The Role of Embodied Conversational Agents in Supporting Older

Adults with Hypertension: A Focus Group Study

Julio Oliveira

, Telmo Silva

, Rita Oliveira

and Elizabeth Furtado

Digimedia, Universidade de Aveiro, Aveiro, Portugal

Doutorado em Informática Aplicada, Universidade de Fortaleza, Fortaleza, Brazil

Keywords: Embodied Conversational Agents, Hypertension, Focus Group, Non-Hedonic Applications.

Abstract: This paper investigates the essential tasks for Embodied Conversational Agents (ECA) to engage Older Adults

in hypertension treatment effectively. The research addresses the growing need for innovative solutions to

support aging populations and explores the potential of ECA to provide cognitive, emotional, and practical

support. The central research question focuses on identifying the key tasks an ECA should perform to assist

Older Adults in managing their hypertension. Through a Focus Group (FG) with healthcare professionals, the

study identifies ten key tasks for ECA support, encompassing pharmacological (medication adherence and

tracking) and non-pharmacological (dietary monitoring, exercise promotion, disease education, and

communication facilitation) aspects of hypertension management. The FG was held online with seven

participants (5 participants and 2 User Experience Specialists) in January 2023. The paper also discusses the

broader challenges and considerations related to ECA engagement for Older Adults, including ethical

concerns, user acceptance, and the importance of personalized and empathetic interactions. The findings

emphasize the need for a tailored ECA design that considers individual patient needs, medical history, and

preferences to maximize effectiveness and promote successful hypertension management.

1 INTRODUCTION

The growing prevalence of the world's ageing

population has intensified the need for innovative

solutions that promote independence, well-being, and

social engagement among Older Adults (OA).

Conversational Systems (CS) with Embodied

Avatars, namely Embodied Conversational Agents -

ECA, have emerged as a promising - and low-cost -

technology to provide cognitive, emotional, and

practical support for this demographic group

(Yaghoubzadeh et al., 2013).

From the perspective of Embodied

Conversational Agents (ECA), it is important to

identify issues like features, recommendations, and

characteristics of the agent and the dialogue

specifically tailored for Older Adults. The primary

challenge lies in identifying and relating the issues

https://orcid.org/0000-0002-3516-0114

https://orcid.org/0000-0001-9383-7659

https://orcid.org/0000-0001-6041-9469

https://orcid.org/0000-0002-1584-3161

associated with this demographic to practical

applications.

Our central research question (RQ) guiding this

inquiry is: “What are the essential tasks for the use of

ECA to engage Older Adults in the Treatment of

Hypertension?”.

From that RQ, this paper seeks to delineate the

main challenges encountered in engaging Older

Adults in using ECA and then identify tasks to be

done with the help of the ECA. The findings highlight

the challenges of identifying an effective tailor for

engaging Older Adults.

The contribution of this work is related to the

development of ECA for the elderly, considering

aspects of health care. The selection of important

tasks can promote an improvement in the

unpredictability of dialogs. The current use of Large

Language Models (LLM) makes it necessary to create

Oliveira, J., Silva, T., Oliveira, R. and Furtado, E.

The Role of Embodied Conversational Agents in Supporting Older Adults with Hypertension: A Focus Group Study.

DOI: 10.5220/0013504300003938

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

In Proceedings of the 11th International Conference on Information and Communication Technologies for Ageing Well and e-Health (ICT4AWE 2025), pages 433-442

ISBN: 978-989-758-743-6; ISSN: 2184-4984

433

prompts that allow greater engagement in the tasks

described.

This paper is organized into four sections. The

next section explores the challenges related to ECA

development, particularly for Older Adults. The third

section examines applications designed to support

Older Adults in non-hedonic contexts. Following this,

we present a dedicated section outlining ten essential

tasks that the ECA should undertake to aid in

managing hypertension treatment. Finally, we

conclude with remarks on future work.

2 OLDER ADULT’S

ENGAGEMENT

The growing adoption of voice technologies,

particularly among the Older Adult population, raises

several ethical issues that require in-depth analysis.

Interaction with conversational voice systems, such

as virtual assistants and chatbots, offers significant

benefits, such as easier access to information and

virtual companionship. However, it is crucial to

consider the ethical aspects inherent in this

relationship to guarantee a safe, respectful, and

beneficial experience for the Older Adult (Patrão

Neves, 2022).

Engagement has been a topic of debate across

various domains, including education, management,

marketing, and social media. It explores how

individuals connect with activities, content, or

communities, influencing their behavior and

experience. Engagement is not just about producing

meaning but also about impacting behavior through

specific patterns. It involves a dynamic interaction

between the user and the context, where cognitive and

emotional responses play crucial roles. This

complexity makes it challenging to define

engagement in a universally applicable manner, as it

is deeply rooted in subjective experiences and

context-specific factors (Zagalo, 2020).

In the field of Human-Computer Interaction

(HCI), engagement has been increasingly recognized

as crucial for designing interactive experiences. It

moves beyond mere usability and functionality to

create meaningful and purposeful interactions.

Engagement is linked to motivational patterns that

drive users to participate and interact continuously,

making it a key factor in maintaining user interest and

satisfaction. It involves three main streams:

progression, which relates to achieving goals;

expression, which involves creativity and self-

expression; and relation, which encompasses social

connections and emotional bonds. This triadic

approach provides a comprehensive framework for

understanding and designing engaging experiences

(Zagalo, 2020).

The measurement of engagement in this study

will be conducted through a motivational

interview(Mercado et al., 2023) with Brazilian and

Portuguese Older Adults. The experiment will utilize

videos of the ECA in conjunction with the scenarios

described in this paper.

ECA collects vast amounts of personal data to

function effectively, including voice recordings,

command history, device information, and even

biometric data such as voiceprint. This data's

indiscriminate collection and long-term storage pose

a significant risk to users' privacy.

Protecting the privacy of elderly users' is one big

challenge related to storing sensitive personal

information, such as health history and lifestyle

habits, and requires robust security measures to

prevent unauthorized access and misuse. Developers

must be transparent about their data collection and

use practices, guaranteeing users' informed consent

(Hadian et al., 2019).

The vulnerability of the Oldeer Adults, often

associated with loneliness and technological

dependence, makes them potential targets for

manipulation and exploitation. We need to be aware

of the risk of using conversational systems for

misleading advertising, selling inappropriate

products and services, or even perpetrating crimes

(Bickmore et al., 2018).

The growing prevalence of the world's ageing

population has intensified the need for innovative

solutions that promote independence, well-being, and

social engagement among Older Adults. ECA have

emerged as a promising - and low-cost - technology

to provide cognitive, emotional, and practical support

for this demographic group (Yaghoubzadeh et al.,

2013).

ECA are designed for natural interaction in

spoken and non-verbal language with the potential to

aid daily tasks, facilitate communication, and even

provide companionship. However, the success of

these technologies depends heavily on ensuring user

adoption and acceptance, a complex issue shaped by

several factors, particularly how dialogues are staged

(Knob et al., 2021).

Creating a successful conversational system for

Older Adults requires a holistic understanding of their

needs and preferences beyond technological

functionality. Several critical factors and Challenges

influence Older Adults' acceptance of and adherence

to conversational systems.

IS4WB_SC 2025 - Special Session on Innovative Strategies to Enhance Older Adults’ Well-being and Social Connections

434

These aspects are not unanimous, but the ones

most often cited in the literature are related below.

Adapting to the needs of Older Adults is crucial

for effective interaction. Agents designed to engage

with human users as interlocutors must be able to

adjust. Research in this area has focused on

developing conversational agents, including virtual

agents, robots (physical agents), and chatbots to

adjust to the users' preferences. These preferences

occur at various levels and are expressed through

different social cues, including verbal and nonverbal

communication and conversational strategies. (Woo

et al., 2024)

To successfully engage in real-time and

demonstrate expressive and adaptive behavior, an

ECA must possess essential functionalities, such as

Perception of the user’s behavior, Agent’s behavior,

and Dialog management.

2.1 Perception of the Older Adult’s

Behavior

An agent interacting with a human user must respond

appropriately to the user, considering their behavior,

including speech and gestures. To achieve this, the

agent must first perceive these signals, which can be

categorized into three components: the content of the

user's speech, the prosody of the speech, and the

user's gestures (Woo et al., 2024).

The content of a user's speech is essential as it

clearly communicates their intentions through

language. This information is crucial for various

automated systems that interact with users, as

evidenced by conversational Artificial Inteligence

(AI) assistants such as Google, Alexa, and Siri. To

effectively capture this speech content, it is necessary

to implement Automatic Speech Recognition (ASR),

also referred to as Speech-to-Text (STT). ASR is a

technology that transcribes spoken audio into written

text, recognizing complete phrases and converting

them in real-time as the user speaks (Woo et al.,

2024)..

User's speech prosody plays a crucial role in

expressing intentions that go beyond mere spoken

words. It includes variations in voice qualities such as

pitch (high or low), loudness (loud or soft), and

duration (fast or slow), all of which are collectively

known as speech prosody. To extract these prosodic

features, audio feature extraction must be performed.

The primary prosodic characteristics identified

include fundamental frequency, which indicates

pitch; loudness, which measures sound energy;

voicing probability, assessing the ratio of unvoiced to

voiced energy; and Mel-frequency Cepstral

Coefficient (MFCC), representing the short-term

power spectrum of sound (Woo et al., 2024).

The user’s gesture needs to be captured to

analyze the perception and generation of facial

gestures, including facial expressions and head/gaze

movements, of both the human user and the agent.

The key features include gaze movements

represented by angles concerning the x and y axes,

head movements characterized by Euler rotations

around the x, y, and z axes, and facial expressions

identified through Action Units (AUs) as defined by

the Facial Action Coding System (FACS) (Ekman &

Friesen, 2019; Woo et al., 2024).

2.2 Agent’s Behavior

The agent's behavior is intricately linked to its ability

to adapt based on both user interactions and its own

historical behaviors. This dual dependency

necessitates a sophisticated perception mechanism

that allows the agent to monitor and evaluate its

actions in relation to its current intentions, such as

synchronizing lip movements with speech output.

Integrating these elements allows the agent to

enhance responsiveness and create a more coherent

interaction experience.

The challenge to achieving this level of self-

awareness is linked to an internal memory system that

retains a record of its most recent behaviors, including

speech content and prosody. This memory needs to

store a large number of instances of user behavior but

also tracks the agent's own displayed behaviors,

thereby facilitating a comprehensive understanding

of its performance and enabling it to adjust its actions

accordingly (Woo et al., 2024; Woo, Grimaldi, et al.,

2023).

The behavior signals perceived by human users

and agents play a crucial role in generating expressive

and adaptive behaviors in artificial agents. The

Augmented Self-Attention Pruning (ASAP) model,

as introduced by Woo et al. (2023), is designed to

create expressive facial gestures for agents that adapt

reciprocally to their interactions. The ASAP model

employs self-attention pruning and data

augmentation techniques to enhance learning

capabilities (Woo, Pelachaud, et al., 2023).

A significant feature of the ASAP model is its

ability to ensure continuity in the generated

behaviors. This is achieved through autoregressive

adaptive online prediction, which allows the model to

maintain a seamless flow in the agent's actions.

(Woo, Pelachaud, et al., 2023).

Integrating the pre-trained ASAP model into the

system significantly advances the rendering of

The Role of Embodied Conversational Agents in Supporting Older Adults with Hypertension: A Focus Group Study

435

expressive and adaptive visual behaviors in agents.

By processing the perceived visual and audio features

from the previous steps of both the human user and

the agent, the model can predict and generate the

agent's behavior for the subsequent time step. This

predictive capability is essential for creating a more

interactive and engaging user experience (Woo,

Pelachaud, et al., 2023).

2.3 Dialog Management

Effective management of dialogue in conversations

relies on a turn-taking mechanism that allows

interlocutors to influence the direction of the

discussion based on the content of their speech. For

an agent to successfully engage in real-time

conversations with human users, it is essential to

replicate this turn-taking behavior.

The process initiates with an automatic speech

recognition (ASR) system that captures the user's

spoken utterance. This captured text is subsequently

transmitted to a dialog engine. By leveraging this

framework, the dialog engine can effectively analyze

the input and determine the most suitable next step in

the conversation, promoting a seamless interaction

between the agent and the user.

There are three key reasons for considering the

inclusion of non-verbal modalities in the dialog. The

first reason is rooted in human factors. In face-to-face

communication, regardless of our language, cultural

background, or age, we all incorporate our facial

expressions and hand gestures as essential

components of our communication. The additional

factors stem from the interplay between computers

and human users. The second concerns the need for

human access to more than one modality in noisy

situations. In situations like real life, non-verbal

modalities come into play: a smile, a shoulder

movement, and so on (Cassell, 1998).

The challenge is to allow the agent to make

gestures and the Older Adult consider it as a part of

the dialog. Cassell suggests the use of Emblems, like

the V of victory, and Propositional Gestures, like the

use of the hand to show the size of a cup. The third

type is Spontaneous Gestures, such as pointing to a

slide, for example (Cassell, 1998).

The integration of gesture and speech is a crucial

consideration. At the word level, it is essential to

temporally align both individual gestures and spoken

words, ensuring that the most dynamic aspect of the

gesture coincides with or precedes the syllable that

carries the greatest intonational emphasis in the

accompanying speech segment. This alignment

presents one of the challenges in achieving effective

integration (Cassell, 1998).

When the Agent articulates a word, various

physical actions need to be made, such as eye blinks,

hand movements, head turns, and brow raises take

place, concluding at the end of the word. This

synchrony is evident across all levels of speech,

including phonemic segments, words, phrases, and

longer utterances (Cassell, 1998). The reality of these

movements can facilitate communication with Older

Adults or give them a false clue about what is

happening.

Although it is crucial for promoting trust and

involvement, it can cause serious problems if

implemented improperly. The main difficulties

identified are related to the subtleties of human

communication and cultural diversity among those

involved in contingency and responsiveness (Kopp &

Hassan, 2022).

Modeling verbal behavior with the nuances of

human emotion is quite complex (Yaghoubzadeh et

al., 2015). However, these subtleties must not be

exaggerated or underestimated, and the system must

respond to the user's needs and current state.

Contextual factors and individual preferences

must be considered to create and implement empathy

through dialogue. Generic approaches can lead to

frustration. Therefore, the system's response and the

level of empathy must be personalized (Yalcin,

2019).

3 NON-HEDONIC

APPLICATIONS

Non-hedonic applications for Older Adults refer to

tools, technologies, and services designed to enhance

their quality of life without focusing primarily on

pleasure or enjoyment. These applications aim to

support daily living, promote health, and foster social

connections (Zhang & Umemuro, 2013).

One of the applications of ECAs for Older Adults

is in the realm of companionship. Loneliness and

social isolation are prevalent issues among the

elderly, often leading to detrimental effects on mental

and physical health (Bousardt, 2022).

Loneliness is a common problem among the Older

Adults, and conversational systems can also offer

virtual companionship through friendly

conversations, games, and hedonic activities

(Esposito et al., 2021). In addition, they can provide

information on cultural events, news, and other

activities of interest. Conversational systems can

IS4WB_SC 2025 - Special Session on Innovative Strategies to Enhance Older Adults’ Well-being and Social Connections

436

offer online courses and training, stimulating

continuous learning and the development of new

skills (Da Paixao Pinto et al., 2021).

ECA can provide a sense of presence and

engagement, offering conversation and interaction

that can alleviate loneliness. By utilizing Natural

Language Processing (NLP) and emotional

recognition, these agents can adapt their responses to

the emotional state of the user, fostering a more

personalized and meaningful interaction. This

companionship can benefit those with limited social

networks or mobility constraints (Kramer et al.,

2022).

However, there is inconsistent evidence on

whether intention translates into actual usage. Since

actual use is essential for realizing health benefits, it

occupies a central role in the model. Increased

engagement with ECA is anticipated to enhance the

quality of the relationship formed, leveraging the

agents' ability to foster empathy. The model posits

that usability and perceived usefulness serve as

precursors to actual use, with improved usability

leading to heightened perceived usefulness (Kramer

et al., 2022).

The classification system proposed by Teixeira et

al. (2012) is utilized to generate hypotheses regarding

which techniques effectively enhance specific

psychological needs. The study's objectives include

determining whether ECA can motivate community-

dwelling Older Adults to modify their dietary habits

and reduce feelings of loneliness, assessing the

mechanisms behind these outcomes, and gaining

insights into the usage of ECA (Teixeira et al., 2012).

The health sector stands out as one of the main

fields of application. These tools can help monitor

chronic conditions, remind Older Adults to take

medication, schedule medical appointments, and

provide information on healthy lifestyle habits

(“Telehealth Innovations in Remote Healthcare

Services Delivery - Global Telehealth 2020,” 2021).

In Health Care, the challenge is related to the type

of content, the security of the information, and the

presence of functionalities related to the collection of

biomedical data. The challenges lie in supporting the

Older Adult with simple medical treatments, whether

drug treatment or lifestyle-related. The use, however,

needs to be more widely accepted to make treatments

effective and assist the medical team in treating

chronic diseases. Even though ECA can make life

easier for Older Adults, they need to trust and

establish a relationship with the system.

The study performed by Car et al. (2020)

identified several different themes: Treatment and

Monitoring, Health Care Services Support,

Education, Lifestyle Behavioral Changes, and

Diagnosis. The healthcare sectors identified for

conversational agent applications were generally

broad, mentioning a few specialties, including mental

health, neurodegeneration, metabolic medicine (such

as obesity and diabetes), and sexual health. Future

applications could expand into other healthcare fields

with the potential for digital health interventions,

such as dermatology, primary care, geriatrics, and

oncology (Car et al., 2020).

ECAs can significantly contribute to the

management of health for senior individuals. Many

Older Adults face chronic health conditions requiring

ongoing monitoring and management. ECAs can

assist in medication reminders, health tracking, and

providing information about medical conditions. By

engaging users in conversations about their health,

these agents can encourage adherence to treatment

plans and promote healthier lifestyle choices.

Furthermore, the ability of ECAs to collect and

analyze data on user behavior can provide valuable

insights for healthcare providers, enabling more

tailored and effective interventions.

Another significant application of ECAs is in

enhancing cognitive engagement among Older

Adults. Cognitive decline is a common concern as

individuals age, and maintaining mental acuity is

essential for overall well-being. ECAs can facilitate

cognitive exercises and games that stimulate mental

activity, helping to keep the mind active and engaged.

By providing a fun and interactive platform for

cognitive training, these agents can motivate Older

Adults to participate in activities that promote

cognitive health. Additionally, the social interaction

provided by ECAs can further enhance cognitive

function by encouraging communication and

socialization.

ECA can serve as a bridge to technology for Older

Adults, who may feel intimidated by modern devices

and applications. ECAs can help older individuals

navigate technology more comfortably by providing

a user-friendly interface that mimics human

interaction. This can increase engagement with digital

resources like telehealth services, online

communities, and educational platforms. Older adults

can access a wealth of information and services to

improve their quality of life as they become more

adept at using technology. Applying Embodied

Conversational Agents to Older Adults presents a

multifaceted approach to addressing this

demographic's unique challenges, ultimately

enhancing their social, emotional, and cognitive well-

being.

The Role of Embodied Conversational Agents in Supporting Older Adults with Hypertension: A Focus Group Study

437

ECA, with special sensors and wearable devices,

can track vital signs, medication adherence, and

physical activity levels. These interactive tools may

assist Older Adults in managing chronic conditions

and proactively maintaining their health.

Remote consultations by telehealth can facilitate

access to healthcare professionals, by enabling Older

Adults to receive medical advice and treatment

without the need to leave their houses, thereby

improving access to Health Care.

The applications can be designed to stimulate

cognitive function through interactive puzzles,

memory games, and problem-solving tasks using

voice and help them maintain mental acuity and delay

cognitive decline.

The possibility of communication with family and

friends through video calling and social media

platforms can be made by supporting ECA to combat

loneliness and promote social engagement.

The offer of smart home technologies to Older

Adults, including fall detection systems, emergency

response buttons, and environmental sensors, ensures

a safe living environment and can be connected to

family members and the health care system.

The assistance of alimentation habits with the

possibility of meal planning and grocery delivery

services tailored to the dietary needs of Older Adults,

ensuring access to healthy and balanced meals, can be

held with the support of nutrition professionals.

Daily exercises specifically designed for Older

Adults can be performed by the ECA, promoting

physical activity and helping to maintain mobility and

strength.

By focusing on these applications of embodied

conversation agents, we can create a supportive

ecosystem that enhances the well-being of Older

Adults, allowing them to live independently and with

dignity.

4 TASKS FOR THE TREATMENT

OF HYPERTENSION

A focus group was conducted to assess the utilization

of ECA among Older Adults within a healthcare

environment. The online session took place in

January 2023 and included five participants, as

detailed in Table 1, along with two UX specialists.

The session lasted a total of 80 minutes.

The criteria used to recruit of participants was by

convenience. Participant (P) P1 is a researcher at

Aveiro University and is a member of a research

project involving Older Adults and technology. P2

has a Master's degree and is in the medical residency

period. P3 is a professor at the University School of

Health, and P4 is also a teacher at the University in

Brazil and works in a government health center. P5

worked in public hospitals in São Paulo, Brazil, and

is currently on a sabbatical leave. All participants

agreed to have their records taken for research

purposes.

Table 1. Participants of Focus Group.

Sub

ect Rule Gender, Countr

P1 Gerontolo

ist Female, PT

P2 Recentl

raduated

sician Female, PT

P3 Ph

siothera

ist Male, PT

P4 Cardiolo

ist Female, BR

P5 Geriatric Ph

sician Male, BR

The mediator introduced everyone and requested

authorization to record audio from the FG session.

The discussion was legitimized by explaining the

context of the research and the Research Project being

conducted. The research question, the context of the

study, and the objectives were also presented.

After that, the key concepts of ECAs - Embodied

Conversational Agents, Treatment of Hypertension,

and Older People were shared with participants. The

goal of FG was then, presented:

"The use of ECA by Older Adults with

Hypertension can be realized; If so, what

tasks can assist the patients?"

A video demonstrating an example of an ECA

dialogue was presented to remind an elderly woman

to take her heart medication. The actress featured in

the video was selected for convenience, ensuring that

the identified tasks were relevant to technologically

literate seniors facing challenges with self-managing

their hypertension treatment. Additionally, the actress

signed a document granting permission for her

appearance to be shown to the focus group

participants.

After the video, participants started with the

considerations about the use of ECA using the

Persona shown in Figure 1. Then, the mediator

opened the space for each participant to make

considerations about the theme and indicate the tasks

that the assistant could support.

The method used to select the tasks involved

brainstorming. The facilitator began by explaining the

goal and inquired about what ECA should and should

not do. Each participant contributed a set of tasks,

leading to a discussion. As the tasks were presented,

some were excluded while others emerged. The input

from medical participants highlighted certain risks

and the need to avoid specific tasks.

IS4WB_SC 2025 - Special Session on Innovative Strategies to Enhance Older Adults’ Well-being and Social Connections

438

Magda Barbosa | 76yo | Single

“I like Technology, and I use it to

improve my daily life”

She lives alone, can't remember to

take medication, loves salty food,

and she is sedentary.

Has a Google Nest to listen to

music and ask for forecasts daily.

She writes down some sticky notes

to remind herself of her

medications.

Figure 1. Persona to be considered to use an ECA.

The UX specialists actively engaged in the

discussion by assessing the usability of the ECA in

executing the proposed tasks. They provided support

to the medical participants, and the diverse insights

shared underscored the necessity to refine the

objectives and methodologies for each task to

promove engagement.

A total of ten tasks were identified that could

benefit from ECA support, which included two

pharmacological tasks (T1 and T2) and eight non-

pharmacological tasks (T3 to T10). The proposed

tasks are detailed in Table 2.

Task 1 involves following up on medications,

while Task 2 focuses on tracking the amount of each

medication, as suggested by P1 and P3. However, it

is important to avoid using typical alarms to remind

Older Adults. Instead, a more empathetic dialogue

should be established to achieve the desired goal.

The third task, while categorized as non-

pharmacological, involves the medical consultation.

Participants noted the challenges Older Adults face in

remembering all the information provided by the

doctor. Important details may be overlooked as the

sessions extend. A significant advancement is the

utilization of comprehensive records of all

recommendations, which can and should be leveraged

to inform other tasks, whether to determine the type

of advice to offer or to issue alerts.

Other tasks can assist Older Adults in staying on

track. Regular monitoring of blood pressure (T5) is

essential, as it helps maintain heart health and identify

potential issues early. Furthermore, proposing and

following up on physical activities (T6) can

significantly enhance overall well-being; simple

exercises such as walking or stretching can make a

substantial difference. The frequency of blood

pressure monitoring and the amount of exercise

should be coordinated with the health team’s

guidance. The cardiologist emphasized that an

analysis of all comorbidities must be carried out

before defining an exercise plan for a patient with

hypertension.

Suggest and assist in developing comprehensive

diets that are nutrient-dense, as proper nutrition is

crucial for sustaining health (T7). Participants

cautioned against utilizing the Medical Record (MR)

solely to offer recommendations that do not consider

other comorbidities. This task is bound by clear

ethical constraints, as the information contained in

medical records may not always be accessible on

platforms beyond hospital settings (Xie et al., 2018).

The Task number 8 identify daily activities that

may impact your treatment. To inform of how certain

habits or routines affect the health helps to make

better choices. However, the participants were not all

convinced that this task could be performed by a

computer device. The inherent risks of an elderly

person's routine must be considered on a case-by-case

basis and cannot be performed based on random

suggestions or suggestions of dubious origin.

Continuous learning from tasks that are already

performed and suggestions made by the medical team

can make this complex task more feasible.

One aspect reported in connection with this task is

related to a possible characteristic of the ECA in

monitoring the elderly person in the tasks they

perform and reacting autonomously without the

elderly person's request. From the participants' point

of view, the elderly person will not authorize or feel

comfortable, if they do authorize, for the equipment

to monitor them. Therefore, the only way would be to

question the user and analyse what they say.

Disease literacy is crucial for effective treatment.

Task 9 focuses on delivering relevant news to Older

Adults concerning HTA health treatment.

Suggestions can be made through discussions on

various topics, and within the bounds of medical

advice, these can help prevent misguided actions.

Participants in the focus group are particularly

mindful of the information level that needs to be

conveyed to ensure it is effective for each patient's

medical condition.

General information can lead to confusion and

may impede patients' decision-making.

Consequently, the proposed task should be

approached with care, taking into account the

knowledge, culture, and educational background of

each individual user. Coronary risks identified by

Participant P4 must be considered to prevent the

establishment of inappropriate guidelines.

Additionally, it is essential to recognize the

significant risks associated with other comorbidities

on an individual level before developing a specific

scenario.

The Role of Embodied Conversational Agents in Supporting Older Adults with Hypertension: A Focus Group Study

439

The last task proposed for the ECA is related to

keeping the elderly adult connected with the entire

care team, whether formal or informal. The

possibility of opening direct video connections with

nurses, caregivers, relatives and even doctors can

prevent, among other things, emergency situations

from being ignored.

A set of assumptions were reported by the

participants: the availability of the medical team, the

level of knowledge of caregivers and relatives and

also ethical and safety aspects related to video

contacts. They unanimously indicated that this

function can enhance user engagement by allowing

other communications not necessarily related to

health.

Table 2. Tasks proposed to be supported by the ECA.

# Tas

T1 Follow up on medication schedules

T2 Keep track of the number of pills of each medication

in the home pharmacy

T3 Record and transcript the audio of the physician’s

appointment, identifying the key information to

personalize the other tasks

T4 Keep track of the amount of sodium ingeste

T5 Monitor blood pressure

T6 Propose and follow up on physical activities

T7 Suggest and help with elaborate diets

T8 Identify daily activities that may impact on treatment

T9 Inform about the disease

T10 Allow contact with the medical team, caregivers, and

family members

The two pharmacological tasks proposed for

medication monitoring are usual and can be

performed without the need for an ECA. However, an

empathic approach was indicated so that the patient

does not feel charged about medication schedules and

amounts in an untimely manner. Suggestions are

pointed out for the use of other subjects to make a rich

and meaningful dialogue. In the participants'

speeches, it is crucial that proactive interactions by

ECA are conducted with care to avoid instilling fear

in the dialogue, especially when such interactions are

considered untimely.

The analysis reveals that the urgency of various

tasks is not uniform across all patients. Each older

adult presents unique limitations, characteristics, and

needs as determined by their physician. These

individual factors significantly influence the

proposed execution of tasks T3 to T10. It is essential

to assess the patient's environment, along with their

medical prescriptions and guidelines, prior to

implementation. The primary rationale for

customizing tasks lies in the directives provided by

the physician, which are informed by the patient's

health history.

In the FG audio transcription stage, considerations

were made and identified. Three groups of

considerations were identified: Agent Characteristics,

Aspects related to the use of technology by the Older

Adults and Tasks to be supported by the ECA.

The characteristics of the ECA agent referenced

by the participants converged on a humanized avatar,

with physical characteristics reminiscent of someone

trustworthy and with some authority. The need for an

avatar adapted to the context of the Older Adults was

reiterated to the patient to be motivated.

The need for a physical avatar was identified as an

aspect that can negatively influence interaction. For

this feature to become an advantage, a connection

with someone from the patient's real context is

needed. Impersonation (sense of humor), voice,

accent and intonation were cited.

5 FINAL REMARKS

This study explored the potential of ECA to support

Older Adults in managing hypertension, a prevalent

health concern within aging populations. Through a

Focus Group with Health Care professionals and UX

Specialists, we identified ten key tasks that ECAs

could effectively address, ranging from medication

reminders and tracking to promoting healthy lifestyle

choices like diet and exercise.

The results highlight the ability of ECA to connect

technological progress with the unique requirements

of Older Adults, providing a tailored and user-

friendly method for managing chronic diseases. The

focus group participants highlighted the importance

of a humanized and trustworthy ECA avatar,

emphasizing the need to consider the agent's

characteristics and dialogue carefully to ensure user

acceptance and engagement.

Participants were chosen based on convenience

and the nature of their work. Since they are located in

different countries, we scheduled an online meeting

at a time that was convenient for everyone.

The research also highlights the broader

challenges of developing and deploying ECA for

Older Adults. Ethical considerations, particularly

regarding data privacy and security, emerged as

crucial concerns. Protecting sensitive health

information and ensuring transparency in data

collection practices are paramount for building trust

and fostering a positive user experience.

Furthermore, the study emphasized the

importance of tailoring ECA to individual needs and

IS4WB_SC 2025 - Special Session on Innovative Strategies to Enhance Older Adults’ Well-being and Social Connections

440

preferences, acknowledging the diverse range of

abilities and technological literacy among Older

Adults. A one-size-fits-all approach is unlikely to be

effective, and personalized interactions are essential

for maximizing engagement and promoting long-

term adherence to hypertension management plans.

Beyond the specific context of hypertension, this

work contributes to a broader understanding of how

ECA can be leveraged to support healthy aging. The

identified tasks and considerations are applicable to a

range of non-hedonic applications, including

companionship, cognitive stimulation, and access to

telehealth services.

By providing a user-friendly interface and

personalized support, ECAs can empower Older

Adults to maintain their independence, improve their

well-being, and actively participate in their own care.

Future research should focus on developing and

evaluating ECAs in real-world settings, exploring

their long-term impact on health outcomes and user

satisfaction.

In conclusion, this study provides valuable

insights into the essential tasks and key

considerations for developing effective ECAs for

Older Adults with hypertension. The ten tasks

identified by the focus group offer a practical

framework for designing and implementing ECAs

that can positively impact hypertension management

and overall well-being. By addressing ethical

concerns and prioritizing personalized interactions,

ECAs hold significant promise as a valuable tool for

supporting healthy aging and empowering Older

Adults to live more independent and fulfilling lives.

6 FUTURE WORK

The subsequent steps involve presenting the

identified tasks to the elderly population. This will be

conducted through motivational interviews with both

Brazilian and Portuguese seniors, as a functional

prototype is currently unavailable.

Additionally, the process includes the

development of a doctoral thesis focused on

Information and Communication on Digital

Platforms at the University of Porto and the

University of Aveiro.

REFERENCES

Bickmore, T. W., Trinh, H., Olafsson, S., O’Leary, T. K.,

Asadi, R., Rickles, N. M., & Cruz, R. (2018). Patient

and consumer safety risks when using conversational

assistants for medical information: An observational

study of siri, alexa, and google assistant. Journal of

Medical Internet Research, 20(9).

https://doi.org/10.2196/11510

Bousardt, H. M. (2022). Using Embodied Conversational

Agents to Combat Loneliness in Older Adults: A

Scoping Review [Master]. University of Twente.

Car, L. T., Dhinagaran, D. A., Kyaw, B. M., Kowatsch, T.,

Joty, S., Theng, Y. L., & Atun, R. (2020).

Conversational agents in health care: Scoping review

and conceptual analysis. In Journal of Medical Internet

Research (Vol. 22, Issue 8). JMIR Publications Inc.

https://doi.org/10.2196/17158

Cassell, J. (1998). Embodied Conversation: Integrating

Face and Gesture into Automatic Spoken Dialogue

Systems. ’.

Da Paixao Pinto, N., Dos Santos Franca, J. B., De Sa Sousa,

H. P., Vivacqua, A. S., & Garcia, A. C. B. (2021).

Conversational Agents for Elderly Interaction.

Proceedings of the 2021 IEEE 24th International

Conference on Computer Supported Cooperative Work

in Design, CSCWD 2021, 61–66.

https://doi.org/10.1109/CSCWD49262.2021.9437883

Ekman, P., & Friesen, W. V. (2019). Facial Action Coding

System. In PsycTESTS Dataset.

https://doi.org/10.1037/t27734-000

Esposito, A., Amorese, T., Cuciniello, M., Riviello, M. T.,

& Cordasco, G. (2021). How Human Likeness, Gender

and Ethnicity affect Elders’Acceptance of Assistive

Robots. Proceedings of the 2020 IEEE International

Conference on Human-Machine Systems, ICHMS 2020,

1–6.

https://doi.org/10.1109/ICHMS49158.2020.9209546

Hadian, M., Altuwaiyan, T., Liang, X., & Li, W. (2019).

Privacy-preserving voice-based search over mHealth

data. Smart Health, 12, 24–34.

https://doi.org/10.1016/j.smhl.2018.04.001

Knob, P., Dias, W. S., Kuniechick, N., Moraes, J., & Musse,

S. R. (2021). Arthur: a new ECA that uses Memory to

improve Communication. Proceedings - 2021 IEEE

15th International Conference on Semantic Computing,

ICSC 2021, 163–170.

https://doi.org/10.1109/ICSC50631.2021.00036

Kopp, S., & Hassan, T. (2022). The Fabric of Socially

Interactive Agents: Multimodal Interaction

Architectures. In The Handbook on Socially Interactive

Agents (pp. 77–112). ACM.

https://doi.org/10.1145/3563659.3563664

Kramer, L. L., Van Velsen, L., Clark, J. L., Mulder, B. C.,

& De Vet, E. (2022). Use and Effect of Embodied

Conversational Agents for Improving Eating Behavior

and Decreasing Loneliness Among Community-

Dwelling Older Adults: Randomized Controlled Trial.

JMIR Formative Research, 6(4).

https://doi.org/10.2196/33974

Mercado, J., Espinosa-Curiel, I. E., & Martínez-Miranda, J.

(2023). Embodied Conversational Agents Providing

Motivational Interviewing to Improve Health-Related

Behaviors: Scoping Review. In Journal of Medical

The Role of Embodied Conversational Agents in Supporting Older Adults with Hypertension: A Focus Group Study

441

Internet Research (Vol. 25, Issue 1). JMIR Publications

Inc. https://doi.org/10.2196/52097

Patrão Neves, M. do C. (2022). Criteria for Ethical

Allocation of Scarce Healthcare Resources: Rationing

vs. Rationalizing in the Treatment for the Elderly.

Philosophies, 7(6).

https://doi.org/10.3390/philosophies7060123

Teixeira, V. D., Pires, C. G., Pinto, F. M., Freitas, J., Dias,

M. S., & Rodrigues, E. M. (2012). Towards elderly

social integration using a multimodal human-computer

interface. Proceedings of the 2nd International Living

Usability Lab Workshop on AAL Latest Solutions,

Trends and Applications, AAL 2012, in Conjunction

with BIOSTEC 2012, 3–13.

https://doi.org/10.5220/0003852800030013

Telehealth Innovations in Remote Healthcare Services

Delivery - Global Telehealth 2020. (2021). In C. H. M.

E. L. van den B. C. G. Anthony J. Maeder (Ed.), Studies

in Health Technology and Informatics (Vol. 277).

Woo, J., Grimaldi, M., Pelachaud, C., & Achard, C. (2023).

IAVA: Interactive and Adaptive Virtual Agent.

Proceedings of the 23rd ACM International Conference

on Intelligent Virtual Agents, IVA 2023.

https://doi.org/10.1145/3570945.3607326

Woo, J., Pelachaud, C., & Achard, C. (2023). ASAP:

Endowing Adaptation Capability to Agent in Human-

Agent Interaction. Proceedings of the 28th

International Conference on Intelligent User

Interfaces, 464–475.

https://doi.org/10.1145/3581641.3584081

Woo, J., Shidara, K., Achard, C., Tanaka, H., Nakamura,

S., & Pelachaud, C. (2024). Adaptive virtual agent:

Design and evaluation for real-time human-agent

interaction. International Journal of Human Computer

Studies, 190.

https://doi.org/10.1016/j.ijhcs.2024.103321

Xie, C., Yang, P., & Yang, Y. (2018). Open Knowledge

Accessing Method in IoT-Based Hospital Information

System for Medical Record Enrichment. IEEE Access,

6, 15202–15211.

https://doi.org/10.1109/ACCESS.2018.2810837

Yaghoubzadeh, R., Kramer, M., Pitsch, K., & Kopp, S.

(2013). Virtual Agents as Daily Assistants for Elderly

or Cognitively Impaired People. In Lecture Notes in

Computer Science (including subseries Lecture Notes

in Artificial Intelligence and Lecture Notes in

Bioinformatics): Vol. 8108 LNAI (pp. 79–91).

https://doi.org/10.1007/978-3-642-40415-3_7

Yaghoubzadeh, R., Pitsch, K., & Kopp, S. (2015). Adaptive

Grounding and Dialogue Management for Autonomous

Conversational Assistants for Elderly Users (pp. 28–

38). https://doi.org/10.1007/978-3-319-21996-7_3

Yalcin, O. N. (2019). Evaluating Empathy in Artificial

Agents. 2019 8th International Conference on Affective

Computing and Intelligent Interaction, ACII 2019,

290–296. https://doi.org/10.1109/ACII.2019.8925498

Zagalo, N. (2020). Engagement Design (D. Tan & J.

Vanderdonckt, Eds.). Springer International

Publishing. https://doi.org/10.1007/978-3-030-37085-5

Zhang, J., & Umemuro, H. (2013). Older adults’ hedonic

and eudaimonic well-being in information and

communication technology activities.

Gerontechnology, 10(4), 231–243.

https://doi.org/10.4017/gt.2012.10.4.004.00

IS4WB_SC 2025 - Special Session on Innovative Strategies to Enhance Older Adults’ Well-being and Social Connections

442