ArticlePDF Available

Addressing Challenges in Promoting Healthy Lifestyles: The AI-Chatbot Approach

  • Università degli Studi di Trento, Fondazione Bruno Kessler

Abstract and Figures

Healthy lifestyles promotion is the main objective of primary care interventions, starting from the pediatric age, were overweight is nowadays exposing about one third of children to the risk of developing chronic diseases, such as diabetes. Recent years have seen a blast of mHealth apps for health promotion, targeting in particular nutrition and dietary behaviour change. However, reviews show difficulties in the adoption and effective usage of these applications in telemedicine and by the population in general, due to a lack of evidence-based content and strategies provided (e.g., by commercial apps) or lack of sufficient user engagement with the apps. Nutrition apps typically require self-reporting of food intake by the user which is often seen as a burden and a cause of abandonment of the app. However, current wave of research has taken up the challenge of promoting healthy lifestyles with advances in artificial intelligence (AI). This paper focus on AI chatbots as an innovative approach offering more simplicity and facilitating long-term adherence to health promotion interventions. Conversational assistants provide the advantage of being deployed in smartphones and laptops within a wide variety of applications. We will particularly focus on harnessing the power of intelligent chatbot systems to provide behaviour change interventions in telemedicine for healthy lifestyle promotion. We describe an application scenario for an AI-chatbot delivering support to nutrition education that could help to overcome current limitations of similar mHealth solutions provided for healthy lifestyles and contribute to more effective public health interventions in this application domain.
Content may be subject to copyright.
 
Healthy lifestyles promotion is the main objective of primary care
interventions, starting from the pediatric age, were overweight is
nowadays exposing about one third of children to the risk of
developing chronic diseases, such as diabetes. Recent years have
seen a blast of mHealth apps for health promotion, targeting in
particular nutrition and dietary behaviour change. However,
reviews show difficulties in the adoption and effective usage of
these applications in telemedicine and by the population in
general, due to a lack of evidence-based content and strategies
provided (e.g., by commercial apps) or lack of sufficient user
engagement with the apps. Nutrition apps typically require
self-reporting of food intake by the user which is often seen as a
burden and a cause of abandonment of the app. However, current
wave of research has taken up the challenge of promoting healthy
lifestyles with advances in artificial intelligence (AI). This paper
focus on AI chatbots as an innovative approach offering more
simplicity and facilitating long-term adherence to health
promotion interventions. Conversational assistants provide the
advantage of being deployed in smartphones and laptops within a
wide variety of applications. We will particularly focus on
harnessing the power of intelligent chatbot systems to provide
behaviour change interventions in telemedicine for healthy
lifestyle promotion. We describe an application scenario for an
AI-chatbot delivering support to nutrition education that could
help to overcome current limitations of similar mHealth solutions
provided for healthy lifestyles and contribute to more effective
public health interventions in this application domain.
Categories and Subject Descriptors
H.5.m. Information interfaces and presentation (e.g., HCI):
General Terms
Design, Human Factors.
Keywords: Behaviour change interventions, nutrition
education, chatbots, health promotion, sustainability.
Today there is an increasing awareness in public health that
preventing unhealthy lifestyles in the population, from the
pediatric to adult life, can bring enormous advantages in reducing
the occurrence of risk conditions, such as obesity that leads to
chronic diseases (e.g., diabetes) which represents a burden for
healthcare system worldwide [1]. Recent years have seen an
increasing offer of mHealth applications providing nutrition
education as well as promoting healthy diet and physical activity.
However, recent reviews show that most apps fail to sustain
long-term users adherence, often due to a lack of evidence-based
strategies and quality of the information provided [2-3-4].
Messaging platforms, such as Facebook messenger, WeChat, and
WhatsApp are quite popular and users tend to spend more time
chatting than using other applications [5]. With the introduction of
services, like Siri and Google Now or Google Allo people are
getting used to interacting with applications designed to support
smart coaching. Recently, Facebook launched their chatbot
platform to host businesses and provide services to customers
directly from their platform through a simple conversation. The
idea of “assistant-as-app” is to create user-virtual assistant
interaction with either voice or natural text messaging to
accomplish increasingly complex tasks with minimal effort.
Notwithstanding, the increasing number of chatbot solutions built
for health and lifestyle are still amateu r and need improvement
with respect to supporting complex behaviours, such as healthy
dietary adherence. To this extend, the existing solutions mainly
targeted food recommendation and physical activity tracking. The
majority of the approaches are focused on rule-based
architectures, where the chatbot intelligence is pre-defined and for
each user action, the bot triggers a response from a predefined list
of answers. Moreover, a few approaches have focused on
combining various profiling factors and techniques to achieve a
sustainable health plan. These factors are related to, for example,
user personality and providing a personalised plan that fits with
their preferences. In addition, sentiment analysis and user
emotional condition are important factors to consider to improve
the effectiveness and quality of user experience with the chatbot.
Integrating the chatbot design with appropriate behaviour change
strategies and techniques (BCTs) can help to effectively deploy
these solutions to support behavioural interventions for healthy
lifestyles and prevention by healthcare providers.
In this paper, we propose a chatbot system to promote healthy and
sustainable eating behaviour as a possible application scenario for
supporting primary care interventions to prevent weight gain in
the adult population. The bot acts as a bi-directional channel
leading to the adoption of more healthy habits regarding diet,
physical activity as well as food purchasing and preparation. The
presented system is unique in the way it collects data about user’s
relevant practices to provide personalised recommendation that
fits with their preferences and profiles. Moreover, since eating
behaviours and practices are strongly correlated with user’s
emotional and choice dynamics, the bot collects data to determine
user intent and perform sentiment analysis, to provide a more
tailored coaching towards a sustainable healthy diet. Finally, our
approach is meant to support behavioural interventions following
the ‘Efficiency Model of Support’ [6] where virtual and human
coaching are combined to facilitate users’ transition to healthier
lifestyles in the most efficient way.
Current research shows that most chatbot systems are used to
access information or communicate simple instructions to the
audience. A few studies have addressed the challenge of adapting
conversational interfaces to detect user sentiment and interpret
their intention. The existing studies on chatbot systems for diet
and lifestyle are either providing a way to track meals or provide
recommendation [7]. There is still room to improve these systems
with respect to bot effectiveness and user’s needs. In this work,
we propose a chatbot architecture that combines advanced feature
integration (e.g., machine learning models) to collect and
understand various features about the users. In this way, the bot
can be trained to understand the user state with respect to a given
plan and trigger the right action that can better fit with their
preferences. Moreover, the bot is able to trigger and notify a
healthcare expert (e.g., nutritionist) to facilitate intervention
whenever relevant. The healthcare expert can track user activities
through a web application.
The envisioned chatbot will combine multiple technical and
behavioural aspects to provide the necessary intelligence while
interacting with the user. These technologies include: the anatomy
and finite state machine for the bot design, the framework to
provide rule-based reasoning for the bot, the approach for
sentiment analysis, the behaviour change techniques to understand
user behaviour at a certain stage, and the intent detection
Chatbot System Anatomy and Finite State
Chatbot System Anatomy
Conversational agents can be of type voice-enabled or text-based,
and sometimes a combination of the two. Voice enabled can listen
to and respond in spoken language, whereas text-based can read
and respond to typed messages and requests. All conversational
agents rely on a set of core underlying technologies in order to
understand natural language input and human intend, and hence
engage in a human-like conversational. Figure 1 below shows the
key technology steps to be considered when developing
conversational agents.
Figure 1. Conversational Agents System Anatomy.
These above steps can vary depending on the application domain.
However, the order defined above better fits our chatbot scenario.
The Intent Detection categorizes the request into predefined
intents. The intent reflects what the user is trying to say or
achieve, and hence prescribes an action that defines the desired
outcome. The Role Detection assigns predefined categories to
entities of particular type. For example, for the request “I had a
burger for lunch”, the entity “burger” can be labeled food and the
entity “lunch” can be labeled meal time. The Entity Resolution
matches the identified entity with a real world object or concept.
For example, resolving an entity “beef burger” into a burger. The
Question Answering identifies the best answer for the request
based on knowledge base or acquired data. Finally, the Dialogue
Management tracks the context of the conversation and formulates
the appropriate response to the user.
Chatbot System Finite State Machine
The chatbot system finite state machine refers to the number of
states a user has to follow while interacting with the chatbot to
accomplish the action (Figure 2, Table 1). The system follows a
rule-based logic, however it relies on some APIs to add the
intelligence layer on top of the system.
Figure 2. The Finite State Automaton for Health & Lifestyle Chatbot
Where the conversation resides.
Where the bot provides necessary
information about the possible
The bot provides information to user based
on their request
The bot provides the activities for users
based on their health condition (e.g., user’s
daily diet or physical activity plan).
The bot provides a feedback to the user
about their performance.
Provides activities other than the original
When the user successfully fulfils the given
activity and reaches the goal.
The system will try to provide other
suggestions to the user, if one will work then
it will switch to Satisfying state and
External Actor
If changing suggestion won’t work, then
system will request information from the
External Actor state, which will involve a
human actor in the process.
Where the last goal-fulfilment task has been
When the bot closes the operation.
Table 1. The Description for Each Finite State.
Chatbot System Architecture
The chatbot follows a rule-based logic to handle various user
requests, and calls API services when providing the response to
such requests. We use Microsoft Bot Framework to provide a
scaffolding to host message-handling logic and plumbing to
integrate with various bot client endpoints. This framework
simplifies the task of setting up a server process which listens for
incoming text messages. Since rule-based frameworks provide no
AI capabilities to parse or classify incoming messages, we will
use API services, such as MonkeyLearn to perform more
advanced analysis (e.g., Sentiment Analysis and Intent Detection).
For example, if the user types “I feel happy with the current plan”,
the bot can extract the mood of the user with respect to the given
plan, in this case “Positive, with a certain probability”, for the
system architecture see Figure 3.
Figure 3. The Chatbot System High-Level Architecture.
Random Access Navigation (R.A.N.)
To provide as natural as possible interaction, the bot has to be
intelligent enough to detect how each user asks a question. To
achieve this, we followed the framework described by Shane Mac
[8] to provide random navigation for the bot to detect intentions
from user’s requests. This model outperforms fixed decision trees
and can complete more complex tasks while reduce friction [8].
The idea with this model is to detect all parameters required to
perform an intent with a context, allowing the user to change their
mind without going back, and it works seamlessly with web
views. In Figure 4 we provide an illustration of how an R.A.N
model behaves.
Figure 4. The Random Access Navigation Model.
An example of how such model works is provided below
regarding a user’s checking for grocery stores nearby:
I want to go grocery shopping this afternoon in Trento.
• What grocery stores are there in Trento on Friday?
• Show me grocery shops in Trento.
Within the eHealth and mHealth fields there is a subset of
technologies specifically designed to deliver behavior change
support to improve users’ health, and they are called behavioral
intervention technologies (BITs) [9]. Human support has been
integrated into BITs in different ways (text messaging, email,
phone calls provided by supporters with varying expertise
including therapists, nurses, trainees etc.). Integrating human
support, however, requires developed models for providing this
assistance. The Efficiency Model of Support [6] has been
proposed as a framework to guide the actions of supporters
delivering BITs by helping them to effectively manage the
interplay between information and intervention. Efficiency is
defined as the ratio of the outcome of an intervention relative to
the human resources required to deliver it, since each decision
corresponds to supporting that intervention (what, when, how
much, who provides it) represents a trade-off between devoting
additional resources and accruing additional benefits. According
to the model, decisions should be based in the consideration of
why people may fail to benefit from BITs and five categories of
possible failure points are proposed: usability, engagement, fit,
knowledge and implementation. These points should be taken into
consideration as important targets for support.
In the case of a telemedicine intervention for nutrition education
the deployment of a chatbot may lower usability barriers for users,
since conversational agents are considered among the most
intuitive to use kinds of BITs, requiring less learning effort by the
user. A chatbot would also effectively support engagement of the
user by prompting and providing healthy recommendations at the
time and place when food choices are made (e.g., before/during
daily meals, when cooking or shopping for food). By helping
users to unobtrusively keeping track of their food intake over the
week the chatbot can be an ideal solution to help them acquire an
accurate knowledge of healthy nutrition guidelines and to turn this
knowledge into healthy habits and practices of daily life. The
user-chatbot interaction may also provide an opportunity to easily
collect data on user difficulties by following the behavioural
intervention so as to inform healthcare provider (e.g., nutritionist,
pediatricians etc.) about these difficulties and focusing the
discussion on the critical points of the intervention for the patient,
thus better personalizing and improving the efficiency of the
support provided.
Chatbots can offer a lot of benefits in the telemedicine domain
both for healthcare providers and patients. For example, by using
chatbots patients’ data logging and assistance can be more
engaging [9]. Conversational agents are effective in making
healthcare topics easier to understand (e.g., physical activity
promotion, hospital discharge instruction, explanation of medical
documents, and family health history-tracking). Bots are effective
in providing nutrition education and primary care services can
harness the simplicity and age friendliness offered by bots in
various health interventions.
The kind of chatbot proposed for this application scenario can
achieve scalability and user engagement with simple message
exchange. We sum up the envisaged support provided to a user
Achieve healthy eating habits: By exchanging text
messages with the bot, user can receive immediate
actionable feedback to improve eating habits.
Be more active: The bot tracks user activities and helps
to acquire healthy habits.
Maintain healthy habits: By providing automation and
coach support, the chatbot can teach users how to stick
to their new habits over time.
Personalised support: The bot can check the user
adherence and encourage relevant activities that fit with
their personal plan.
Track user sentiment: The bot can collect emotional
data about users to understand their mental/mood
condition at a certain stage during their behavior change
Use Case Scenario
To better present the way a chatbot support may fit into a
telemedicine domain, we provide a use case scenario of the
user-chatbot interaction aimed at improving the user’s nutrition
and at implementing more sustainable food choices.
“Sara is 35 years old, she is living and working in Trento- Italy.
She is busy throughout the week days and has no idea where to
get veggies from the closest grocery shop. Moreover, she doesn’t
like to go to city often. Lately, she started having inactive life due
to her job condition and as a result she put on weight. Sometimes,
this condition and her work put her in a lot of stress which affects
her overall life and daily dietary habits.
Sara decides to change this, she consults with our chatbot and
starts using the chatbot application by simply searching the name
of the bot. Initially, the bot asks some questions about Sara’s
preferences and other diet and health related information. Later,
the bot starts to suggest support and recommendations, such as
KM0 grocery shops and optimal diet that goes well with her
condition and preferences. Later, the bot collects data about her
compliance with the suggestions and about her feeling. All these
data are communicated to the healthcare expert who can provide
alternative plans or intervene with Sara whenever needed. Sara is
better able to keep track of her meals and exercise by interacting
regularly with the bot.
The bot analyses her meals and activity, and then sets goal and
checks in periodically to keep her on track. Moreover, Sara
provides information to the bot about her breakfast,
and the bot
explains the healthiness of her meals; by recording information
about her activity, the bot is able to create an updated picture of
Sara’s condition and feed her with proper support.”
4. Design Challenges
There are still several challenges in applying the described chatbot
approach to support BITs in telemedicine. The conversations
supported between chatbot and user generally cannot be very
complex and they require increasing resources when expanding
the chatbot domain focus. This issue might be partially overcome
by future developments in the architectural design of
conversational agents. Some limitations in the interaction with the
bot are also due to the the singular and plural conversational
forms management. For example, if we want the bot to extract
entities, such as “pizza”, we need to provide the plural form as
well, “pizzas”. The synonyms, hypernyms and hyponyms which
are Natural Language Processing and ontology challenges are
among the complex limitations that most chatbots suffer today.
For example, if the user reports soda as a beverage, but the
chatbot only knows specific terms such as coca-cola or pepsi, that
are hyponyms of soda, then it can’t provide a sound reply to the
user request [11]. Other challenges are related to the privacy and
security of the data collected, since when relying on APIs to
perform sentiment analysis and tracking user habits, it’s important
to secure the user data collected.
Chatbots cannot replace humans [12 -13-14], but they can provide
an interesting channel to deliver behavioral interventions
combined to human support. The kinds of information collected
and provided by the chatbot can effectively complement the
support delivered by telemedicine services. This is likely to
increase the quality and efficiency of healthcare resources
deployed. Effective mechanisms enabling fallback in case the bot
is not able to properly manage a certain situation are to be
designed to facilitate human expert intervention. Future research
will need to tackle these challenges by means of empirical testing
in the field of these solutions.
The biggest advantages of chatbots include being able to reach a
broad audience on messenger apps, as well as the ability to
automate personalised messages. They can also improve
efficiency of resource allocation in the healthcare domain by
offloading clinicians and human operators from tasks that can be
Chatbots represent an innovative approach to address challenges
in the telecare and prevention domains. If well-designed and
implemented, chatbots can increase users’ engagement and
self-empowerment, by providing a better experience and save
costs for the healthcare system. However, designing an intelligent
chatbot that responds well to user demands is not trivial. The
chatbot ecosystem is moving very fast and new features are being
released every day by the numerous existing platforms. In this
paper we discussed an application scenario regarding a chatbot
system to promote healthy nutrition. The system adapts machine
learning techniques provided by API services to improve the bot
intelligence. We proposed a chatbot approach over mHealth
alternative solutions because it can fit well with usability
requirements that are essential for supporting interaction by
different target user groups. The bot provides a large set of
relevant functionalities ranging from food recommendation to
help shopping in a sustainable way. The fact that tech giants like
Google, Facebook, Microsoft, IBM and Amazon are giving
increasing attention to chatbot solutions is a strong indicator that
this technology will play a key role in the future. Currently, there
is an incredible amount of platforms and tools providing
different complexity levels, expressive powers and integration
capabilities. Which one to choose depends on the contextual
design and application scenario considered. Future work is still
needed to advance the Artificial Intelligence techniques required
to obtain intelligent chatbot solutions for the healthcare domain,
especially in terms of machine learning components, sentiment
analysis and intent detection.
[1] World Health Organization
[2] Breton ER, Fuemmeler BF, Abroms LC. 2011.Weight
loss-there is an app for that! But does it adhere to
evidence-informed practices? Transl Behav Med 2011
Dec;1(4):523-529 Doi= 10.1007/s13142-011-0076-5
[3] Cowan LT, Van Wagenen SA, Brown BA, Hedin RJ,
Seino-Stephan Y, Hall PC, et al. 2013.Apps of steel: are
exercise apps providing consumers with realistic
expectations?: a content analysis of exercise apps for
presence of behavior change theory. Health Educ Behav
2013 Apr;40(2):133-139. Doi= 10.1177/1090198112452126
[4] Jeon E., Park H., Min Y. H., & Kim H. Y. 2014. Analysis of
the information quality of Korean obesity-management
smartphone applications. Healthcare informatics research
(1), 23-29.
[5] Do, T. M. T., Blom, J., and Gatica-Perez, D. Smartphone
usage in the wild: a large-scale analysis of applications and
context. In Proceedings of the 13th international conference
on multimodal interfaces, ACM (2011), 353–360.
[6] Schueller, S.M., Tomasino, K.N. and Mohr, D.C., 2016.
Integrating human support into behavioral intervention
technologies: The efficiency model of support. Clinical
Psychology: Science and Practice
[7] Forman, G. 2003. An extensive empirical study of feature
selection metrics for text classification. J. Mach. Learn. Res.
3 (Mar. 2003), 1289-1305.
[8] Mac Follow, S. There are a dozen ways to order a coffee.
why do dumb bots only allow one?, 2017.
[9] Mohr, D. C., Burns, M. N., Schueller, S. M., Clarke, G., &
Klinkman, M. 2013. Behavioral intervention technologies:
Evidence review and recommendations for future research in
mental health. General Hospital Psychiatry, 35, 332–338.
[10] Consolvo, S., Klasnja, P., McDonald, D. W., Landay, J. A.,
et al. Designing for healthy lifestyles: Design considerations
for mobile technologies to encourage consumer health and
wellness. Foundations and Trends in Human–Computer
Interaction 6, 3–4 (2014), 167–315.
[11] Brinkman, W.-P. Virtual health agents for behavior change:
Research perspectives and directions.
[12] Fadhil, A., Matteotti, C., Armellin, G., Villafiorita, A., and
Betti, D. Coach me: a platform for promoting healthy
lifestyle. In Proceedings of the 18th International Conference
on Human-Computer Interaction with Mobile Devices and
Services Adjunct, ACM (2016), 1077–1080.
[13] Brown, L. D., Hua, H., and Gao, C. 2003. A widget
framework for augmented interaction in SCAPE. In
Proceedings of the 16th Annual ACM Symposium on User
Interface Software and Technology (Vancouver, Canada,
November 02 - 05, 2003). UIST '03. ACM, New York, NY,
1-10. DOI=
[14] Alvarez-Godinez, Eduardo, and Sumithra Bhakthavatsalam.
"Anna: A Nutrition-Facts Dialogue System."
... The conversational function, aka Chatbots, is crucial to check, obtain feedback and make recommendations for improvements. A conversational function represents an innovative approach to address challenges in telecare and prevention domains in healthcare [5]. ...
... It is also a technology to prevent diseases through the analysis of the lifestyle of modern people. This technology is a platform technology that can be given market value [5]. ...
... Further developed chatbot conversation flows, and its better interpretation of users' input might increase MYA's usability [15], which in turn could improve its acceptability and the users' confidence in its proper function. Fadhil et al. [18] associate a better chatbot experience that promotes user engagement with a well-designed and implemented chatbot. Since MYA was a prototype with a simulated activity tracker, the identification and reporting of issues by the participants was expected. ...
Conference Paper
Full-text available
Fully automated self-help interventions integrated with social media chatbots could serve as highly cost-effective physical activity promotion tools for a large population. We have developed MYA, a Telegram-based chatbot for increasing physical activity. The objective of this study was to assess the usability of MYA. To identify usability issues, we recruited volunteers and asked them to interact with MYA and to answer the Chatbot Usability Questionnaire. Thirty volunteers participated in the study, 83.3% agreed MYA was welcoming during initial setup and 63.3% agreed MYA was very easy to use. MYA was perceived as realistic and engaging, easy to navigate, and its responses were useful, appropriate, and informative (all 53.3%). However, 63.3% of respondents agreed MYA failed to recognize most of their inputs, and 43.3% claimed it would be easy to get confused when using MYA. Although the results are encouraging, it remains unclear if a social media chatbot can motivate people to increase their physical activity. MYA has the potential to do that, with improvements in functionalities like challenge personalization. The efficacy of these approaches should be studied in a clinical trial.
... Messaging platforms instead (e.g. Facebook messenger, WhatsApp) are quite popular and users tend to spend more time chatting than using other applications [15]. ...
Conference Paper
Artificial intelligence is transforming healthcare with a profound paradigm shift impacting diagnostic techniques, drug discovery, health analytics, interventions and much more. In this paper we focus on exploiting AI-based chatbot systems, mainly based on machine learning algorithms and Natural Language Processing, to understand and respond to needs of patients and their families. In particular, we describe an application scenario for an AI-chatbot delivering support to pregnant women, mothers, and families with young children, by giving them help and instructions in relevant situations.
... Barak et al. (2009) reported that the ability of chatbots to create empathy and react to emotions resulted in higher compliance with therapeutic treatments [3]. Healthcare focused chatbots facilitate increased user engagement and increased usability [10] and may also solve issues with text message-based systems, such as 24-hour availability and automated messages sounding less human [9]. In response to growing waiting lists and difficulties in accessing mental health services, developed iHelpr, a self-help mental health chatbot. ...
Conference Paper
Full-text available
Chatbots are becoming increasingly popular as a human-computer interface. The traditional best practices normally applied to User Experience (UX) design cannot easily be applied to chatbots, nor can conventional usability testing techniques guarantee accuracy. WeightMentor is a bespoke self-help motivational tool for weight loss maintenance. This study addresses the following four research questions: How usable is the WeightMentor chatbot, according to conventional usability methods?; To what extend will different conventional usability questionnaires correlate when evaluating chatbot usability?; And how do they correlate to a tailored chatbot usability survey score?; What is the optimum number of users required to identify chatbot usability issues?; How many task repetitions are required for a first-time chatbot users to reach optimum task performance (i.e. efficiency based on task completion times)? This paper describes the procedure for testing the WeightMentor chatbot, assesses correlation between typical usability testing metrics, and suggests that conventional wisdom on participant numbers for identifying usability issues may not apply to chatbots. The study design was a usability study. WeightMentor was tested using a pre-determined usability testing protocol, evaluating ease of task completion, unique usability errors and participant opinions on the chatbot (collected using usability questionnaires). WeightMentor usability scores were generally high, and correlation between questionnaires was strong. The optimum number of users for identifying chatbot usability errors was 26, which challenges previous research. Chatbot users reached optimum proficiency in tasks after just one repetition. Usability test outcomes confirm what is already known about chatbots - that they are highly usable (due to their simple interface and conversation-driven functionality) but conventional methods for assessing usability and user experience may not be as accurate when applied to chatbots.
... A text-based analysis of the intent to churn is even more relevant today in the context of the chatbot explosion (Hill et al., 2015;Fadhil and Gabrielli, 2017;Argal et al., 2018). Chatbots are becoming one of the main means of textual communication with the evolution of automation processes. ...
Conference Paper
Very recently, ESEAP mutual authentication protocol was designed to avoid the drawbacks of Wang et al. protocol and highlights that the protocol is protecting all kind of security threats using informal analysis. This work investigates the ESEAP protocol in security point of view and notices that the scheme is not fully protected against stolen verifier attack and does not provide user anonymity. Furthermore, the same protocol has user identity issues, i.e., the server cannot figure out the user identity during the authentication phase. Later we discuss the inconsistencies in the security analysis of ESEAP presented by RESEAP.
The Handbook of Research on Lifestyle Sustainability and Management Solutions Using AI, Big Data Analytics, and Visualization offers extensive research on lifestyle management and services that contribute towards indication, detection, conduction, protection, and technological enhancement including machine learning, deep learning, artificial intelligence, big data analytics, and visualization. It also provides mechanisms that can improve lifestyle monitoring and help in increasing the immunity of the human body. Covering topics such as big data, robot therapy, and wearable technology, it is ideal for students, researchers, technologists, IT specialists, computer engineers, systems engineers, data scientists, doctors, hospital administrators, engineers, academicians, and technology providers.
Full-text available
Artificial Intelligence (AI) technology is transforming the healthcare sector. However, despite this, the associated ethical implications remain open to debate. This research investigates how signals of AI responsibility impact healthcare practitioners’ attitudes toward AI, satisfaction with AI, AI usage intentions, including the underlying mechanisms. Our research outlines autonomy, beneficence, explainability, justice, and non-maleficence as the five key signals of AI responsibility for healthcare practitioners. The findings reveal that these five signals significantly increase healthcare practitioners’ engagement, which subsequently leads to more favourable attitudes, greater satisfaction, and higher usage intentions with AI technology. Moreover, ‘techno-overload’ as a primary ‘techno-stressor’ moderates the mediating effect of engagement on the relationship between AI justice and behavioural and attitudinal outcomes. When healthcare practitioners perceive AI technology as adding extra workload, such techno-overload will undermine the importance of the justice signal and subsequently affect their attitudes, satisfaction, and usage intentions with AI technology.
Full-text available
Context. Asynchronous messaging is increasingly used to support human–machine interactions, generally implemented through chatbots. Such virtual entities assist the users in activities of different kinds (e.g., work, leisure, and health-related) and are becoming ingrained into humans’ habits due to factors including (i) the availability of mobile devices such as smartphones and tablets, (ii) the increasingly engaging nature of chatbot interactions, (iii) the release of dedicated APIs from messaging platforms, and (iv) increasingly complex AI-based mechanisms to power the bots’ behaviors. Nevertheless, most of the modern chatbots rely on state machines (implementing conversational rules) and one-fits-all approaches, neglecting personalization, data-stream privacy management, multi-topic management/interconnection, and multimodal interactions. Objective. This work addresses the challenges above through an agent-based framework for chatbot development named EREBOTS. Methods. The foundations of the framework are based on the implementation of (i) multi-front-end connectors and interfaces (i.e., Telegram, dedicated App, and web interface), (ii) enabling the configuration of multi-scenario behaviors (i.e., preventive physical conditioning, smoking cessation, and support for breast-cancer survivors), (iii) online learning, (iv) personalized conversations and recommendations (i.e., mood boost, anti-craving persuasion, and balance-preserving physical exercises), and (v) responsive multi-device monitoring interface (i.e., doctor and admin). Results. EREBOTS has been tested in the context of physical balance preservation in social confinement times (due to the ongoing pandemic). Thirteen individuals characterized by diverse age, gender, and country distribution have actively participated in the experimentation, reporting advancements in the physical balance and overall satisfaction of the interaction and exercises’ variety they have been proposed.
Full-text available
Objective: The objective of this study was to assess whether a version of the Smoke Free app with a supportive chatbot powered by artificial intelligence (versus a version without the chatbot) led to increased engagement and short-term quit success. Methods: Daily or non-daily smokers aged ≥18 years who purchased the 'pro' version of the app and set a quit date were randomly assigned (unequal allocation) to receive the app with or without the chatbot. The outcomes were engagement (i.e. total number of logins over the study period) and self-reported abstinence at a one-month follow-up. Unadjusted and adjusted negative binomial and logistic regression models were fitted to estimate incidence rate ratios (IRRs) and odds ratios (ORs) for the associations of interest. Results: A total of 57,214 smokers were included (intervention: 9.3% (5339); control: 90.7% (51,875). The app with the chatbot compared with the standard version led to a 101% increase in engagement (IRRadj = 2.01, 95% confidence interval (CI) = 1.92-2.11, p < .001). The one-month follow-up rate was 10.6% (intervention: 19.9% (1,061/5,339); control: 9.7% (5,050/51,875). Smokers allocated to the intervention had greater odds of quit success (missing equals smoking: 844/5,339 vs. 3,704/51,875, ORadj = 2.38, 95% CI = 2.19-2.58, p < .001; follow-up only: 844/1,061 vs. 3,704/5,050, ORadj = 1.36, 95% CI = 1.16-1.61, p < .001). Conclusion: The addition of a supportive chatbot to a popular smoking cessation app more than doubled user engagement. In view of very low follow-up rates, there is low quality evidence that the addition also increased self-reported smoking cessation.
Full-text available
A growing number of interventions use websites, mobile applications, and wearable devices to deliver and enhance mental health treatments. These technologies are used more often and are more effective when provided along with human support. Integrating human support, however, requires developed models for providing this support. This article presents the Efficiency Model of Support, a new model for understanding the provision of human support in the context of behavioral intervention technologies. The Efficiency Model of Support defines the ratio of benefit accrued from an intervention to resources devoted to it as a critical consideration in support provision. The Efficiency Model of Support serves to consolidate the current findings and guide future research and practice with regard to human support and technology.
Conference Paper
Full-text available
Promoting health and wellness reflects a holistic approach to maintain the overall wellbeing of the nation. This paper discusses challenges in dietary adherence and reviews approaches in promoting healthy diet, physical activity, and healthy lifestyle presented in the literature. After discussing persisting challenges, we propose our future approach and contribution to overcome these challenges.
Full-text available
This study analyzed smartphone obesity-management applications developed in Korea and the quality of the information that they provide. Obesity-management smartphone applications were searched using the keywords 'obesity + management,' 'weight + management,' 'weight + loss,' 'weight + exercise,' 'weight + diet,' 'weight + calories,' and 'diet,' with a search application programming interface (provided by Apple) between September 23 and September 27, 2013. These applications were then classified according to their main purpose, type of interventions used, price, type of developer, and user ratings. The information quality of the applications was analyzed using the Silberg scale. In total, 148 smartphone applications for obesity management were found. The main purpose of most of these applications (70.95%) was to provide information regarding weight control. The most frequently used intervention (34.62%) was to provide information on exercise management. More than half of the applications (58.78%) were free of charge. The mean of users' rating of these applications was 3.68 out of 5. The quality of information provided by these applications was evaluated as 4.55 out of 9: specifically, 1.79 out of 3 for authorship, 0.22 out of 2 for attribution, 1.29 out of 2 for disclosure, and 1.25 out of 2 for currency. Only three of the applications (2.88%) had a score on the Silberg scale greater than or equal to 7 points. The findings of this study suggest that the quality of information provided by smartphone applications in the healthcare domain urgently need to be evaluated to prevent users being misinformed by these applications.
Full-text available
Little is known about how much smartphone apps for weight control adhere to evidence-informed practices. The aim of this study was to review and summarize the content of available weight control apps. Information on content, user rating, and price was extracted from iTunes on September 25, 2009. Apps (n = 204) were coded for adherence to 13 evidence-informed practices for weight control. Latent class analysis was used to identify subgroups of apps based on endorsement practices. Only a small percentage of apps had five or more of the 13 practices (15%). Latent class analysis revealed three main types of apps: diet, physical activity, and weight journals (19%); dietary advice and journals (34%); and weight trackers (46%). User ratings were not associated with apps from these three classes. Many apps have insufficient evidence-informed content. Research is needed that seeks to develop, improve, and evaluate these apps.
Full-text available
Machine learning for text classification is the cornerstone of document categorization, news filtering, document routing, and personalization. In text domains, effective feature selection is essential to make the learning task efficient and more accurate. This paper presents an empirical comparison of twelve feature selection methods (e.g. Information Gain) evaluated on a benchmark of 229 text classification problem instances that were gathered from Reuters, TREC, OHSUMED, etc. The results are analyzed from multiple goal perspectives-accuracy, F-measure, precision, and recall-since each is appropriate in different situations. The results reveal that a new feature selection metric we call 'Bi-Normal Separation' (BNS), outperformed the others by a substantial margin in most situations. This margin widened in tasks with high class skew, which is rampant in text classification problems and is particularly challenging for induction algorithms. A new evaluation methodology is offered that focuses on the needs of the data mining practitioner faced with a single dataset who seeks to choose one (or a pair of) metrics that are most likely to yield the best performance. From this perspective, BNS was the top single choice for all goals except precision, for which Information Gain yielded the best result most often. This analysis also revealed, for example, that Information Gain and Chi-Squared have correlated failures, and so they work poorly together. When choosing optimal pairs of metrics for each of the four performance goals, BNS is consistently a member of the pair-e.g., for greatest recall, the pair BNS + F1-measure yielded the best performance on the greatest number of tasks by a considerable margin.
Full-text available
Objective: A technical expert panel convened by the Agency for Healthcare Research and Quality and the National Institute of Mental Health was charged with reviewing the state of research on behavioral intervention technologies (BITs) in mental health and identifying the top research priorities. BITs refers to behavioral and psychological interventions that use information and communication technology features to address behavioral and mental health outcomes. Method: This study on the findings of the technical expert panel. Results: Videoconferencing and standard telephone technologies to deliver psychotherapy have been well validated. Web-based interventions have shown efficacy across a broad range of mental health outcomes. Social media such as online support groups have produced disappointing outcomes when used alone. Mobile technologies have received limited attention for mental health outcomes. Virtual reality has shown good efficacy for anxiety and pediatric disorders. Serious gaming has received little work in mental health. Conclusion: Research focused on understanding reach, adherence, barriers and cost is recommended. Improvements in the collection, storage, analysis and visualization of big data will be required. New theoretical models and evaluation strategies will be required. Finally, for BITs to have a public health impact, research on implementation and application to prevention is required.
Full-text available
Objective: To quantify the presence of health behavior theory constructs in iPhone apps targeting physical activity. Methods: This study used a content analysis of 127 apps from Apple's (App Store) Health & Fitness category. Coders downloaded the apps and then used an established theory-based instrument to rate each app's inclusion of theoretical constructs from prominent behavior change theories. Five common items were used to measure 20 theoretical constructs, for a total of 100 items. A theory score was calculated for each app. Multiple regression analysis was used to identify factors associated with higher theory scores. Results: Apps were generally observed to be lacking in theoretical content. Theory scores ranged from 1 to 28 on a 100-point scale. The health belief model was the most prevalent theory, accounting for 32% of all constructs. Regression analyses indicated that higher priced apps and apps that addressed a broader activity spectrum were associated with higher total theory scores. Conclusion: It is not unexpected that apps contained only minimal theoretical content, given that app developers come from a variety of backgrounds and many are not trained in the application of health behavior theory. The relationship between price and theory score corroborates research indicating that higher quality apps are more expensive. There is an opportunity for health and behavior change experts to partner with app developers to incorporate behavior change theories into the development of apps. These future collaborations between health behavior change experts and app developers could foster apps superior in both theory and programming possibly resulting in better health outcomes.
As the rates of lifestyle diseases such as obesity, diabetes, and heart disease continue to rise, the development of effective tools that can help people adopt and sustain healthier habits is becoming ever more important. Mobile computing holds great promise for providing effective support for helping people manage their health in everyday life. Yet, for this promise to be realized, mobile wellness systems need to be well designed, not only in terms of how they implement specific behavior-change techniques but also, among other factors, in terms of how much burden they put on the user, how well they integrate into the user's daily life, and how they address the user's privacy concerns. Designing for all of these constraints is difficult, and it is often not clear what tradeoffs particular design decisions have on how a wellness application is experienced and used. In this monograph, we provide an account of different design approaches to common features of mobile wellness applications and we discuss the tradeoffs inherent in those approaches. We also outline the key challenges that HCI researchers and designers will need to address to move the state of the art for mobile wellness technologies forward.
We introduce Anna, a nutrition-facts dialogue system built with a combination of freely available components and some ad hoc code. In this paper, we describe its domain, its over-all architecture, and its major components.
Conference Paper
This paper presents a large-scale analysis of contextualized smartphone usage in real life. We introduce two contextual variables that condition the use of smartphone applications, namely places and social context. Our study shows strong dependencies between phone usage and the two contextual cues, which are automatically extracted based on multiple built-in sensors available on the phone. By analyzing continuous data collected on a set of 77 participants from a European country over 9 months of actual usage, our framework automatically reveals key patterns of phone application usage that would traditionally be obtained through manual logging or questionnaire. Our findings contribute to the large-scale understanding of applications and context, bringing out design implications for interfaces on smartphones.