Personalized Nutritional Guidance System to Prevent Malnutrition in

Pluripathological Older Patients

Jordi Torres

, Garazi Artola

1,2,3

, Nekane Larburu

1,2

, Amaia Agirre

, Elixabete Narbaiza

Idoia Berges

and Ainhoa Lizaso

Vicomtech Foundation, Basque Research and Technology Alliance (BRTA), Mikeletegi 57,

20009 Donostia-San Sebastián, Spain

Biodonostia Health Research Institute, (Bioengineering Area), eHealth Group, 20014 San Sebastián, Spain

Computer Languages and Systems Department, Faculty of Informatics, University of the Basque Country UPV/EHU,

20018 Donostia-San Sebastián, Spain

Grupo Ausolan, 20500, Arrasate, Spain

Keywords: Decision Support System, Nutritional Recommendation, Malnutrition, Comorbidity, Ageing.

Abstract: Malnutrition is a frequent problem in the elderly population, who usually is affected by one or more

pathologies. The health status of these patients can get worsened if malnutrition is left untreated. Nutritional

guidelines have been developed to fulfil the nutritional needs derived from certain pathologies, but still are

not easy to use. Digital tools can help implement and use these guidelines in real clinical scenarios. Current

solutions are designed around a single pathology or specific scenario, but the pluripathologic scenario presents

a challenge when it comes to provide nutritional support. In this paper, we present an adaptative tool that

provides personalized nutritional recommendations for pluripathological patients in an efficient way, and can

be extended to include other pathologies.

1 INTRODUCTION

Nutrition is often focused on healthy habits to prevent

obesity, diabetes and other diseases. However,

advanced age brings with it a series of physiological

(e.g., swallowing problems or inability to prepare the

meals) and psychological (e.g., social isolation)

changes that predispose people to inappropriate

eating habits and/or imbalances between nutrient

intake and the needs of the individual for an optimal

physical status that leads in malnutrition (Agarwal,

Miller, Yaxley, & Isenring, 2013). In addition,

pathological problems (e.g., several medications) that

accompany aging are considered one of the main

reasons negatively affecting the motivation to eat and

thus being possible causes of malnutrition within

elderly population. It is triggered by loss,

dependency, loneliness, and chronic illness, and

potentially impacts on higher morbidity and

mortality. Without intervention, it presents a

downwards trajectory leading to poor health and

decreased quality of life. That is why it is essential to

consider new multidisciplinary approaches and the

use of nutritional formulas to solve disease-related

malnutrition, as it is a health problem of high

prevalence and high costs for public health.

Nutritional guidelines have been developed to

cope with the needs related with specific pathologies

(NHLBI, 1998; Kushi, et al., 2012; Joint WHO/FAO

Expert Consultation on Diet, Nutrition and the

Prevention of Chronic Diseases, 2003). However, this

type of traditional nutrition evaluations not only

require the use of numerous tables and lists to provide

sufficient nutritional recommendations for each

patient, but are also very time-consuming due to

cross-referencing and calculations (Kuo, et al., 2018).

In this context, digital tools can help to develop more

competitive and personalized nutritional

recommender systems.

Different approaches have been recently carried

out with the objective of offering this type of decision

support systems both to patients (direct end-users) or

to health professionals or caregivers. For instance,

this recent study (Taweel, et al., 2016) presents the

design of a system that enables homecare

management in the context of self-feeding and

malnutrition prevention through balanced nutritional

338

Torres, J., Artola, G., Larburu, N., Agirre, A., Narbaiza, E., Berges, I. and Lizaso, A.

Personalized Nutritional Guidance System to Prevent Malnutrition in Pluripathological Older Patients.

DOI: 10.5220/0011113700003188

In Proceedings of the 8th International Conference on Information and Communication Technologies for Ageing Well and e-Health (ICT4AWE 2022), pages 338-343

ISBN: 978-989-758-566-1; ISSN: 2184-4984

intake. The design employs a service-based system

that incorporates several services including

monitoring of activities, nutritional reasoning for

assessing feeding habits, and diet recommendation

for food planning. In another study (Kuo, et al.,

2018), a clinical nutritional information system was

implemented to help hospital dietitians perform their

daily work more effectively in terms of time

management and paper work. The system mainly

targets in-patients who require cancer-nutrition

counselling.

As far as we can find in the state of the art, some

studies focus their work in supporting patients that

suffer from a specific disease or health problem

(Paulsen, Varsi, & Andersen, 2021), while the others

use patients’ general information (e.g., demographic

or socioeconomic data) without considering any

clinical data (Leipold, et al., 2018). Nevertheless, the

clinical information or health status of an older adult

is a variable that is essential to be considered for

obtaining a reliable nutritional recommendation. In

this sense, approaches that support not only single

pathologies but different combinations of

comorbidity cases are needed.

In addition, the use of technologies that pursue

knowledge sharing and interoperability is more and

more important in this kind of DSSs. That is why

recent approaches integrate semantic web

technologies into their systems (Espín, Hurtado, &

Noguera, 2016), allowing the inference of new

knowledge that could not be extracted from

traditional nutritional databases.

In this paper, the implementation of a solution that

offers personalized nutritional recommendations

taking into account not only the personal information

(e.g., malnutrition risk, gender) but also the

comorbidity profile of an elderly patient is presented.

This solution provides health professionals and

caregivers with a nutritional decision support system

that considers not only the different nutritional needs,

but also the whole environment of an elderly patient,

such as socio-demographic factors (e.g., sex, age...),

psychosocial factors (e.g., psychosocial disorders),

and morbidity factors (diseases).

2 METHODOLOGY

In this study we present the following approach to

develop a personalized nutritional guidance system,

which consists in three different modules: (i)

nutritional recommendations for each patient profile,

(ii) nutritional rule flow to generate the personalized

recommendations and (iii) nutritional ontology for

standardize the nutritional knowledge across the

platform.

2.1 Nutritional Recommendations

Nutritional guidelines were developed along with

experienced nutritionists on the domain. These

recommendations were generated for a set of possible

profiles considering i) the age and sex of the user, ii)

the malnutrition risk, iii) the need for texture adapted

meals, and iv) the intake level.

The structure of the recommendation provided by

our solution is divided as follows:

• Diet: Daily intake of several nutrients such as

carbohydrates, fats, cholesterol, proteins, fiber,

liquids, or salt; and daily caloric and protein

distribution.

• Fortification: Complement meals that help

reach the daily caloric and protein intake

values. These caloric and protein enrichments

are given by means of natural foods. If natural

fortification is not possible, it is carried out

artificially, through protein and/or caloric

modules.

• Food and liquid adaptation: Mainly texture

adaptations in order to ease the swallowing

process (i.e., solid food grinding). If these

adaptations entail low energy and nutritional

intake, Enteral Nutrition is recommended. In

case the patient's intestinal tract is not

functional, Parenteral Nutrition is prescribed.

• Supplementation: Recommendation of specific

oral nutritional supplements such as

hypercaloric and hyperprotein formulas.

• Enteral nutrition: Recommendations for

patients in need of enteral nutrition.

• Follow-up: Re-evaluation of the nutritional

status. For instance, the continuation,

reduction, or withdrawal of the

supplementation can be considered in this area.

Each of these areas is formed by different

attributes. For instance, the Diet area is divided into

different recommendations, each one containing the

specific information about a different nutrient (see

Figure 1).

Figure 1: Example of a diet recommendation.

Personalized Nutritional Guidance System to Prevent Malnutrition in Pluripathological Older Patients

339

The presented structure is the core of the

nutritional recommendations. For each pathology and

user profile included in our platform, the first step

was to develop a complete set of recommendations

for the single pathology case. In the following step,

the combination of pathologies was tackled, taking

the single pathology case plans as the basis. In these

pluripathological cases, the nutrition plans were

generated by i) the combination of single pathology

nutrition plans (i.e., nutritional needs from pathology

1 and supplement intake from pathology 2, ii) the

generation of specific plans for each combination of

pathologies, and iii) the re-use of the nutrition plans

generated for the monopathological case if no

adaptation needs to be done for a specific

pluripathological case.

These recommendations are represented in a more

visual form to the users in a web application described

in Section 3. In Section 2.2 a detailed explanation on

how the generated nutritional recommendations were

introduced in a rule flow process is presented, for

both single and pluripathological cases.

2.2 Nutritional Rule Flow

The business logic of the presented platform consists

of a business process combined with a rule engine that

evaluates the patients’ data and generates the

personalized recommendation in return. The logic

implemented in this process consists, overall, on a

series of nodes that direct the process execution to the

set of nutrition recommendations that belong to the

pathology or combination of pathologies of the

evaluated patient. The specific recommendation is

generated based on the input data, as the rule engine

evaluates it and returns the nutritional

recommendation triggered by the patients’ data.

Our work is based on the previous work carried

out in (Torres, Artola, & Naiara, 2020) where a

platform for the development of domain independent

rule based clinical decision support systems (CDSS)

is described. From that starting point, jBPM processes

were included in the developed rule engine (Drools).

The BPMN Eclipse plugin was used to design and

implement the flows, and to link them with the

generated rules.

Three flows were configured, one for patients

with one pathology, and the rest for pluripathological

patients suffering from 2 or 3 of the included

pathologies. The general structure of the flows is the

same, an entry node, a set of decisional gateways and

business rule tasks, and the end nodes for each

possible pathological pathway (see Figure 2).

Figure 2: jBPMN process for the combination of 2

pathologies. Example of the combinations with Diabetes

and Alzheimer.

The path followed to reach an end node is what

we have called pathological pathway, theoretically,

for each pathology and its combinations, a

pathological pathway would need to be defined along

with its defined set of rules, but our solution reduced

the final amount of sets of rules needed by making

use of the execution of specific rule flow groups in

more than one pathway. The cases where the

nutritional recommendations coincide were grouped

under the same rule group attribute, reducing the total

amount of nutrition plans and rules that were needed

and avoiding duplicates.

2.3 Ontology

An ontology was developed for the nutrition

recommendations formalization process. The

variables formalized were differentiated between the

variables related with the patient and the ones related

with the nutritional recommendation. The ontology

was generated using Protégé, and it can be easily

updated with new pathologies apart from the ones

specified in Section 3.

The ontology was used as the basis from which to

formalize the diets generated by the nutritionist into

business rules. Due to the vast number of

recommendations that were entered into the system,

the use of an ontology reduces the risk of human

errors in the formalization process.

ICT4AWE 2022 - 8th International Conference on Information and Communication Technologies for Ageing Well and e-Health

340

Figure 3: Structure of the ontology used to formalize the

nutrition recommendations.

3 USE CASE

The presented solution was developed within the

NUTRIGEP project. In this project, a platform for the

management of elderly pluripathologic patients at

risk of suffering malnutrition was developed. The

main goals of the project are i) to determine the

prevalence of malnutrition in admitted patients, ii) to

analyse the possible relationship between different

risk factors and the nutritional status of patients

during admission, and iii) to correct each diagnosed

malnutritional state through the management and

personalization of diets and specific

recommendations of food and/or supplements. The

solution hereby presented corresponds with the third

goal in charge of providing the specific nutritional

recommendations that aim to improve the nutritional

state of the users.

The pathologies that were included in the

NUTRIGEP platform are representative of the ones

present among the elderly, chronic patients of the

Asunción Klinika (Tolosa, Spain), the hospital

involved in the project: diabetes, Alzheimer,

Parkinson, Chronic Obstructive Pulmonary Disease

(COPD), cardiac failure, cancer and psychosocial

disorders. For each of the included pathologies and its

combinations, a nutrition plan was developed (see

Section 2.1), and modelled into the rule flow process

(see Section 2.2). Combinations of pathologies were

limited to all possible combinations up to 3

pathologies in order to set the number of possible

diets.

Table 1: Summary of Nutrigep use case data.

Item Description Amount

Pathologies Pathologies considered

in Nutrigep

Possible

pathology

combinations

Total amount of

possible combinations

of pathologies

Nutrition plans Amount of generated

nutrition plans

Nutrition

recommendations

/plan

Amount of

recommendations per

nutrition plan

378

Total nutrition

recommendations

Total amount of

recommendations of the

Nutrigep platform

7938

The Nutrigep project is aimed at a specific

population, the elderly. Due to this, for the

development of the recommendations a minimum age

of 65 was considered. The other factors taken into

consideration to provide personalized nutrition

recommendations were the sex, risk of malnutrition,

need for food and liquid adaptation, the pathology(es)

of the patient and a valuation of the quantity of food

eaten normally by the user (see Table 2 for more

detail).

Table 2: Description of patient’s data used in the Nutrigep

project.

Variable Description Values

Age Age of the

patient

>65

Sex Sex of the

patient

Male/Female

Risk of

malnutrition

Qualitative

assessment of

risk of

malnutrition

High / medium / low

Pathology Pathology or

combination of

pathologies

Alzheimer / Diabetes

/ COPD / Cardiac

failure / Psychosocial

disorders / Cancer /

Parkinson

Intake

valuation

Qualitative

estimation of

the amount of

food consumed

by the patient

Appropriate /

Moderate / Low

Texture Texture

adaptations

needed by the

patient

Normal / Soft /

Lightly thickened /

Little thick / Wet

chopped / Blended-

moderately thickened

/ Extremely thick

mash/pudding

Personalized Nutritional Guidance System to Prevent Malnutrition in Pluripathological Older Patients

341

Figure 4: Example of the visualization of a nutritional recommendation in the Nutrigep web application for health

professionals.

The personalized nutritional recommendations

are visualized in a web application designed for

health professionals to aid them in the management

of their patients, where the developed

recommendation system was integrated. In this

platform, the clinicians can select the patient they

need to evaluate from the list of all the patients in the

database, and receive the personalized nutritional

recommendations given by our solution (Figure 4).

The clinical value of this recommender system

will be assessed in a clinical validation study that will

be carried out in Asuncion Klinika. This validation

will enrol twenty older patients suffering from one or

more of the pathologies included in the developed

system. The health professionals responsible for these

patients will access the web platform developed for

the Nutrigep project and enter the data of the patients

for the generation of personalized nutritional

recommendations. In the following three months, the

health status of the patients will be monitored and a

final evaluation will be done to assess the impact that

the nutritional guidance of Nutrigep can have on it.

4 CONCLUSIONS AND FUTURE

WORK

This paper presents a solution that can prevent

malnutrition in the geriatric environment,

contributing to the good nutritional management of

the elderly in order to improve their state of health.

It provides health professionals and caregivers

with a tool that offers personalized nutritional

recommendations for elderly pluripathologic patients

with risk of suffering from malnutrition. Not only the

personalized diet plans are generated for a specific

pathology and the patient’s malnutrition risk, but also

our solution generates specific diet plans for patients

that suffer from a combination of the diagnosed

pathologies, a typology of patients for which,

according to our research, current platforms cannot

properly offer nutritional recommendations.

Furthermore, the logic of the system can be easily

followed thanks to the visual representation of the

business processes that is being used. The system was

implemented in a real clinical scenario, where a

clinical study is going to be carried out. The clinical

study will show the impact of the developed platform

ICT4AWE 2022 - 8th International Conference on Information and Communication Technologies for Ageing Well and e-Health

342

and the developed nutritional guidelines in the

patients’ health status.

We have identified some future work to improve

the current platform. Firstly, in order to widen the

scope of potential users, other pathologies should also

be included. Furthermore, recommendations should

include other information related with the health

status, such as physical or wellbeing

recommendations, in order to provide not only diet

recommendations, but a more holistic set of

recommendations that could help improve the overall

wellbeing of the patients. Lastly, in order to make the

system easy to configure, a web authoring tool is

being developed where the user will be able to define

the different combinations of pathologies that are

supported by the platform. The diet plans will also be

generated using the same tool, as a result of this, the

scalability of the system will be increased as the

addition of new logic and new recommendations will

be supported from the same tool, reducing the work

needed to update the platform.

ACKNOWLEDGEMENTS

This study was supported by the grant ZL 2019/00647

NUTRIGEP from Eusko Jaurlaritza (Basque

Government) and the European Union under the

European Regional Development Fund (ERDF). The

funding sources had no involvement in the collection,

analysis and interpretation of data; in the writing of

the report; or in the decision to submit the article for

publication.

The study complies with the current laws of Spain

and Europe.

All authors declare that they have no competing

interests.

REFERENCES

Agarwal, E., Miller, M., Yaxley, A., & Isenring, E. (2013).

Malnutrition in the elderly: A narrative review.

Maturitas, 76, 296-302. doi:https://doi.org/10.1016/

j.maturitas.2013.07.013

Espín, V., Hurtado, M. V., & Noguera, M. (2016). Nutrition

for Elder Care: a nutritional semantic recommender

system for the elderly. Expert Systems, 201-210.

doi:10.1111/exsy.12143

Joint WHO/FAO Expert Consultation on Diet, Nutrition

and the Prevention of Chronic Diseases. (2003). Diet,

Nutrition And The Prevention Of Chronic Diseases.

Geneva: WHO technical report series.

Kuo, S.-E., Lai, H.-S., Hsu, J.-M., Yu, Y.-C., Zheng, D.-Z.,

& Hou, T.-W. (2018). A clinical nutritional information

system with personalized nutrition assessment.

Computer Methods and Programs in Biomedicine, 155,

209-216.

doi:https://doi.org/10.1016/j.cmpb.2017.10.029

Kushi, L. H., Doyle, C., McCullough, M., Rock, C. L.,

Demark-Wahnefried, W., Bandera, E. V., Guidelines,

T. A. (2012). American Cancer Society Guidelines on

Nutrition and Physical Activity for Cancer Prevention.

CA: A Cancer Journal for Clinicians, 30-67.

doi:10.3322/caac.20140

Leipold, N., Madenach, M., Schäfer, H., Lurz, M.,

Terzimehic, N., Groh, G., Krcmar, H. (2018). Nutrilize

a Personalized Nutrition Recommender System: an

Enable Study. HealthRecSys.

NHLBI, O. E. (1998). Clinical Guidelines on the

Identification, Evaluation, and Treatment of

Overweight and Obesity in Adults. Bethesda (MD):

National Heart, Lung, and Blood Institute. Retrieved

from https://www.ncbi.nlm.nih.gov/books/NBK2003/

Paulsen, M., Varsi, C., & Andersen, L. (2021). Process

evaluation of the implementation of a decision support

system to prevent and treat disease-related malnutrition

in a hospital setting. BMC Health Serv Res, 21, 281.

doi:https://doi.org/10.1186/s12913-021-06236-3

Taweel, A., Barakat, L., Miles, S., Cioara, T., Anghel, I.,

Tawil, A.-R. H., & Salomie, I. (2016). A service-based

system for malnutrition prevention and self-

management. Computer Standards & Interfaces, 225-

233. doi:https://doi.org/10.1016/j.csi.2016.03.005

Torres, J., Artola, G., & Naiara, M. (2020). A Domain-

Independent Semantically Validated Authoring Tool

for Formalizing Clinical Practice Guidelines. Studies in

Health Technology and Informatics, 517-521.

Personalized Nutritional Guidance System to Prevent Malnutrition in Pluripathological Older Patients

343