation. The calculation can be

done using equation (11).

𝑁



 𝑃



𝑥 𝑡𝑜𝑡𝑎𝑙 𝑜𝑓 𝑖𝑛𝑑𝑖𝑣𝑖𝑑𝑢𝑎𝑙 𝑥 15 (11)

Step 2 : Determine the position of

individual and gen to be mutated randomly.

Step 3 : Performing mutation. The gen

will be mutated if the gen in the next chromosome has

the same type. For example, if generated gen is gen

with type of staple food in breakfast, then the gen will

be exchanged with the gen with the type of staple food

in lunch. And then the gen with the same type will be

exchange to chromosome of breakfast. It aims to keep

the calorie of every type of food as needed.

3 SYSTEM DESIGN

Figure 3 shows the general architecture of our

proposed method.

Figure 3: General architecture

From the figure above, the iteration numbers of

initial inputs to calculate the calorie needs from the

patients by implementing the Genetic Algorithm

(GA) by generating the number of selecting the 7

(seven) best individuals (solutions) with the lowest

fitness score, performing one-point crossover to the

individuals with the lowest score, and performing

random mutation to the chromosome. The output of

this system is a schedule menu in a week for liver

patients.

4 SYSTEM TESTING

The testing is conducted 8 times with different

number of initial individual and generation. 10 initial

individual is tested with the number of generation

respectively 60, 80, and 100. It is also conducted with

20 initial individual in the next testing. Then, 30

initial individual is tested with 60 and 80 generations.

The result of this research is schedule of diet for liver

patients.

Table 3:

Testing Result with 10 Initial Individual

Number of Generation Minimum fitness score

60 74.5

80 13.5

100 11.5

Table 4:

Testing Result with 20 Initial Individual

Number of Generation Minimum fitness score

60 74.5

80 13.5

100 11.5

Table 5:

Testing Result with 30 Initial Individual

Number of Generation Minimum fitness score

60 16.5

80 4.1

5 RESULT AND EVALUATION

From all the testing results, it can be concluded that

the greater the number of generation, the greater the

evolution of individual that causes the more

possibility of fitness score approaches or equals to 0

(zero). It means that the calorie total of the generated

food menu will correspond to the total of patient’s

daily calorie needs.

6 CONCLUSIONS

Based on all testing result of diet scheduling system

for liver patients, there are several obtained

ICOSTEERR 2018 - International Conference of Science, Technology, Engineering, Environmental and Ramiﬁcation Researches

1908

conclusions. First, the greater the number of

population, the more the possibility of fitness score

changes into approaching best fitness score. The

greater the number of generation, the greater the

evolution of individual that causes the more

possibility of fitness score approaches or equals to 0

(zero). Best individual has the smallest fitness score.

The last solution of scheduling diet can change every

time running the system. It is caused by the initial

population generated randomly, so that the generated

fitness score in the solution of scheduling diet

becoming more varied.

For further research, adding the generated food

menu can be done to make it more varied and increase

the appetite of liver patients, but still limit the food

that contains meat. Another optimization algorithm

can be used to improve the effectiveness of the

obtained results.

ACKNOWLEDGEMENTS

This research was supported by Universitas Sumatera

Utara. All the faculty, staff members and laboratory

technicians of Information Technology Department.

REFERENCES

Dixit, V., Pruthi, J., 2014. “Review of image processing

techniques for automatic detection of tumor in human

live”, International Journal of Computer Science and

Mobile Computing, vol. 3, pp. 371-378.

Kargulewicz, A., Stankowiak-Kulpa, H., Grzymisławski,

M., 2014. “Dietary recommendations for patients

with nonalcoholic fatty liver disease,” Prz

Gastroenterol, vol. 9, pp. 18-23

Norouzi, S., Nematy, M., Zabolinezhad, H., Sistani, S.,

Etminani, K., 2016. “Food recommender systems for

diabetic patients: a Narrative review,” Reviews in

Clinical Medicine.

El-Dosuky, M. A., Rashad, M. Z., Hamza, T. T., El-

Bassiouny, A. H., 2012. “Food recommendation

using ontology and heuristics,” in International

Conference on Advanced Machine Learning

Technologies and Applications, p. 423-429.

Chen, R. C., Lin, Y. D., Tsai, C. M., Jiang, H., 2013.

“Constructing a diet recommendation system based

on fuzzy rules and knapsack method,” in

International Conference on Industrial, Engineering

and Other Applications of Applied Intelligent

Systems, p. 490-500.

Kumar, M., Husian, M., Upreti, N., Gupta, D., 2010.

“Genetic algorithm: Review and application,”

International Journal of Information Technology and

Knowledge Management, vol. 2, pp. 451-454.

Skinner, B., Yuan, S., Huang, S., Liu, D., Cai, B.,

Dissanayake, G., Pagac, D., 2013. “Optimisation for

job scheduling at automated container terminals

using genetic algorithm,” Computers & Industrial

Engineering, vol. 64, pp. 511-523.

Filho, M. G., Barco, C. F.. Neto, R. F. T., 2014. “Using

Genetic Algorithms to solve scheduling problems on

flexible manufacturing systems (FMS): a literature

survey, classification and analysis,” Flexible Services

and Manufacturing Journal, vol. 26, pp. 408-431.

Xu, Y., Li, K., Hu, J., Li, K., 2014. “A genetic algorithm

for task scheduling on heterogeneous computing

systems using multiple priority queues,” Information

Sciences, vol. 270, pp. 255-287.

Genetic Algorithm for Scheduling Diet Case: Liver Patient

1909