Designing a Digital Personal Coach to Promote a Healthy Diet and

Physical Activity Among Patients After Cardiothoracic Surgery

Ana Martins

1,3 a

, Isabel Nunes

3 b

, Lu

ıs Velez Lap

2,3 c

and Ana Londral

1,2 d

Value for Health CoLAB, Lisboa, Portugal

Comprehensive Health Research Center, Nova Medical School, Nova University of Lisbon, Lisboa, Portugal

Department of Mechanical and Industrial Engineering, NOVA School of Science and Technology, Nova University of

Lisbon, Lisboa, Portugal

Keywords:

Digital Coach, Text-Message Intervention, Conversational Agents, Behavior Change Theory, Cardiothoracic

Surgery.

Abstract:

This position paper describes the design of a text-message intervention that uses behavior change theory to

help design the intervention to improve patients’ diet and physical activity. The text-message intervention aims

to guide patients to improve their self-efﬁcacy in managing their disease to reduce complications and hospital

readmissions and improve health-related outcomes, well-being, and quality of life. Design Science Research

methodology is used to support problem deﬁnition, design, development, and evaluation of a sustainable,

useful, and valuable artifact.

1 INTRODUCTION

A healthy lifestyle can reduce the burden associated

with chronic diseases by preventing their appearance,

and promoting health maintenance. Protective behav-

iors should be taught to chronic patients to improve

their quality of life and long-term well-being. This

can reduce the workload of healthcare professionals

by making patients active agents in their health (Bauer

et al., 2014; Kontis et al., 2014). Behavior change in-

terventions can guide the process of change and help

maintain a healthy behavior (Grimmett et al., 2019).

The use of conversational agents can be an afford-

able and efﬁcient way to personalize care by deliv-

ering personalized content based on what is reported

and needed by the patient (Chaix et al., 2019; Fitz-

patrick et al., 2017; Stephens et al., 2019; Ghandehar-

ioun et al., 2019).

Cardiovascular disease (CVD) is the biggest cause

of mortality and morbidity worldwide (WHO, 2021).

A prospective cohort study found that cardiovascular

disease events are caused by modiﬁable risk factors.

The predominant factors were hypertension, high

https://orcid.org/0000-0002-4890-3941

https://orcid.org/0000-0002-0428-0930

https://orcid.org/0000-0003-0506-1294

https://orcid.org/0000-0002-8002-6790

low-density lipoprotein (LDL) cholesterol, household

pollution, and tobacco use (Yusuf et al., 2020).

Cardiothoracic surgery has decreased mortality

and morbidity of cardiovascular patients, allowing pa-

tient’s to live longer. However, because surgery is

a stressful event (Gardner et al., 2005; Pinto et al.,

2016; Tadas et al., 2021), it may result in a failure

to fully understand the medical information provided

and compromise health outcomes. These facts have

led researchers to focus on ways to increase patients’

awareness of their disease to improve their well-being

and quality of life and prevent further events from oc-

curring (Tully, 2013). Pervasive technology might be

used to improve modifying risk factors and quality of

life.

Behavior change interventions (BCIs) systemati-

cally organize actions that target speciﬁc behaviors

with the intent to improve a target behavior in a tar-

get population (Michie et al., 2011). Interventions

to change behavior are complex due to the interact-

ing components they are composed of. In the design

phase, speciﬁcations must be understandable to en-

sure proper evaluation of effectiveness.

The Behavior Change Wheel (BCW) summarizes

19 behavior change frameworks and was developed

using expert consensus and validation (Michie et al.,

2011) to guide researchers in the design and evalua-

tion phases. The framework divides intervention de-

Martins, A., Nunes, I., Lapão, L. and Londral, A.

Designing a Digital Personal Coach to Promote a Healthy Diet and Physical Activity Among Patients After Cardiothoracic Surgery.

DOI: 10.5220/0011776800003414

In Proceedings of the 16th International Joint Conference on Biomedical Engineering Systems and Technologies (BIOSTEC 2023) - Volume 5: HEALTHINF, pages 595-602

ISBN: 978-989-758-631-6; ISSN: 2184-4305

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

595

sign into three phases: (1) understanding the behav-

ior, (2) identifying intervention options, and (3) iden-

tifying content and implementation options. In the

ﬁrst phase, the deﬁnition of the problem in behav-

ioral terms is speciﬁed, the target behavior is selected

and speciﬁed, and a needs assessment is performed.

The behavioral analysis is performed using the model

of capacity, opportunity, and motivation (COM-B).

Capability stands for the psychological and physical

ability to perform a certain activity. Opportunity is

characterized as the external factors that could trig-

ger/enable a particular behavior, and motivation is the

conscious and unconscious cognitive processes that

lead the agent to perform a particular behavior in-

tuitively or rationally. Phase 2 establishes the in-

tervention functions and policy categories are iden-

tiﬁed. Education, persuasion, incentivization, coer-

cion, training, restriction, environmental structuring,

modeling, and enablement are intervention functions

that can be used to achieve a speciﬁc target behavior

(Michie et al., 2011). In the last phase, the content

and implementation options are clariﬁed. Behavior

change techniques (BCTs) should be deﬁned in this

phase.

BCTs are observable, replicable, and irreducible

components that aim to redirect causal processes that

regulate target behavior. They can give us cues about

how the content delivered by the conversational agent

inﬂuences the receiver (Michie et al., 2013). Mecha-

nisms of action (MoAs) give us a better understand-

ing of how BCTs act on individuals to produce the

expected effects. Based on the studies published to

date, the Human Behavior Change Project has devel-

oped a tool that maps BCTs to MoAs. This allows the

MoA underlying the change to be identiﬁed (Carey

et al., 2018; Johnston et al., 2021).

In this position paper, we describe the methodol-

ogy we used to develop a pilot study aimed at inte-

grating behavior change theory into a digital solution

to promote healthy eating and regular physical activ-

ity. Future work is also explained.

2 METHODS

Design Science Research Methodology (DSRM) is

used to ensure that a successful artifact created to ad-

dress a problem that meets the interests of all stake-

holders is produced (Hevner et al., 2004). A suc-

cessful DSRM cycle draws on previous literature and

lends itself as a solution to a speciﬁc problem whose

utility, quality, and effectiveness will be rigorously

evaluated (Hevner et al., 2004). The development

of the artifact should uphold on existing theories and

knowledge to a speciﬁc problem (Peffers et al., 2007).

We followed a nominal process based on previous re-

search involving 6 steps (Peffers et al., 2007) as de-

picted in Figure 1.

2.1 Problem Identiﬁcation, Context

Characterization, and Motivation

Consistent with the DSRM, problem identiﬁcation

and motivation deﬁnition are critical to the sustain-

ability of the artifact. This step was divided into 4

activities, namely context characterization, data anal-

ysis, literature review, and semi-structured interview

with clinical team.

We obtained an initial description of the context

in which the artifact would be used during a meet-

ing with the researchers involved in an earlier research

project to develop a telemonitoring service to monitor

patients undergoing cardiothoracic surgery (Londral

et al., 2022). An additional meeting with the clini-

cal team helped understanding how the follow-up af-

ter the telemonitorization period was done and what

were the major concerns of the clinical team after the

telemonitorization period.

Following the context characterization, data anal-

ysis of electronic health records (EHR) was con-

ducted to better understand the demographic charac-

teristics of the study population, such as age, gen-

der, income, education, occupation, and additional

medical information on pre-existing conditions to

surgery and type of procedure. In addition, patient-

reported outcomes (PROMs) collected by nurses dur-

ing follow-up interviews were also analyzed.

At this point qualitative studies regarding experi-

ences after surgery (Gardner et al., 2005; Pinto et al.,

2016; Tadas et al., 2021) where studied to understand

what are patient’s experiences after surgery. Addi-

tionally, risk modifying factors of cardiovascular dis-

ease were assessed (Yusuf et al., 2020).

After data analysis and discussion of ﬁndings with

the research team, semi-structured interviews were

conducted with 7 healthcare providers to understand

(1) what are risk modifying factors for the population

considered important to the clinical team, (2) what is

important for patients to have a better quality of life,

(3) what are the barriers and facilitators to activities

relevant to increasing patient well-being, (4) whether

there are personality traits in this population that with

better health outcomes, (5) how the follow-up service

works, (6) how information is communicated to pa-

tients (literacy strategies used), and (7) how this ser-

vice can be improved. Interviews were performed us-

ing LimeSurvey (Schmitz, 2012).

Scale-IT-up 2023 - Workshop on Best Practices for Scaling-Up Digital Innovations in Healthcare

596

Figure 1: Methodology for developing a valuable artifact for follow-up in the cardiothoracic surgery service.

2.2 Deﬁne the Objectives for a Solution

The speciﬁcation of the target behaviors for the in-

tervention focused on understanding which behaviors

can have a broader impact in the modifying risk fac-

tors for cardiovascular disease.

Ten patients who had participated in the telemon-

itoring pilot project mentioned earlier (Londral et al.,

2022) were interviewed a year and a half after surgery

to collect their experiences during the recovery pe-

riod. The semi-structured interview consisted of 8

questions, according to the recommendations of the

clinical team, who advised us to ask only a few ques-

tions, otherwise, the patient would lose interest. In

this interview, the COM -B model is used to identify

barriers and facilitators to healthy eating and physi-

cal activity. So the questions focused on ﬁnding out

how the surgery changed their lives, what their quality

of life is like today, what motivates them to maintain

healthy habits and follow the clinical team’s recom-

mendations, whether they adhere to diet and physical

activity recommendations. We also asked patients if

they could recommend changes to the service. In this

way, we were able to conduct a behavioral analysis

of the patients who participated in the telemonitoring

project in accordance with the phase 1 of the BCW

framework. The main barriers and motivators for ini-

tiating and maintaining a particular behavior were as-

signed to the intervention function layer of the BCW.

The objectives were set after 2 meetings with the

researchers involved in the study (engineers, medical

doctors, and nurses) where all the information gath-

ered and a possible objective for a solution were pre-

sented. We foresee the need to redeﬁne the objective

after each iteration of the study.

A systematic literature review was conducted to

determine how to develop conversational agents that

deliver personalized and automated content to pa-

tients with chronic diseases.

2.3 Design and Development

In designing and developing the solution, we focused

on integrating the results of the previously deﬁned

target behavior with the intervention functions previ-

ously identiﬁed by the clinical team and the results of

the systematic literature review.

2.4 Demonstration

The ﬁrst iteration of the DSRM will be completed

with a pilot study of 5 patients who agree to partic-

ipate.

2.5 Evaluation

The evaluation process was deﬁned based on the tar-

get behavior previously established in the goal deﬁni-

tion phase. Our goal was to deﬁne methods currently

used in the service to assess patient recovery in or-

der to increase the feasibility of the artifact. We also

wanted to measure engagement, acceptance, and sat-

isfaction.

3 RESULTS

3.1 Problem Identiﬁcation and

Motivation

The cardiothoracic surgery service follows patients

for one year after surgery. During this time, nurses

conduct follow-up visits to assess patients’ recovery

based on physical activity, pain, diet, body mass in-

dex, medications, health monitoring, social support,

tobacco use, and quality of life. Regular appointments

are also scheduled with the patient’s assigned medical

doctor. In addition, patients who are willing to partic-

ipate in the telemonitoring study, if selected for the

intervention group, can beneﬁt from the telemonitor-

ing service three months after surgery, as shown in

Figure 2.

EHRs provided by the hospital showed that only

1% of patients had no preexisting condition before

surgery. Pre-existing conditions or risk factors were

arterial hypertension (17.7%), hypercholesterolemia

(14.6%), diabetes (6.9%), persistent atrial ﬁbrillation

(3.8%), previous myocardial infarction (3.2%), smok-

ing (3.1%), and more. According to the clinical team,

Designing a Digital Personal Coach to Promote a Healthy Diet and Physical Activity Among Patients After Cardiothoracic Surgery

597

most patients are treated as having a chronic dis-

ease after surgery and there is a high risk of read-

mission 1 year after surgery. Therefore, the service

provides specialist appointments, uses various tech-

niques to improve patients’ literacy, conducts tele-

monitoring, does regular follow-up interviews, and

distributes brochures on nutrition and physical activ-

ity to help patients properly manage their condition.

The semi-structured interviews with the clinical

team revealed that risk factors are important to im-

prove health-related outcomes. Managing chronic

conditions such as diabetes, hypertension, LDL, a

healthy diet, regular physical activity, good mental

health, and a social role are factors that positively

inﬂuence recovery. Barriers to implementing protec-

tive behaviors include poor health literacy skills, lack

of motivation, lack of support and specialized help,

employment, anxiety, fear, pain, and physical capa-

bility. The clinical team was unable to identify any

unique personality traits in this population. The clin-

ical team suggested developing an automated system

that would not increase the workload of the clinical

team, predict what the patient’s needs would be at

home after the telemonitoring period, and send per-

sonalized messages to encourage healthy behaviors.

Doctors also say that it is very common for pa-

tients to come back to the hospital after 1 year of

surgery due to complications. This might be cause

by lack of self-management skills and distress at that

moment.

For what was said, we hypothesized that a prob-

lem could be associated with the lack self-efﬁcacy in

self-managing the disease. Thus, the use of pervasive

technology could be important to provide support,

and valuable information for managing their disease

(Halldorsdottir et al., 2020; Akinosun et al., 2021).

3.2 Deﬁne the Objectives for a Solution

The target behaviors for the intervention were identi-

ﬁed by ﬁrst analyzing all the modifying risk factors

to the target population, which can be split into three

types of factors: (1) behavioral factors (ie, tobacco

use, alcohol, diet, physical activity, and sodium in-

take), (2) metabolic factors (ie, lipids, blood pressure,

diabetes, obesity), (3) socioeconomic and psychoso-

cial factors (ie, education, symptoms of depression),

(3) grip strength, and household and ambient pollu-

tion (Yusuf et al., 2020). We chose diet and physical

activity as target behaviors for the intervention since

they have a great impact in other factors, such as to-

bacco use, sodium intake, lipids, blood pressure, dia-

betes, obesity, and symptoms of depression (Warbur-

ton and Bredin, 2017; Cena and Calder, 2020).

The semi-structured interviews with patients re-

vealed that personalized advice on physical activity

and healthy eating could increase their well-being and

help them recover. The main barriers to engaging

in these behaviors were occupation, lack of motiva-

tion and physical capability, health status, medication,

weight, weather, and mental health. The main pro-

moters were social support, health beliefs, identiﬁca-

tion with the behavior, pets, and the presence of mon-

itoring devices to track activity. We focused in the

results of the semi-structured interviews and results

from two studies to identify the determinants of hav-

ing a healthy diet and being physically active (Brug,

2008; Sherwood and Jeffery, 2000). We found that the

most important determinants for physical activity are

self-efﬁcacy, health-related outcomes, social interac-

tion, stage of change, and social support. While, for

having a healthy diet, the most important were moni-

toring, personal satisfaction, social comparison, stage

of change, and social support.

The systematic literature review showed that to

carry out a fully-automated BCI it is necessary to set

up a data management platform and a communica-

tion system, incorporate algorithms that can automate

content delivery, specify the intended personalization

features, and ﬁnd proper instruments to evaluate the

intervention. Previous studies carried out BCI to im-

prove physical activity, diet, enhance disease moni-

toring, and self-management (Mayberry et al., 2021;

Kelly et al., 2020; Polgreen et al., 2018; Cadilhac

et al., 2020; Chokshi et al., 2018; Horner et al., 2017;

Thiengwittayaporn et al., 2021). Three of those de-

veloped their own data management platforms (Pol-

green et al., 2018; Thiengwittayaporn et al., 2021;

Cadilhac et al., 2020), while the remaining used com-

mercially available platforms (Mayberry et al., 2021;

Kelly et al., 2020; Horner et al., 2017; Chokshi et al.,

2018). Every study used rule-based models to au-

tomate content delivery. Personalization features in-

cluded patient-centered goals (Mayberry et al., 2021;

Kelly et al., 2020; Cadilhac et al., 2020), barriers to

behavior change (Kelly et al., 2020), timing (May-

berry et al., 2021; Kelly et al., 2020; Cadilhac et al.,

2020) and frequency (Kelly et al., 2020) of message

delivery, functional ability (Cadilhac et al., 2020),

disease stage (Thiengwittayaporn et al., 2021), com-

munication channel (Mayberry et al., 2021; Chokshi

et al., 2018; Cadilhac et al., 2020), feedback based

on activity (Polgreen et al., 2018; Chokshi et al.,

2018; Horner et al., 2017), baseline demographic data

(Horner et al., 2017), and stage of behavior change

(Horner et al., 2017). Nevertheless, personalization

algorithms are not well documented and could be en-

hanced using other approaches that automatically un-

Scale-IT-up 2023 - Workshop on Best Practices for Scaling-Up Digital Innovations in Healthcare

598

derstand what are the patient’s needs and adjust con-

tent.

Based on the above, we have the following objec-

tives for the solution:

• The end users are healthcare providers and pa-

tients after telemonitoring or surgery who want to

receive messages to improve their ability to self-

manage their condition.

• There should be no additional burden on health-

care providers.

• A ”virtual coach” CA sends text messages to

improve self-efﬁcacy and help patients maintain

healthy behaviors (diet and physical activity).

• Intervention functions will include education, per-

suasion, training, and enablement.

• The language used should be encouraging, posi-

tive, friendly, polite, and lighthearted.

• Messages should be written using behavior

change techniques to allow for replicability.

• Content, communication channel, and dose

should be personalized.

• The follow-up questionnaire already used by the

clinical team should be used to measure the effec-

tiveness of the artifact.

• Improve self-management skills and prepare pa-

tients for the time when they are not supported by

the team (1 year after surgery).

• A platform should be developed that allows

healthcare professionals to select messages sug-

gested by an automation algorithm.

• This platform should be modular and reusable for

evidence-based behavior change interventions.

3.3 Design and Development

The patient journey in cardiothoracic surgery after the

integration of this artifact is shown in Figure 2. Typ-

ically, patients are admitted to the hospital one day

before surgery (preoperative). There, they receive

instructions to facilitate their postoperative recovery.

After surgery, the patient stays in the hospital for 5-7

days, depending on how the surgical procedure went

(hospitalization). After discharge from the hospital,

study participants are selected by a risk assessment

model to use a remote patient monitoring kit (home

telemonitoring). After 3 months of telemonitoring,

patients are asked if they would like to receive text

messages to help them engage in healthy behaviors

(text message intervention).

The text messages were developed using a text

message bank that has already been used with patients

with diabetes and developed with BCTs (MacPherson

et al., 2021). Therefore, we used these messages as

a reference to create a text message bank. The text

message bank was translated into Portuguese, adapted

to the population of the study, and validated by the

nurses to ensure the reliability of the content. This

database may grow as we determine the need for

more messages, which should happen iteratively as

the study progresses.

A platform is being developed to automate the

conversational agent to be used during the pilot study.

During the pilot study, we will use a ﬁxed-frame

dialog to send content to the patient. An example of

the weekly dialog ﬂow can be found in Table 1.

Table 1: Example of a weekly dialog ﬂow.

(Monday) Getting rid of old habits is difﬁcult,

start with simple things: try to eat fruit when you

feel like something sweet or drink water instead of

juice.

(Tuesday) The small steps can lead to big changes

in your health. Every step you take this week is a

step towards a healthier you.

(Wednesday) Think where, when and how you

will exercise today!

(Thursday) Try swapping reﬁned carbohydrates

(eg white bread) for whole grain equivalents (eg

whole wheat bread).

Table 2: Example of a weekly status questionnaire [MoA -

mechanisms of action] UX - user experience).

[MoA/UX] Message (expected answer)

[Knowledge] A balanced diet means eating all

foods in the same amount. (T/F)

[Attitudes towards behavior] I don’t like the taste

of healthy food. (T/F)

[Attitudes towards behavior] I would have more

energy for my family and friends if I exercised

regularly. (T/F)

[Beliefs about consequences] Personally, I exer-

cise to improve my health. (T/F)

[Intention] I intend to eat more fruit. (T/F)

[UX] Do you consider the information sent to you

this week to be relevant?

[UX] Would you like to maintain the frequency

with which we send you messages? If not, indi-

cate the number of times per week.

Designing a Digital Personal Coach to Promote a Healthy Diet and Physical Activity Among Patients After Cardiothoracic Surgery

599

Figure 2: The patient journey at Santa Marta’s hospital will comprise 5 stages: (a) preoperative, (b) hospitalization, (c) home

telemonitoring, (d) text-message intervention, (e) regular follow-up.

3.4 Demonstration

The demonstration of the solution will be ﬁrstly be

done with a small pilot study with 5 patients and dur-

ing a month. The main goal is to understand the

usability, the engagement, and the satisfaction with

the intervention. The text-intervention diagram is de-

picted in Figure 3.

Figure 3: Text-intervention diagram.

3.5 Evaluation

Questionnaires will be used to improve patient expe-

rience (dose, timing, and relevance will be assessed

weekly). An example of a weekly questionnaire is

shown in Table 2. Patient progress will be evaluated

using data from the follow-up questionnaire already

used by the clinical team to assess the patient’s recov-

ery.

4 DISCUSSION AND

CONCLUSION

This paper focuses on explaining the methodology

that led to the development of a text messaging inter-

vention. DSRM was used to ensure the sustainability

and usability of solutions developed to address real-

world problems, such as the one we identiﬁed: lack

of self-efﬁcacy in managing a cardiovascular disease

after cardiothoracic surgery. To do this, it is essential

to engage with all stakeholders involved. Behavior

change theory is used to guide the design of the inter-

vention by guiding a behavior analysis, and help de-

ﬁne the target behavior, developing message content,

and personalize the intervention.

The goal of this work is to help patients improve

their quality of life and well-being and take control of

their disease after the follow-up period in the hospital.

The platform to be developed will be designed to be

reusable for future interventions and ensure usability

by paying attention to the user experience. This text

messaging intervention has not yet been evaluated and

no preliminary results are available.

Currently, we are focusing on developing algo-

rithms to personalize content selection. Unsupervised

algorithms will be used to understand patient needs

using data from previous patients. The next step will

be to develop a modular platform that will allow the

clinical team to monitor the diet and physical activity

Scale-IT-up 2023 - Workshop on Best Practices for Scaling-Up Digital Innovations in Healthcare

600

of patients enrolled in the study and send text mes-

sages to patients who are not enrolled in the study.

ACKNOWLEDGEMENTS

The authors would like to acknowledge the cardio-

thoracic service of the Santa Marta Hospital for their

contributions to carry out the study. This research has

been supported by the project DSAIPA/AI/0094/2020

from the Fundac¸

ao para a Ci

encia e Tecnologia AI 4

COVID-19 Program.

REFERENCES

Akinosun, A. S., Polson, R., Diaz Skeete, Y., De Kock,

J. H., Carragher, L., Leslie, S., Grindle, M., and

Gorely, T. (2021). Digital technology interventions

for risk factor modiﬁcation in patients with cardiovas-

cular disease: Systematic review and meta-analysis.

JMIR mHealth and uHealth, 9(3):e21061.

Bauer, U. E., Briss, P. A., Goodman, R. A., and Bowman,

B. A. (2014). Prevention of chronic disease in the 21st

century: elimination of the leading preventable causes

of premature death and disability in the USA. The

Lancet, 384(9937):45–52.

Brug, J. (2008). Determinants of healthy eating: motiva-

tion, abilities and environmental opportunities. Family

Practice, 25:i50–i55.

Cadilhac, D. A., Andrew, N. E., Busingye, D., Cameron,

J., Thrift, A. G., Purvis, T., Li, J. C., Kneebone, I.,

Thijs, V., Hackett, M. L., Lannin, N. A., Kilkenny,

M. F., and on behalf of the ReCAPS investigators

(2020). Pilot randomised clinical trial of an eHealth,

self-management support intervention (iVERVE) for

stroke: feasibility assessment in survivors 12–24

months post-event. Pilot and Feasibility Studies,

6(1):172.

Carey, R. N., Connell, L. E., Johnston, M., Rothman, A. J.,

de Bruin, M., Kelly, M. P., and Michie, S. (2018). Be-

havior change techniques and their mechanisms of ac-

tion: A synthesis of links described in published inter-

vention literature. Annals of Behavioral Medicine.

Cena, H. and Calder, P. C. (2020). Deﬁning a healthy diet:

Evidence for the role of contemporary dietary patterns

in health and disease. Nutrients, 12(2):334.

Chaix, B., Bibault, J.-E., Pienkowski, A., Delamon, G.,

Guillemass

e, A., Nectoux, P., and Brouard, B. (2019).

When chatbots meet patients: One-year prospective

study of conversations between patients with breast

cancer and a chatbot. JMIR Cancer, 5(1):e12856.

Chokshi, N. P., Adusumalli, S., Small, D. S., Morris, A.,

Feingold, J., Ha, Y. P., Lynch, M. D., Rareshide, C.

A. L., Hilbert, V., and Patel, M. S. (2018). Loss-

framed ﬁnancial incentives and personalized goal-

setting to increase physical activity among ischemic

heart disease patients using wearable devices: The

ACTIVE REWARD randomized trial. Journal of the

American Heart Association, 7(12):e009173.

Fitzpatrick, K. K., Darcy, A., and Vierhile, M. (2017). De-

livering cognitive behavior therapy to young adults

with symptoms of depression and anxiety using a fully

automated conversational agent (woebot): A random-

ized controlled trial. JMIR Mental Health, 4(2):e19.

Gardner, G., Elliott, D., Gill, J., Grifﬁn, M., and Crawford,

M. (2005). Patient experiences following cardiotho-

racic surgery: An interview study. European Journal

of Cardiovascular Nursing, 4(3):242–250.

Ghandeharioun, A., McDuff, D., Czerwinski, M., and

Rowan, K. (2019). EMMA: An emotion-aware well-

being chatbot. In 2019 8th International Confer-

ence on Affective Computing and Intelligent Interac-

tion (ACII), pages 1–7. IEEE.

Grimmett, C., Corbett, T., Brunet, J., Shepherd, J., Pinto,

B. M., May, C. R., and Foster, C. (2019). Systematic

review and meta-analysis of maintenance of physical

activity behaviour change in cancer survivors. Inter-

national Journal of Behavioral Nutrition and Physical

Activity, 16(1):37.

Halldorsdottir, H., Thoroddsen, A., and Ingadottir, B.

(2020). Impact of technology-based patient education

on modiﬁable cardiovascular risk factors of people

with coronary heart disease: A systematic review. Pa-

tient Education and Counseling, 103(10):2018–2028.

Hevner, March, Park, and Ram (2004). Design science

in information systems research. MIS Quarterly,

28(1):75.

Horner, G. N., Agboola, S., Jethwani, K., Tan-McGrory, A.,

and Lopez, L. (2017). Designing patient-centered text

messaging interventions for increasing physical activ-

ity among participants with type 2 diabetes: Qualita-

tive results from the text to move intervention. JMIR

mHealth and uHealth, 5(4):e54.

Johnston, M., Carey, R. N., Connell Bohlen, L. E., John-

ston, D. W., Rothman, A. J., de Bruin, M., Kelly,

M. P., Groarke, H., and Michie, S. (2021). Devel-

opment of an online tool for linking behavior change

techniques and mechanisms of action based on trian-

gulation of ﬁndings from literature synthesis and ex-

pert consensus. Translational Behavioral Medicine,

11(5):1049–1065.

Kelly, J. T., Conley, M., Hoffmann, T., Craig, J. C., Tong,

A., Reidlinger, D. P., Reeves, M. M., Howard, K., Kr-

ishnasamy, R., Kurtkoti, J., Palmer, S. C., Johnson,

D. W., and Campbell, K. L. (2020). A coaching pro-

gram to improve dietary intake of patients with CKD:

ENTICE-CKD. Clinical Journal of the American So-

ciety of Nephrology, 15(3):330–340.

Kontis, V., Mathers, C. D., Rehm, J., Stevens, G. A., Shield,

K. D., Bonita, R., Riley, L. M., Poznyak, V., Beagle-

hole, R., and Ezzati, M. (2014). Contribution of six

risk factors to achieving the 25x25 non-communicable

disease mortality reduction target: a modelling study.

The Lancet, 384(9941):427–437.

Londral, A., Azevedo, S., Dias, P., Ramos, C., Santos, J.,

Martins, F., Silva, R., Semedo, H., Vital, C., Gualdino,

A., Falc

ao, J., Lap

ao, L. V., Coelho, P., and Fragata,

Designing a Digital Personal Coach to Promote a Healthy Diet and Physical Activity Among Patients After Cardiothoracic Surgery

601

J. G. (2022). Developing and validating high-value pa-

tient digital follow-up services: a pilot study in cardiac

surgery. BMC health services research, 22(1):680.

MacPherson, M., Cranston, K., Johnston, C., Locke, S.,

and Jung, M. E. (2021). Evaluation and reﬁnement

of a bank of SMS text messages to promote behav-

ior change adherence following a diabetes prevention

program: Survey study. JMIR Formative Research,

5(8):e28163.

Mayberry, L. S., Berg, C. A., Greevy, R. A., Nelson,

L. A., Bergner, E. M., Wallston, K. A., Harper, K. J.,

and Elasy, T. A. (2021). Mixed-methods random-

ized evaluation of FAMS: A mobile phone-delivered

intervention to improve family/friend involvement in

adults’ type 2 diabetes self-care. Annals of Behavioral

Medicine, 55(2):165–178.

Michie, S., Richardson, M., Johnston, M., Abraham, C.,

Francis, J., Hardeman, W., Eccles, M. P., Cane, J.,

and Wood, C. E. (2013). The behavior change tech-

nique taxonomy (v1) of 93 hierarchically clustered

techniques: Building an international consensus for

the reporting of behavior change interventions. An-

nals of Behavioral Medicine, 46(1):81–95.

Michie, S., van Stralen, M. M., and West, R. (2011). The

behaviour change wheel: A new method for character-

ising and designing behaviour change interventions.

Implementation Science, 6(1):42.

Peffers, K., Tuunanen, T., Rothenberger, M. A., and Chat-

terjee, S. (2007). A design science research method-

ology for information systems research. Journal of

Management Information Systems, 24(3):45–77.

Pinto, A., Faiz, O., Davis, R., Almoudaris, A., and Vincent,

C. (2016). Surgical complications and their impact

on patients’ psychosocial well-being: a systematic re-

view and meta-analysis. BMJ Open, 6(2):e007224.

Polgreen, L. A., Anthony, C., Carr, L., Simmering, J. E.,

Evans, N. J., Foster, E. D., Segre, A. M., Cremer,

J. F., and Polgreen, P. M. (2018). The effect of au-

tomated text messaging and goal setting on pedome-

ter adherence and physical activity in patients with di-

abetes: A randomized controlled trial. PLOS ONE,

13(5):e0195797.

Schmitz, C. (2012). LimeSurvey: An Open Source survey

tool. LimeSurvey Project, Hamburg, Germany.

Sherwood, N. E. and Jeffery, R. W. (2000). The behav-

ioral determinants of exercise: implications for physi-

cal activity interventions. Annual Review of Nutrition,

20:21–44.

Stephens, T. N., Joerin, A., Rauws, M., and Werk, L. N.

(2019). Feasibility of pediatric obesity and predi-

abetes treatment support through tess, the AI be-

havioral coaching chatbot. Translational Behavioral

Medicine, 9(3):440–447.

Tadas, S., Pretorius, C., Foster, E. J., Gorely, T., Leslie,

S. J., and Coyle, D. (2021). Transitions in technology-

mediated cardiac rehabilitation and self-management:

Qualitative study using the theoretical domains frame-

work. JMIR Cardio, 5(2):e30428.

Thiengwittayaporn, S., Wattanapreechanon, P., Sakon, P.,

Peethong, A., Ratisoontorn, N., Charoenphandhu, N.,

and Charoensiriwath, S. (2021). Development of a

mobile application to improve exercise accuracy and

quality of life in knee osteoarthritis patients: a ran-

domized controlled trial. Archives of Orthopaedic and

Trauma Surgery.

Tully, P. J. (2013). Quality-of-life measures for cardiac

surgery practice and research: a review and primer.

The Journal of Extra-Corporeal Technology, 45(1):8–

15.

Warburton, D. E. and Bredin, S. S. (2017). Health bene-

ﬁts of physical activity: a systematic review of current

systematic reviews. Current Opinion in Cardiology,

32(5):541–556.

WHO (2021). Cardiovascular diseases (CVDs).

Yusuf, S., Joseph, P., Rangarajan, S., Islam, S., Mente, A.,

Hystad, P., Brauer, M., Kutty, V. R., Gupta, R., Wiel-

gosz, A., AlHabib, K. F., Dans, A., Lopez-Jaramillo,

P., Avezum, A., Lanas, F., Oguz, A., Kruger, I. M.,

Diaz, R., Yusoff, K., Mony, P., Chifamba, J., Yeates,

K., Kelishadi, R., Yusufali, A., Khatib, R., Rahman,

O., Zatonska, K., Iqbal, R., Wei, L., Bo, H., Rosen-

gren, A., Kaur, M., Mohan, V., Lear, S. A., Teo, K. K.,

Leong, D., O’Donnell, M., McKee, M., and Dagenais,

G. (2020). Modiﬁable risk factors, cardiovascular dis-

ease, and mortality in 155 722 individuals from 21

high-income, middle-income, and low-income coun-

tries (PURE): a prospective cohort study. The Lancet,

395(10226):795–808.

Scale-IT-up 2023 - Workshop on Best Practices for Scaling-Up Digital Innovations in Healthcare

602