Wellbeing Recommender System, a User-Centered Framework for

Generating a Recommender System for Healthy Aging

Jordi Torres

, Meritxell Garcia

, Garazi Artola

1,2 c

, Teresa Garcia-Navarro

Isabel Amaya

, Nekane Larburu

1,2 f

and Cristina Martin

1,3 g

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

Mikeletegi 57, 20009 Donostia-San Sebastián, Spain

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

Faculty of Engineering, University of Deusto, Av.Universidades, 24, 48008, Bilbao, Spain

Keywords: Healthy Aging, Recommender System, Quality of Life, Synthetic Data Generation.

Abstract: The needs of the currently aging population require new technologies to support them in order to offer them

a decent quality of life. Different interventions have been proposed in the last years to face this challenge,

where recommender systems are gaining strength. The general objective of these systems is to promote the

adoption of healthy habits among the end users, but sometimes they show limitations in the fulfilment of this

goal. To overcome these limitations, our approach offers an easy to maintain, interoperable, and personalized

recommender system capable of providing recommendations based on individuals’ daily activity data. A

methodology is presented for the generation and management of wellbeing recommendations, which are then

tested using a synthetically generated dataset that simulates a variety of user categories. With the evaluation

of this data, a technical validation is carried on to assess the performance and scalability of our developed

system.

1 INTRODUCTION

The current trend of population aging, especially in

developed countries, will pose several challenges to

our society, comprising from changes in the structure

of health and social services, as well as the financial

system and labour markets (Ahtonen, 2012).

Predictions indicate an increase of the burden of age-

related expenditures in state budgets. Increasing life

expectancy comes with a variety of changes in the

care of elderly people as we know it today, with

growing evidence that a shift from targeting

individual diseases to postponing physical

deterioration and comorbidities (Goldman et al.,

2013). Considering that the elderly population is the

group that requires the most from health services and

https://orcid.org/0000-0003-4818-7620

https://orcid.org/0000-0002-7035-3835

https://orcid.org/0000-0002-8194-2467

https://orcid.org/0000-0003-4409-0506

https://orcid.org/0000-0002-1073-7116

https://orcid.org/0000-0003-0248-7783

https://orcid.org/0000-0002-3919-2738

with age is affected by more comorbidities,

interventions that can lead to an improvement of

quality of life that can turn into an increase in the

healthy life-expectancy (Södergren, 2013), which is

the years that a person lives free of disability (Jagger

& Robine, 2011). In return, the burden to the health

services is reduced as the onset of disability is delayed

(Beltrán-Sánchez et al., 2015; Mehta & Myrskylä,

2017), and reduced in a shorter span of time. It is

known that this can be modified by some factors such

as health habits, among others (Fried, 2000).

In this sense, different approaches have been

carried out in the last years to promote wellbeing and

ensure healthy lives for our elderly. Among them, the

most studied approach is probably the use of

recommender systems, which have proven to be

118

Torres, J., Garcia, M., Artola, G., Garcia-Navarro, T., Amaya, I., Larburu, N. and Martin, C.

Wellbeing Recommender System, a User-Centered Framework for Generating a Recommender System for Healthy Aging.

DOI: 10.5220/0011760600003476

In Proceedings of the 9th International Conference on Information and Communication Technologies for Ageing Well and e-Health (ICT4AWE 2023), pages 118-125

ISBN: 978-989-758-645-3; ISSN: 2184-4984

 2023 by SCITEPRESS – Science and Technology Publications, Lda. Under CC license (CC BY-NC-ND 4.0)

useful for this purpose (Ceron-Rios et al., 2017),

being a well-functioning form of offering tailored

health interventions (Azmi et al., 2019b; Park et al.,

2012; Sezgin & Özkan, 2013). Although some of

these systems are already in use, they present several

limitations regarding i) the need of relevant validation

studies to demonstrate their usability (Azmi et al.,

2019a; Martinho et al., 2019), ii) interoperability and

the use of standards for communication with other

clinical platforms (Hors-Fraile et al., 2018), iii) the

need of continuously updating knowledge base

(Berner & Lande, 2016), or iv) user-based

personalization (Rist et al., 2018). In view of all this,

our approach attempts to address these limitations by

presenting an interoperable, technically validated,

and multidisciplinary tool that provides personalized

recommendations to elderly users. The system is

technically validated using stochastic, synthetic data,

which allows to assess the potential benefits of the

presented tool in the target population, the elderly.

This paper presents the methodology used for the

development of our Wellbeing Recommender System

(WRS) and the assessment of its potential to aid users,

describing its components and functionalities, and

showing the steps followed for its technical validation

with synthetic data. In addition, the results and

conclusions of the approach are also presented.

2 MATERIALS AND METHODS

In this section, the methodology followed for the

generation of the dataset used for the validation of the

WRS can be found (Section 2.1), the creation and

management of the recommendations is described

(Section 2.2.1), the data evaluation process by a rule

engine is detailed (Section 2.2.2), the modelling of the

data used by the platform is explained (Section 2.2.3)

and lastly the communication protocols and standards

of the platform are detailed (Section 2.2.4. All these

principal components and the general workflow of the

WRS are represented in the next Figure 1.

2.1 Synthetic Data Generation

In this study, we tested the WRS using a stochastic,

synthetic database named "Synthetic Database for

Recommendation System 2022 (SDRS2022)" which

simulates various user types. In subsection 2.1.1 the

methodology followed to define the categories of

users is explained, and in subsection 2.1.2 the

definition of the metrics and values that were

generated is described.

2.1.1 Elderly Population Characterization

The generated dataset simulates profiles of potential

users of the WRS. With the purpose of obtaining a

real representation of the future users, we based the

data generation on the population groupings defined

in the SHAPES project, whose aim is to build a large-

scale platform EU standardized open platform for

long-term active and healthy aging. Using the

personas method (Pérez-Montoro, 2017) eight

mutually exclusive types of personas were defined:

 Active: older adults between 65-75 years old

with good health and an active lifestyle. They

can be either retired or still working and have

an active social life. Smoking, alcohol and

caffeine consumption are closely related

(Torres-Collado et al., 2018);

 Chronic: older adults from 65 on that suffers

of multimorbid conditions such as diabetes and

oncological disease. Even though their

limitations, they try to maintain their autonomy

and active life (Arnautovska et al., 2018);

 Musculoeskeletal: older adults that suffer

mobility difficulties. They have high risk of

falls and the fear of falling limits their daily life

activities. Sleep duration is associated with the

musculoskeletal pain (Lavigne et al., 2011).

They try to maintain their autonomy but there

is a risk of social isolation (Auais et al., 2018);

 Neurodegenerative lonely elders with

memory decline, isolated from society and with

need of homecare (McCabe et al., 2014). Past

high alcohol consumption habits are related

with the prevalence of neurodegenerative

diseases (Kamal et al., 2020). Sleep patterns

become altered due to the neurodegenerative

process (Owen & Veasey, 2020);

Figure 1: Data execution workflow using the WRS.

Wellbeing Recommender System, a User-Centered Framework for Generating a Recommender System for Healthy Aging

119

 Lonely: older adults socially isolated, with no

support that needs homecare assistance.

Through physical activity, they can reduce their

loneliness feeling (Pels & Kleinert, 2016),

which in turn is closely related to their sleep

quality(Jia & Yuan, 2020);

 Drug dependency: older adults that have high

alcohol consumption habits. They have a high

percentage of hospitalizations and hospital

visits (Choi et al., 2015), and it is considered

that they can maintain a controlled autonomy.

Abuse of alcohol severely affects sleep

quality(Devenney et al., 2019);

 Fragility: older adults over 85 years old with

high fragility suffering from falls, exhaustion

and weight loss (Scheibl et al., 2019) They are

highly dependent and need of a professional or

informal caregiver;

 Deafblind: elders characterized by their

difficulties in socialising. They have a minimal

dependency condition (Bodsworth et al., 2011).

Due to their disability, can develop high

alcohol consumption habits (Fellinger et al.,

2012).

Not only were these categories defined, but also how

belonging to one of them influenced the metrics of the

study. For this, a research was done in the literature

to identify the variables and values needed to model

the categories, as detailed in the following subsection.

2.1.2 Metrics for Synthetic Users

To create stochastic, synthetic data, metrics were used

that represent the habits and lifestyle of the elderly

population. The criteria for these metrics were based

on literature related to physical activity, sleep, and

liquid intake. For physical activity, the elderly were

grouped into five age groups (65 to over 85 years old)

based on the difference in expected daily steps

(Tudor-Locke et al., 2013). For sleep data, the

literature suggests a decline in sleep duration with

aging (Faubel et al., 2009), and a range of less than 7

hours or more than 9 hours was used as a high-risk

symptom or alteration in sleep duration, respectively.

For liquid intake, the elderly tend to consume less

water as they age and in Europe, liquid intake is not

uniform due to differences in beverage categorization

(EFSA Panel on Dietetic Products, Nutrition, and

Allergies (NDA), 2010). The European Food Safety

Authority (EFSA) analyzed reports to create

reference values for liquid intake. The values were

used to make reference values for water, alcohol, and

caffeine-based beverages to monitor daily drinking

behavior and anticipate potential health problems

such as alcoholism and hypertension.

2.2 WRS Components

The WRS design is composed of the following

modules: (i) the wellbeing recommendations

manager to create and maintain the content of the

different recommendations that the system returns,

(ii) the wellbeing rule engine to generate the

personalized recommendations by evaluating user’s

data, (iii) a wellbeing ontology to keep the knowledge

representation homogeneous across the platform and

(iv) an interoperability module that transforms the

input/output data following clinical standard

communication protocols such as FHIR HL7 so that

the system can be integrated with other platforms. In

the next subsections, each component is described

more in detail.

2.2.1 Wellbeing Recommendations Manager

The recommendations included in the WRS are not

limited to a single domain; due to the possible use of

different sources that can be used to gather data, the

format used to store the recommendations in a digital

format is domain independent.

In this paper, the recommendations included are

divided into 3 groups, depending on the target that the

recommendations are associated with. These

categories are (i) liquid intake, (ii) physical activity

and (iii) sleep. For each one, specific

recommendations were defined based on agreed

criteria published and accepted in the literature (see

Section 2.1 for more details). Once identified, the

wellbeing recommendations were modelled into

several rules, which are the knowledge base used by

the rule engine described in subsection 2.2.2.

These rules are defined in a domain independent

format, so that the same representation format can be

re-used in different domains, not only the three

specified ones. Furthermore, the recommendations

can be returned in different languages, thus

broadening the potential user’s population that could

benefit from this solution. The generated

recommendations can be displayed to the end-users in

different devices (i.e., users’ phone, computer, etc),

as the communications to/from the WRS are managed

via a REST API. As a result, if an organization plans

to integrate a recommender system for its users,

employees, etc, it is possible by sending the data in a

HTTP request, as detailed in subsection 2.2.4, and

then visualize the answer as needed, either in a web,

or as notifications of a mobile application for

ICT4AWE 2023 - 9th International Conference on Information and Communication Technologies for Ageing Well and e-Health

120

instance. Lastly, since recommendations may become

obsolete, or new recommendations may want to be

introduced into the system, a web-based rule

authoring tool (AT) was integrated to manage the

formalized content in a simple way by any user. This

tool was developed in (Torres et al., 2020), and was

used to first introduce the rules, and later edit them

when needed. This is a need that is critical for the

adoption of the system as the contents need to be

updated with new evidence otherwise they will

become outdated (Sim et al., 2001).

All the introduction of data was done using the

aforementioned tool, which in conjunction with the

ontology presented in subsection 2.2.3, eased the

introduction of the knowledge into the system, while

the reducing the possibility of input errors of human

origin. An example of the use of the AT for the

formalization of rules is shown in Figure 2, where the

definition of the conditions that conform are rule is

done using the interface of the AT. Individual

conditions or groups of conditions can be defined, as

well as the relation between them. In a similar way,

recommendations are introduced using the same

interfaces.

2.2.2 Wellbeing Rule Engine

The number of rules formalized in the WRS can grow

substantially over time as new content is added into

its knowledge base. As a result of it, a correct

management of the rules and its execution is crucial

to ensure the scalability of the system. As mentioned

before, the base of the WRS is the work presented in

(Torres et al., 2020), where a business rule engine

(Drools) is used to perform the mass execution of the

rules that conform the knowledge base.

Once users’ raw data is received, it is transformed

from the input FHIR message to internal JSON

instances. For liquid intake data, measurements an

aggregation by drink (i.e., water, beer, tea, juice, etc)

and by the type of the drink (alcoholic beverages,

carbonated drinks, caffeine beverages, etc). The use

of business rule engines allows the platform to

perform well, independently of the size of the

knowledge base, guaranteeing the scalability of our

system.

2.2.3 Data Model

The WRS system uses an ontology to keep its

knowledge base homogenous and easy to access for

new rule introduction. The ontology helps reduce

errors in representing the knowledge of the protocols

in rules and the information for both the conditions

and recommendations of the rules is modeled in it.

The ontology can be used by other solutions that need

the same data, as it can be accessed through a REST

API. Additionally, the information can be coded in

different languages to better fit the user's

sociolinguistic profile.

2.2.4 Interoperability

Interventions aiming to improve the habits of the

users can employ one or more component/platform to

support the patient during the intervention process.

As a result of this, the need to interchange data

between different systems becomes a necessary

feature that enhances the interoperability of the

developed platforms. In the present work, the issue of

interoperability was addressed adopting the FHIR

standard developed by HL7 for exchanging clinical

data between systems. All the interchanges of data,

both input and output requests, are done via FHIR

resources. These resources are processed by a parser

component that serializes/deserializes the data

between the FHIR and the internal formats used to

manage the users’ data.

Figure 2: Introduction of conditions using the integrated Authoring Tool.

Wellbeing Recommender System, a User-Centered Framework for Generating a Recommender System for Healthy Aging

121

Although the WRS is presented as a standalone

solution, it is designed to be capable of working in

conjunction with other components, as was the case

in the ecosystem of digital solutions developed under

the SHAPES project, where the interchange of

clinical data between the different components of the

ecosystem is done via, among others, FHIR

resources.

3 RESULTS AND DISCUSSION

With the profiles defined with the persona

methodology and the identified metrics described in

Section 2.1, a dataset of 100 subjects between 65 and

90 years of age has been created, with the following

characteristics (see Table 1).

Table 1: Statistical description of the used dataset.

Variable Mean and SD

Age (years) 78.24 ± 8.4

Gender (%)

53 (Female)

(

Male

)

Physical condition

14 (Active)

10 (Deafblind)

17 (Drug dependency)

12 (Musculoeskeletal)

10 (Lonely)

14 (Fragility)

16 (Chronic)

(

Neurode

enerative

)

Ste

s 4689.14 ± 8434.1

Sleep (hours) 6.08 ± 2.2

Water intake (ml) 1205.07 ± 1012.3

Alcohol intake (ml) 386.58 ± 195.3

Caffeine intake

(

)

933.48 ± 358.5

The process of validation consisted of the

evaluation of the data by the WRS system for each

patient, the storage of the generated

recommendations, and lastly, an analysis of the

correlation between the different profiles of the

generated users and the type of the wellbeing

recommendations obtained.

As a result of the evaluation of the SDRS2022

dataset, a total of 523 recommendations were

obtained. This showed at first that the system was

capable of dealing with high amounts of data in a

reduced amount of time, but it was not easy to check

whether the recommendations obtained were related

with the different patient profiles. An aggregation of

the recommendations based on the profile of the user

that generated them was performed, grouping first the

recommendations belonging to each of the profiles

described in Section 2.1, followed by a categorization

of the rules depending on the nature of its

recommendation. The categorization was different

for each of the wellbeing recommendation categories.

In the case of the sleep recommendations, they were

grouped depending on the number of hours slept. If

the patient slept the recommended hours, it was

considered as a sleep notification, if it was not the

case, it was considered a warning, differentiating

between excess (sleep excess warning), and lack of

slept hours (sleep lack warning). A similar

categorization was done with the activity

recommendations, differentiating between

recommendations when the end-user reached

(activity_notification), and did not reach its daily step

goal (activity_warning). Lastly, in the liquid intake

case, the recommendations were grouped depending

on the class of drink, that is, recommendations

including alcoholic beverages were grouped under

alcohol warnings, and recommendations of

caffeine/theine-containing drinks such as coffee, tea

or carbonated drinks were grouped as caffeine

warnings. The obtained results regarding the

recommendation type and patient profile can be seen

in Figure 3.

The recommendations of the system reflected the

main characteristics of the user profiles, as the highest

amount of alcohol-related recommendations took

place in the Drug dependency category which also

presented over sleep disturbances and low physical

activity as described in Section 2.1. Similarly, the

Fragility users were not able to meet the daily step

target, which can be correlated with their frail

physical condition.

ICT4AWE 2023 - 9th International Conference on Information and Communication Technologies for Ageing Well and e-Health

122

Figure 3: Average number of categorized recommendations grouped by user profile.

Other observed behaviours match Section 2.1.

Sleep disturbances are seen in the Neurodegenerative

group, high physical activity disturbances in Chronic,

and poor sleep quality in Lonely. Active,

Neurodegenerative, and Lonely groups had high

alcohol consumption. Chronic's low average

recommendations could be due to heterogeneity, and

similar recommendations for Active,

Neurodegenerative, and Lonely groups suggest a

need to improve profile modelling to prevent overlap.

Although results show that the recommendations

are in line with the habits described in Section 2.1, a

more real categorization of the patients can be done

by not limiting them to just one category, resulting in

a more accurate approximation of the WRS value.

The technical performance of the system was

evaluated by measuring the time it took to process the

SDRS2022 dataset (parse the FHIR observations,

evaluate data, and return the recommendations in

FHIR format). The evaluation was done using a PC

with 16GB of RAM and an Intel i5-9400F processor

and the results for 100 users are presented in Table 2.

Table 2: Technical characteristics of the evaluation process.

Users 100

Total observations 1500

Number of generated recommendations 523

Recommendations

use

5.23

Evaluation time (ms) 1.5

4 CONCLUSIONS

In this paper we present a rule-based recommender

system (WRS) capable of providing multidomain

wellbeing recommendations for the elderly

population. The aim of this system is to help users to

follow healthy lifestyle habits that will help them

improve their quality of life.

The WRS is integrated with a rule authoring tool

(AT) that allows for updating its knowledge base with

the latest evidence on wellbeing recommendations.

The system is designed to maintain consistency

across its knowledge base by incorporating an

ontology that defines the variables necessary for

defining wellbeing rules.

The system's value was assessed using a

synthetically generated dataset to i) test its

performance with high volumes of data and ii) ensure

proposed interventions matched various potential

end-users. Technical measurements showed the WRS

has no scalability issues as the entire evaluation of the

SDRS2022 dataset took less than two milliseconds,

including serialization/deserialization of FHIR

resources.

The potential impact of the formalized

recommendations was evaluated by analyzing the

triggered recommendations for each user profile. The

analysis showed that the formalized rules are useful

Wellbeing Recommender System, a User-Centered Framework for Generating a Recommender System for Healthy Aging

123

in providing information about users' habits.

However, the dataset used can be improved by not

limiting users to a single category and by improving

the methods of data generation to simulate more

realistic users.

Future research will develop a methodology to

analyze behavioral changes over time. The WRS

conforms to clinical standards, and new features can

be added easily to improve the impact on the end-

user’s life. The generation of synthetic data will also

be enhanced to simulate users' habits and enable

multi-category inclusion. This will allow to test the

methodology for detecting changes in behavior.

ACKNOWLEDGEMENTS

This project has received funding and clinical

professionals’ advice by gewi-Institut für

Gesundheitswirtschaft e.V. under the European

Union’s Horizon 2020 research and innovation

programme under Grant Agreement No 857159. 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

Ahtonen, A. (2012). Healthy and active ageing. 4.

Arnautovska, U., O’Callaghan, F., & Hamilton, K. (2018).

Behaviour change techniques to facilitate physical

activity in older adults: What and how. Ageing and

Society, 38(12), 2590–2616.

Auais, M., French, S., Alvarado, B., Pirkle, C., Belanger,

E., & Guralnik, J. (2018). Fear of Falling Predicts

Incidence of Functional Disability 2 Years Later: A

Perspective From an International Cohort Study. The

Journals of Gerontology: Series A, 73(9), 1212–1215.

Azmi, A. K., Abdullah, N., & Emran, N. A. (2019a). A

Collaborative Filtering Recommender System Model

for Recommending Intervention to Improve Elderly

Well-being. International Journal of Advanced

Computer Science and Applications (IJACSA), 10(6),

Article 6.

Azmi, A. K., Abdullah, N., & Emran, N. A. (2019b). A

Recommender System Model for Improving Elderly

Well-Being: A Systematic Literature Review. 11, 22.

Beltrán-Sánchez, H., Soneji, S., & Crimmins, E. M. (2015).

Past, Present, and Future of Healthy Life Expectancy.

Cold Spring Harbor Perspectives in Medicine, 5(11),

a025957.

Berner, E. S., & Lande, T. J. L. (2016). Overview of

Clinical Decision Support Systems. In Clinical

Decision Support Systems (pp. 1–17). Springer, Cham.

Bodsworth, S. M., Clare, I. C. H., Simblett, S. K., &

Deafblind UK. (2011). Deafblindness and mental

health: Psychological distress and unmet need among

adults with dual sensory impairment. British Journal of

Visual Impairment, 29(1), 6–26.

Ceron-Rios, G. M., Lopez-Gutierrez, D. M., Dıaz-Agudo,

B., & Recio-Garcıa, J. A. (2017). Recommendation

System based on CBR algorithm for the Promotion of

Healthier Habits.

Choi, N., Marti, C., DiNitto, D., & Choi, B. (2015). Alcohol

Use as Risk Factors for Older Adults’ Emergency

Department Visits: A Latent Class Analysis. Western

Journal of Emergency Medicine, 16(7), 1146–1158.

Devenney, L. E., Coyle, K. B., Roth, T., & Verster, J. C.

(2019). Sleep after Heavy Alcohol Consumption and

Physical Activity Levels during Alcohol Hangover.

Journal of Clinical Medicine, 8(5), Article 5.

EFSA Panel on Dietetic Products, Nutrition, and Allergies

(NDA). (2010). Scientific Opinion on Dietary

Reference Values for water. EFSA Journal, 8(3).

Faubel, R., López-García, E., Guallar-Castillón, P.,

Graciani, A., Banegas, J. R., & Rodríguez-Artalejo, F.

(2009). Usual sleep duration and cognitive function in

older adults in Spain: Association between usual sleep

duration and cognitive function. Journal of Sleep

Research, 18(4), 427–435.

Fejer, R., & Ruhe, A. (2012). What is the prevalence of

musculoskeletal problems in the elderly population in

developed countries? A systematic critical literature

review. Chiropractic & Manual Therapies, 20(1), 31.

Fellinger, J., Holzinger, D., & Pollard, R. (2012). Mental

health of deaf people. Lancet (London, England),

379(9820), 1037–1044.

Fried, L. P. (2000). Health Promotion for Older Adults:

What Is the Potential? 26.

Goldman, D., Cutler, D., Rowe, J. W., Michaud, P.-C.,

Sullivan, J., Peneva, D., & Olshansky, S. J. (2013).

Substantial Health and Economic Returns From

Delayed Aging May Warrant a New Focus for Medical

Research. Health Affairs (Project Hope), 32(10), 1698–

1705.

Hors-Fraile, S., Rivera-Romero, O., Schneider, F.,

Fernandez-Luque, L., Luna-Perejon, F., Civit-Balcells,

A., & de Vries, H. (2018). Analyzing recommender

systems for health promotion using a multidisciplinary

taxonomy: A scoping review. International Journal of

Medical Informatics, 114, 143–155.

Jagger, C., & Robine, J.-M. (2011). Healthy Life

Expectancy. In R. G. Rogers & E. M. Crimmins (Eds.),

International Handbook of Adult Mortality (pp. 551–

568). Springer Netherlands.

Jia, G., & Yuan, P. (2020). The association between sleep

quality and loneliness in rural older individuals: A

cross-sectional study in Shandong Province, China.

BMC Geriatrics, 20(1), 180.

Kamal, H., Tan, G. C., Ibrahim, S. F., Shaikh, Mohd. F.,

Mohamed, I. N., Mohamed, R. M. P., Hamid, A. A.,

ICT4AWE 2023 - 9th International Conference on Information and Communication Technologies for Ageing Well and e-Health

124

Ugusman, A., & Kumar, J. (2020). Alcohol Use

Disorder, Neurodegeneration, Alzheimer’s and

Parkinson’s Disease: Interplay Between Oxidative

Stress, Neuroimmune Response and Excitotoxicity.

Frontiers in Cellular Neuroscience, 14.

Lavigne, G. J., Nashed, A., Manzini, C., & Carra, M. C.

(2011). Does sleep differ among patients with common

musculoskeletal pain disorders? Current Rheumatology

Reports, 13(6), 535–542.

Martinho, D., Carneiro, J., Novais, P., Neves, J., Corchado,

J., & Marreiros, G. (2019). A Conceptual Approach to

Enhance the Well-Being of Elderly People. In P. Moura

Oliveira, P. Novais, & L. P. Reis (Eds.), Progress in

Artificial Intelligence (pp. 50–61). Springer

International Publishing.

McCabe, C., Dinsmore, J., Brady, A. M., Mckee, G.,

O’Donnell, S., & Prendergast, D. (2014). Using Action

Research and Peer Perspectives to Develop Technology

That Facilitates Behavioral Change and Self-

Management in COPD. International Journal of

Telemedicine and Applications, 2014, 1–10.

Mehta, N., & Myrskylä, M. (2017). The Population Health

Benefits Of A Healthy Lifestyle: Life Expectancy

Increased And Onset Of Disability Delayed. Health

Affairs, 36(8), 1495–1502.

Owen, J. E., & Veasey, S. C. (2020). Impact of sleep

disturbances on neurodegeneration: Insight from

studies in animal models. Neurobiology of Disease,

139, 104820.

Park, D. H., Kim, H. K., Choi, I. Y., & Kim, J. K. (2012).

A literature review and classification of recommender

systems research. Expert Systems with Applications,

39(11), 10059–10072.

Pels, F., & Kleinert, J. (2016). Loneliness and physical

activity: A systematic review. International Review of

Sport and Exercise Psychology, 9(1), 231–260.

Pérez-Montoro, M. (2017). Navigation design and SEO for

content-intensive websites: A guide for an efficient

digital communication. Elsevier.

Rist, T., Seiderer, A., & André, E. (2018). Providing Life-

Style-Intervention to Improve Well-Being of Elderly

People. In E. Clua, L. Roque, A. Lugmayr, & P. Tuomi

(Eds.), Entertainment Computing – ICEC 2018 (pp.

362–367). Springer International Publishing.

Scheibl, F., Farquhar, M., Buck, J., Barclay, S., Brayne, C.,

& Fleming, J. (2019). When Frail Older People

Relocate in Very Old Age, Who Makes the Decision?

Innovation in Aging, 3(4), igz030.

Sezgin, E., & Özkan, S. (2013). A systematic literature

review on Health Recommender Systems. 2013 E-

Health and Bioengineering Conference (EHB), 1–4.

Sim, I., Gorman, P., Greenes, R. A., Haynes, R. B., Kaplan,

B., Lehmann, H., & Tang, P. C. (2001). Clinical

Decision Support Systems for the Practice of Evidence-

based Medicine. Journal of the American Medical

Informatics Association

, 8(6), 527–534.

Södergren, M. (2013). Lifestyle predictors of healthy

ageing in men. Maturitas, 75(2), 113–117.

Torres, J., Artola, G., & Muro, N. (2020). A Domain-

Independent Semantically Validated Authoring Tool

for Formalizing Clinical Practice Guidelines. Studies in

Health Technology and Informatics, 270, 517–521.

Torres-Collado, L., García-de la Hera, M., Navarrete-

Muñoz, E. M., Compañ-Gabucio, L. M., Gonzalez-

Palacios, S., & Vioque, J. (2018). Coffee Drinking and

Associated Factors in an Elderly Population in Spain.

International Journal of Environmental Research and

Public Health, 15(8), Article 8.

Tudor-Locke, C., Schuna, J. M., Barreira, T. V., Mire, E.

F., Broyles, S. T., Katzmarzyk, P. T., & Johnson, W. D.

(2013). Normative Steps/Day Values for Older Adults:

NHANES 2005-2006. The Journals of Gerontology

Series A: Biological Sciences and Medical Sciences,

68(11), 1426–1432.

Wellbeing Recommender System, a User-Centered Framework for Generating a Recommender System for Healthy Aging

125