Application of Food Nutrient Composition Analysis and Computer

Algorithm in Nutrition Recommendation System

Shuliang Liu and Jinxu Gao

Weifang Engineering Vocational College, 262500, China

Keywords: Food Nutrient Analysis, Computer Algorithms, Apriori Algorithm, Nutrition Recommendation System

Application.

Abstract: This paper studies the application of food nutrition composition and computer algorithm in nutrition

recommendation system, and uses the Apriori algorithm in computer algorithm to optimize the nutrition

recommendation results. In the process of research, this paper designs a nutritional recommendation computer

algorithm based on Apriori, further improves it, and then realizes a nutritional recommendation system based

on food nutrient analysis and Apriori algorithm based on the integration of various components. In addition,

the system is used for practical applications. According to the experimental data, the recommended

combination of high protein and low fat has met the needs of 85% of users, and the recommended combination

of high fiber and low carb water can effectively help control blood sugar. Comprehensive analysis shows that

the system can provide accurate dietary recommendations based on in-depth nutritional analysis to improve

the health of users.

1 INTRODUCTION

In recent years, due to the enhancement of Chinese

people's health awareness, personalized nutrition

recommendations have become a research hotspot.

There are researchers who can solve the problem of

personalized meal recommendations based on

manual calculations, but these methods are inefficient

and cannot meet individual needs (Beecher, 2024).

Some researchers also use statistical methods, such as

simple regression analysis, to make nutritional

recommendations, but their ability to analyze

complex nutrient combinations is limited (Beltramo,

and Bast, et al. 2023). In addition, some traditional

rule-based systems can provide some

recommendations, but they cannot be adjusted in time

according to the dynamic needs of users (Dunlop, and

Cunningham, et al. 2025). In this paper, computer

algorithms, especially Apriori association rule

mining, are used to analyze the nutritional

composition of food and make personalized

recommendations (Lara-Arevalo, and Laar, et al.

2024). The algorithm can efficiently process large-

scale nutrition data and further generate accurate

recommendations based on individual needs

(Marchese, and Hendrie, et al. 2024). This round

mainly analyzes the relationship between the

enhancement of people's health awareness and the

demand for personalized nutrition recommendation,

and puts forward the shortcomings of data statistical

methods and rule-based traditional systems in the

pertinence of nutrition recommendation, and at the

same time, introduces the value of food nutrition

composition analysis and computer algorithm in

nutrition recommendation, and then determines the

research on the system.

2 RELATED WORKS

2.1 Advantages of Computer

Algorithms

In the food nutrient analysis and nutrition

recommendation system, computer algorithms are

extremely crucial. Based on accurate data processing

and efficient computer technology, the system can

quickly parse huge nutrient data sets (Martins, and

Magnusson, et al. 2023). Computer algorithms

provide the basis for the calculation and analysis of

complex food nutrients and related data, and

effectively improve the accuracy and efficiency of

332

Liu, S. and Gao, J.

Application of Food Nutrient Composition Analysis and Computer Algorithm in Nutrition Recommendation System.

DOI: 10.5220/0013541000004664

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 3rd International Conference on Futuristic Technology (INCOFT 2025) - Volume 1, pages 332-338

ISBN: 978-989-758-763-4

nutrient composition analysis and prediction based on

mechanism optimization and machine learning

methods (Pickford, and McCormack, et al. 2022). In

recommendation systems, common algorithms

include regression analysis, clustering algorithms,

and association rule mining, which are conducive to

discovering hidden patterns in food nutrition content

data and generating targeted recommendations. The

introduction of computer algorithms, especially those

based on data mining, has greatly improved the ability

of the system to process large-scale complex data on

food nutrition content (Silveira, and Valler, et al.

2024). These algorithms can accurately predict the

nutritional composition of food based on the user's

historical data, and have certain learning and

optimization capabilities. Based on continuous

updates and adjustments, the system can revise the

mechanism according to real-time data to ensure the

dynamic update and optimization of recommendation

results. This ability to self-regulate makes the

application of computer algorithms in nutritional

recommendation systems extremely critical (Svarc,

and Jensen, et al. 2022).

2.2 Apriori Algorithm Can be Used for

Estimation of the Frequency of

Occurrence of Nutrient

Combinations

Among many computer algorithms, Apriori

algorithm is an important association rule mining

algorithm, which can find frequent item sets from

large data sets and generate association rules based on

these item sets. It is based on a recursive approach to

find frequently occurring combinations in the data,

such as the nutritional composition of a food, and then

identify effective recommendation patterns (Víquez,

and Morales, et al. 2022). The algorithm first defines

the degree of support, which is mainly used to

measure the frequency of occurrence of one nutrient

combination , and the confidence level, which

measures the probability of another ingredient

appearing when there is a particular nutrient

combination. In the nutrition recommendation

system, the Apriori algorithm provides data support

for personalized dietary recommendations based on

the screening of high-support and high-confidence

combinations of nutrients. Compared to other

traditional algorithms, the Apriori algorithm can

efficiently process multi-dimensional data and

generate personalized nutrition recommendations for

users. The advantage is that it can mine high-value

correlation information from the big data set, and use

this information to optimize the recommendation

mechanism of the system.

3 METHODS

3.1 The Structure of Each Component

of the Nutrition Recommendation

System

The research in this paper needs to construct a

nutritional recommendation system, which includes

six components, each of which has its own content

that needs to be responsible. Specifically, the task of

the data collection component is to collect data

related to the user's food nutrients, such as the user's

dietary history, physical fitness information,

nutritional needs, etc. The widget is based on an

interface to obtain the nutritional content data of food

products, and then maintain the integrity and

accuracy of the data. In addition, the widget interfaces

with external databases, allowing the latest food data

to be updated in real time. The task of the data

preprocessing component is to clean, standardize and

structure food nutrition data to ensure that the data

format is uniform. Data preprocessing mainly focuses

on the processing of missing values, outliers and

standardized nutrient composition data. It is also

possible to select key food ingredients relevant to

nutritional analysis based on feature extraction,

preparing them for subsequent analysis. The task of

the Association Rule Mining component is to apply

the Apriori algorithm to perform frequent itemset

mining and association rule generation. It can be

based on a set minimum level of support and

confidence to provide insight into potential

associations between food nutrients. Another task is

to export the association rules of nutrients to provide

stable, reliable, and comprehensive data support for

personalized recommendations. The task of the

recommendation engine component is to apply the

mined association rules to generate nutritional

recommendations according to the user's individual

needs. The recommendation engine provides users

with suitable food recommendations based on the

user's health goals, such as the user's fat loss goals,

muscle gain goals, etc. It can adjust the recommended

regimen in real-time to meet the different dietary

needs of users. The task of the User Feedback widget

is to collect user feedback on the recommendation

results and apply their feedback to system

optimization. Users can provide feedback on

recommended foods based on ratings, evaluations,

Application of Food Nutrient Composition Analysis and Computer Algorithm in Nutrition Recommendation System

333

etc. User feedback data is fed into the Associated

Rule widget to improve recommendation accuracy

and user experience. The task of the system

management component is to manage and monitor the

system as a whole, such as algorithm tuning, data

update, user management, etc. It ensures the stable

operation of the system and monitors the system

performance, such as recommendation accuracy, user

satisfaction, etc. The system administrator can adjust

the system parameters based on this widget to further

optimize the functionality of the system.

3.2 Mining of Frequent Itemsets and

Design of Computer Algorithms for

Nutritional Recommendations

In the frequent itemset mining stage, the algorithm

mechanism needs to mine the frequent component

combinations from a large number of food nutrient

data. The nutritional content of each food, such as

proteins, fats, carbohydrates, vitamins, minerals, etc.,

can be considered as an item. Apriori taps into these

frequent combinations of nutrients by progressively

filtering out the most supportive items. See Eq. (1) for

this.

|T(X)|

Support(X)

|T|

)

In this formula,

()

Support X

refers to the

degree of support of item X, which is the frequency

of occurrence of the nutrient item set in all foods.

()

Refers to the number of transactions that

contain item set X, in this case the number of foods

containing a specific combination of nutrients.

Refers to the total number of transactions, which is

the total number of all food samples. Based on this

formula, it is possible to calculate the degree of

support for each nutrient combination and identify the

nutrient combinations that occur frequently. For

example, if the combination of high protein and high

fiber is more supportive, it proves that this

combination is frequent in many foods and can be

used as a basis for nutritional recommendations.

After you've identified a frequent itemset, the next

step is to build an association rule. Association rules

are used to explore potential relationships between

nutrients, such as whether a food containing one

ingredient is often accompanied by another. The

strength of this relationship is measured based on the

calculation confidence, for which see Eq. (2).

Support(X Y)

Confidence(X Y)

Support(X)

∪

 =

)

In this formula,

Confidence(X Y)

refers to

the probability that a food containing

X is another

nutrient at the same time

Y given a combination of

nutrients.

Support(X Y)∪

Refers to X Y is the

proportion of food products that contain both and in

the data.

Support(X Y)∪

Refers to the support of

the itemset

X . Is based on the calculation confidence,

it is possible to identify strong associations between

nutrients, such as certain high-protein foods that are

often found to be present with low-fat nutrients, and

corresponding association rules can be generated for

nutrition recommendations.

After the rules are generated, the associated rules

are filtered and refined. Lift is a key indicator of the

strength of association rules, which measures the

actual relevance of a combination of nutritional

recommendations. For this, see Eq. (3).

Confidence(X Y)

Lift(X Y)

Support(Y)



 =

)

In this formula,

Lift(X Y)

refers to the

degree of lift, which represents

is the actual

correlation between and .

Confidence(X Y)

Refers to the confidence level of the rule.

()

Support Y

Refers to the degree of support of item

(Y), which is the proportion of nutrients in a food

. If the lift is greater than 1, there is a strong positive

correlation between the two nutrients and can be used

for nutritional recommendations. For example, if the

system finds that a high-protein food is often

accompanied by a low-fat ingredient and has a high

degree of lifting, it can consider this combination as

part of the recommendation.

3.3 Further Improvement of The

Computer Algorithm for

Nutritional Recommendation

At this stage, the system generates and optimizes

association rules based on frequent scans of nutrient

data. Specifically, it is necessary to optimize the

support, confidence, and promotion thresholds of

rules to ensure that the rules generated by the system

have high relevance and reliability. Training on

multiple food samples allows the system to find high-

INCOFT 2025 - International Conference on Futuristic Technology

334

frequency and high-confidence nutrient combinations

to improve the accuracy of recommendations. For

example, the system will generate personalized

nutrition recommendations based on the user's dietary

history, recommending high-fiber foods to

supplement the user's daily missing nutrients.

In order to improve the efficiency and accuracy of

the mechanism, the generated association rules

should be pruned in the optimization stage. Rule

pruning is based on setting minimum confidence and

lift thresholds to remove redundant, meaningless

rules. For example, some rules have a higher

confidence level of less than 1, indicating that the

nutrient content of these rules is not strongly

correlated. For this, see Eq. (4).

{

Prune(X Y) Confidence(X Y)1 if 0 otherwise \ =  ≥τ

(4)

In this formula, the

is pruning threshold is the

minimum confidence level that is set. Based on

pruning, it can reduce useless association rules and

improve the computational efficiency of the nutrition

recommendation mechanism. For example, the

system can delete food combinations with low

confidence and weak relevance, and retain rules that

are beneficial to the user's health.

Rule merging is another important step in

improving the mechanism, based on the expansion

and merging of similar rules, the system can generate

more flexible and diverse nutrition recommendations.

For example, two similar combinations of nutrients

can be combined into a new, more extensive

recommendation rule, allowing users to have a more

diverse nutritional choice. See Eq. (5) for this.

Combine(X Y, X Y) X X Y

12 12

=∪

(5)

In this formula, X,X

refers to a combination

of different nutrient ingredients that can be combined

to generate a new rule. Based on the consolidation of

rules, the system will increase the flexibility of

recommendations and provide comprehensive

nutritional recommendations based on the

combination of nutrients in different foods.

In the process of improving the mechanism, it is

also necessary to evaluate the mechanism so that the

generated association rules and recommendation

results can meet the expected accuracy and user

satisfaction. Commonly used evaluation metrics are

accuracy, recall, and F1 score. See Eq. (6) for this.

Precision Recall

=×

(1)

In this formula,

Precision

refers to the precision

rate, which indicates the correct proportion of the

recommended nutrient content.

Recall

refers to the

recall rate, which is the proportion of the nutrient that

the system correctly recommends in the user's actual

needs.

Based on the evaluation of the accuracy and

recommendation effect of the system, the parameters

of the mechanism can be further optimized, and the

overall performance of the system can be improved

based on this. For example, if the system finds that

certain recommended rules have low F1 scores, they

will need to be readjusted.

3.4 Integration of the Various

Components of the Nutrition

Recommendation System

In this step, it is necessary to combine the various

components of the nutrition recommendation system

to ensure the smooth flow of data for food nutrition

content analysis and achieve accurate personalized

recommendations. To this end, the data acquisition

component will obtain the user's dietary data and food

nutrition content in real time from external databases

and sensors, and these data will be comprehensively

processed by the data preprocessing component to

become data that is more in line with the system

requirements. The processed data is transferred to the

Association Rule Mining widget, which is tasked

with analyzing the correlation of nutrients in food

using the Apriori algorithm and generating frequent

itemsets and association rules that meet the health

needs of the user. These rules will be passed to the

recommendation engine artifact to provide users with

personalized nutrition recommendations. After that,

the user feedback component will play an effective

role, which will collect feedback data such as user

satisfaction with the recommended food and the

actual effect, and return all this information to the

association rule mining model, and then realize the

update of the rules and the optimization of one set.

Then, the system management component will

coordinate the operation of each component in a

unified manner, and monitor the performance

indicators of the system, such as recommendation

accuracy, response speed and user satisfaction, which

can ensure that the nutrition recommendation system

continues to adapt to the needs of users. Based on this,

the entire nutrition recommendation system can

adjust the recommendation content in real time, and

give more accurate and personalized nutritional

analysis and recommendations.

Application of Food Nutrient Composition Analysis and Computer Algorithm in Nutrition Recommendation System

335

4 RESULTS AND DISCUSSIONG

4.1 Background of the Case

In this application case, this paper uses a food

nutrition analysis and nutrition recommendation

system based on the Apriori algorithm to accurately

analyze the nutritional needs of different user groups,

such as athletes, diabetic patients, the elderly, and

ordinary adults, to achieve personalized

recommendations. According to their different

nutritional needs, the system analyzes the most

suitable combination from the nutrient content of the

food and generates corresponding dietary

recommendations. These combinations generally

contain a variety of nutrients, such as proteins, fats,

carbohydrates, etc., covering 4 types of foods with

high nutrient content. For example, chicken breast

contains 31.0g of protein, 3.6g of fat, etc.

Table 1: Food nutrient composition data

Name of

the foo

Protein (g) Fat (g) Carbohydrates

(g)

Chicken

Breast

31.0 3.6 0

Cauliflower 2.8 0.4 6

Brown

Rice

3.5 1.0 22

Avocado 2.0 15.0 9

Table I shows the nutritional content data of the

foods analyzed by the recommendation system, such

as chicken breast is rich in protein, avocado is more

fat, and brown rice is higher in carbohydrates, which

provide the basis for the system's nutritional

recommendations. The process of recommending

nutritional components is shown in Figure 1.

Figure 1: The process of recommending nutritional

components.

4.2 The Recommended Effect of

Nutritional Components

Table 2: Types of users and nutritional needs

User type Protein

requirement (g)

Carbohydrate

requirement

(g)

Fat requirement

(g)

Jock 150 300 70

Diabetic 60 150 60

Senior

Citizen

50 200 65

Ordinary

Adults

70 250 75

Table 2 shows the daily nutritional requirements

for different user types, such as athletes who need

more protein and people with diabetes who need to

limit carbohydrate intake. Based on these needs, the

system generates personalized dietary

recommendations，The distribution of nutrients is

shown in Figure 2.

Figure 2: The distribution process of nutritional

components.

The nutritional analysis can be integrated with

computers to better recommend ingredients.

4.3 The Overall Effectiveness of the

Nutrient Recommendation System

A comprehensive analysis of the data in the above

three tables shows that the system can be significantly

applied to different user groups. Specifically, high-

protein foods, such as chicken breast, are commonly

INCOFT 2025 - International Conference on Futuristic Technology

336

used in athletes' dietary recommendations, with a

90% accuracy rate for systematically recommended

protein intake. In addition, the system recommended

a high-fiber, low-carbohydrate food combination for

diabetic patients, as shown in Table 3.

Table 3: Association Rule Mining Results

Nutritional Composition

Combination

Support Confidence Lift

High in protein, low in fat 0.52 0.80 1.25

High in fiber, low in carbs 0.43 0.75 1.30

High protein, high fiber 0.48 0.78 1.35

High in vitamin C 0.35 0.60 1.18

Table 3 shows that the nutritional

recommendation system designed in this design has a

high degree of support and confidence based on the

combination of nutrients generated by association

rule mining, such as the combination of high protein

and low fat, so as to provide practical and efficient

nutritional recommendations for athletes and weight

control users with a support level of 0.43 and a

confidence level of 0.75 to ensure effective blood

glucose control. Moreover, dietary recommendations

for the elderly and ordinary adults also show the

effectiveness of the system, The analysis of

nutritional components is shown in Figure 3.

Figure 3: Analysis of nutritional components

From Figure 3, it can be seen that in the process

of recommending nutritional components, the

nutritional components are relatively reasonable and

can meet the actual needs. The fat intake is

systematically optimized to ensure that the daily fat

intake is controlled within a healthy range. User

feedback showed that the system-recommended food

combinations were effective in improving the balance

of nutrient intake, with 80% of users achieving

significant improvements in weight control, blood

sugar regulation, and overall health. Overall, the

nutrition recommendation system is based on an in-

depth analysis of the nutritional content of food to

provide accurate and personalized nutrition

recommendations for different user groups. Its

powerful association rule mining and high-support

and confidence combination recommendation can

finally achieve the user's health goals ， the final

recommended nutritional composition is shown in

Figure 4.

Figure 4: The final recommended nutritional content.

5 CONCLUSIONS

This paper verifies the efficient application of food

nutrient analysis and computer algorithm in nutrition

recommendation system. Based on the Apriori

algorithm, the system can quickly and accurately

analyze complex nutrient combinations in food and

provide precise dietary recommendations based on

user needs. The results show that the algorithm

performs well in processing large-scale nutrition data

and meeting personalized health needs, and provides

strong technical support for the realization of

personalized nutrition management. Moreover, the

system is flexible enough to dynamically adjust based

on user health feedback and continuously optimize

recommendations. This shows that food nutrition

analysis combined with computer algorithms can be

widely used in the field of health management and

improve the effect of personalized recommendation.

The data content of this paper is still relatively

limited, so it has certain limitations and needs to be

further studied in the future.

REFERENCES

Beecher, G.R. (2024). Evolution of food composition

knowledge in the United States from its beginning.

Journal of Food Composition and Analysis, 126.

Beltramo, B., Bast, A., Diliën, H., de Boer, A. (2023).

Nutrient composition of fresh produce-Assessing

variability between European countries to substantiate

Application of Food Nutrient Composition Analysis and Computer Algorithm in Nutrition Recommendation System

337

nutrition and health claims. Journal of Food

Composition and Analysis, 118.

Dunlop, E., Cunningham, J., Adorno, P., Fatupaito, S.,

Johnson, S.K., Black, L.J. (2025). Development of an

updated, comprehensive food composition database for

Australian-grown horticultural commodities. Food

Chemistry, 463.

Lara-Arevalo, J., Laar, A., Chaparro, M.P., Drewnowski,

A. (2024). Nutrient-Dense African Indigenous

Vegetables and Grains in the FAO Food Composition

Table for Western Africa (WAFCT) Identified Using

Nutrient-Rich Food (NRF) Scores. Nutrients, 16(17).

Marchese, L.E., Hendrie, G.A., McNaughton, S.A.,

Brooker, P.G., Dickinson, K.M., Livingstone, K.M.

(2024). Comparison of the nutritional composition of

supermarket plant-based meat and dairy alternatives

with the Australian Food Composition Database.

Journal of Food Composition and Analysis, 129.

Martins, V.A., Magnusson, N., Laska, M. (2023). Go for

Lipids! Food Preferences and Nutrient Composition in

Zoo-Housed White-Faced Sakis, <i>Pithecia

pithecia</i>. International Journal of Primatology,

44(2), 341-356.

Pickford, C., McCormack, L., Liu, Y.B., Eicher-Miller,

H.A. (2022). US Department of Agriculture Food

Composition Databases, the Food and Nutrient

Database for Dietary Studies 2013-2014, and the

National Nutrient Database for Standard Reference

Version 28 Yield Significantly Different Nutrient

Totals of Food Items from Eight Midwestern Food

Pantry Inventories. Journal of the Academy of Nutrition

and Dietetics, 122(7), 1326-+.

Silveira, P., Valler, I.W., Hirano, Z.M.B., Dada, A.N.,

Laska, M., Salazar, L.T.H. (2024). Food preferences

and nutrient composition in captive Southern brown

howler monkeys, <i>Alouatta guariba clamitans</i>.

Primates, 65(2), 115-124.

Svarc, P.L., Jensen, M.B., Langwagen, M., Poulsen, A.,

Trolle, E., Jakobsen, J. (2022). Nutrient content in

plant-based protein products intended for food

composition databases. Journal of Food Composition

and Analysis, 106.

Víquez, C.H., Morales, J.C., Castro, M.M., Herrera, C.C.

(2022). Analysis of methodological components and

available resources in Costa Rica to generate food

composition data. Journal of Food Composition and

Analysis, 106.

INCOFT 2025 - International Conference on Futuristic Technology

338