Fit-Twin: A Digital Twin of a User with Wearables and Context as

Input for Health Promotion

Muhammad Sulaiman, Anne Håkansson and Randi Karlsen

Department of Computer Science, UiT The Arctic University of Norway, Tromsø, Norway

Keywords: Digital Twins, Proactive Health, Wearables, Artificial Intelligence, Health Promotion.

Abstract: Digital health contributes to health promotion by empowering the user with the holistic view of their health.

Health promotion is to enable the user to take control over their health. The availability of wearables has

contributed to the shift in healthcare, that is more connected, predictive, and proactive. Proactive in healthcare

is to predict and prevent a situation, beforehand. This shift in healthcare puts the user in charge of most health-

related decisions. Innovative technologies like AI already contribute to the cause by applying reasoning and

negotiation to the collected health data to provide timely interventions to the user. The availability of real-

time data from sensors that the user wears all the time allows more opportunities with new health insights.

One such prospect is the use of digital twins, which provides personalization and precision. Digital twins also

allow risk-free modelling for more accurate outcomes. A user digital twin is not just a virtual replica, but it

combines all the factors that can impact the user. The context of the user is a prominent factor in healthcare.

The paper establishes the need for digital twins in health promotion. In this paper, a Fit-twin is presented that

mimics a user with wearables and the user context as input. The Fit-twin is implemented using Azure digital

twins, Fitbit charge, and local context API. This allows one-way communication between the user and the

Fit-twin. The outcome is a user digital twin that can be used for health promotion by applying predictive

capabilities.

1 INTRODUCTION

Well-being is the combination of factors associated

with lifestyle, health, living circumstances, and

economic situation (CDC, 2018). The increase in

well-being will have a positive impact on the health

of a user. Boost in well-being directly impact better

productivity in individuals (CDC, 2018).

Health promotion is directly associated with well-

being of an individual. Health promotion is to enable

people to increase control over their health (WHO,

Health promotion). This increased control leads to

user-empowerment which can then improve wellness.

Enabling users to take control of their health is the

shift in healthcare that gives the user an active role.

The paradigm shift is essential because healthcare

is dealing with many obstacles (CDC - Global Health

2022), some of these are related to practical issues:

the shortage of resources, and the hospital's ability to

cope (CDC - Global Health 2022). But some

challenges are part of the health approach, for

example, the reactive approach, which is to wait for

something to happen, an example of crisis

management. This reactive approach (Alexis Wise,

2020) is useful, but it depreciates user-empowerment.

Digital health can contribute to this by supporting and

making healthcare more real-time (Argyres et al.,

2022), providing the tools that enable the user with

the holistic view of their health.

This shift establishes that future healthcare will be

more connected, predictive, and proactive (Deloitte,

2021). Proactive in healthcare is to predict and

prevent a situation beforehand, before becoming sick

(Sulaiman et al., 2021). This enables care that

empowers the user to promote health and well-being.

A definition of healthcare also emphasizes this shift:

"health is the ability to adapt and to self-manage, in

the face of social, physical and emotional challenges"

(Huber et al., 2016). The goal of this shift is to

establish user empowerment to support healthcare.

Digital health can contribute to the cause by

providing the tools (Marwaha et al., 2022). Newly

available devices that the user can wear furnish new

insights into the health information of the user. This

information was not part of the traditional health

systems. The information can be the bio-signals, and

Sulaiman, M., Håkansson, A. and Karlsen, R.

Fit-Twin: A Digital Twin of a User with Wearables and Context as Input for Health Promotion.

DOI: 10.5220/0011735900003414

In Proceedings of the 16th International Joint Conference on Biomedical Engineering Systems and Technologies (BIOSTEC 2023) - Volume 5: HEALTHINF, pages 461-470

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)

461

patterns of the user, for example, the activity data,

heart rate data, sleep data, and change in data after an

intervention.

Wearable technology enables the user to

continuously monitor activities, behavior, daily

patterns, and other parameters. It is a combination of

sensors that a user wears on their body. The common

features are heart rate, activities, sleep data, SpO2,

and body temperature (Loucks et al., 2021). Some

advanced sensors also provide blood pressure,

electrocardiogram (ECG), and ballistocardiogram

(BCG) (Min Wu, 2021). These devices are worn by

the user on the wrist, head, finger, or other suitable

places. The data from wearables provide the basis for

user-empowerment to promote health. Wearables

also allow personalization by considering the

uniqueness of the user. This continuous abundance of

data also fuels the need for innovation with Artificial

Intelligence (AI) to analyze and provide support that

is tailored to user needs.

We are part of the fourth revolution in the

industry, also known as Industry 4.0. This

revolutionized most industries by shifting the focus

from manual to automation. IBM defines Industry 4.0

as "revolutionizing the way companies manufacture,

improve and distribute their products. Manufacturers

are integrating new technologies, including Internet

of Things (IoT), cloud computing and analytics, and

AI and machine learning into their production

facilities and throughout their operations." (IBM,

2022). This evolution also affects healthcare that is

adapting to the evolution by introducing healthcare

4.0 (Li & Carayon, 2021) which brings in new trends

with AI and robotics to provide better and cost-

effective healthcare. The goal is to provide healthcare

accessible to everyone and to implement automation

in clinical decision-making and enable early detection

and feedback.

Industry 4.0 also boosts an application area with

digital twins. Digital twin is defined as "A virtual

model designed to accurately reflect a physical

object. The object being studied — for example, an

airplane — is outfitted with various sensors related to

vital areas of functionality. These sensors produce

data about different aspects of the physical object’s

performance, such as energy output, temperature,

weather conditions and more. This data is then

relayed to a processing system and applied to the

digital copy. " (IBM, digital twin 2022). A digital

twin is not merely a replica, but it considers the

dynamic context and all the factors associated with

the object. The digital twin market is growing and is

estimated to reach USD 48.2 billion by 2026 (IBM,

digital twin 2022).

The future of healthcare is also more personalized

(Deloitte, 2021), meaning every user is unique when

it comes to behavior and internal and contextual

states. The digital twin application in healthcare will

revolutionize personalization and precision medicine.

It will also allow more opportunities for health

promotion and well-being. The real-time data will

lead to more precise interventions and generate

possible improvements in the health of the user.

AI can contribute to this by applying algorithms

to the virtual representation, by analysing the data

from sensors and other heterogeneous sources. The

collected data can be combined to feed an AI model

(Agarwal, 2021) for predictive analytics. The model

can then provide timely interventions back to the real

asset. The data from the real-time object can allow

pattern recognition; to early detect and manage

health. A challenge would be the heterogenous data

streams because of multiple sensors and wearables.

This paper provides insights into innovations in

healthcare with digital twins. It also provides

fragments on how digital twins can be utilized for

health promotion and risk prevention, by

understanding the user as a twin, and the significance

of the context and states of the user. The later part of

the paper furnishes wearables as a source for digital

twins. Ultimately, a proof-of-concept implementation

of Fit-twin that uses Fitbit and Azure digital twins to

implement a digital twin of a user with the context.

The outcome is a Fit-twin that mimics the real user,

to collect data from wearables and the context in real-

time. This allows a one-way communication between

the user and the Fit-twin. Fit-twin will combine

multiple parameters of the user for predictive

analytics with AI to provide timely intervention back

to the user for health promotion.

2 RELATED WORK

A digital twin is not a new concept, in 1991, Micheal

Grieves (Grieves, 2007) defined the concept of the

digital twin in manufacturing: this accentuates it as a

process. The idea of a digital twin can even go back

to 1960 when NASA (Grieves, 2007) used it for a

space mission exploration to simulate.

The current evolution is because of the

availability of sensors, that can connect to the real

source and the computational power of AI to apply

algorithms to the generated data. One such existing

research presented, an intelligent personal MINI-Me

(Håkansson & Hartung, 2014) that allows contextual-

based decision-making for the individual.

Personalized data and context data are combined to

HEALTHINF 2023 - 16th International Conference on Health Informatics

462

create a MINI-Me that can interact with other devices

within its context. This MINI-Me motivated the

formation of Fit-twin, which uses personalized data

from wearables and context data to allow better

decision-making.

In healthcare, there are many application areas for

digital twins - some of these are a work in progress

(Hilary, 2021). The implementation of heart twin

(Coorey et al., 2022), allows cardiologists to monitor

heart performance for early detection. Another study

(Dan Bland, 2018) implemented hospital twins that

are used for workflow analysis, system redesign, and

process improvement methodologies.

Philips used device twins (Philips, 2021) for the

complex equipment to ease the maintenance of these

devices. A symptom tracker (Babylon, 2020) by

Babylon Health introduced the concept of symptom

tracking during covid-19 by connecting to the

wearables of the user. This is somewhat closer to the

concept of the digital twin of the user. The Table 1

shows some examples of digital twins in healthcare.

Table 1: Digital twins in healthcare.

ital twin Pur

ose

Oxygen tank twin Used in hospital to track

real-time availability of

tanks.

Precision-medicine A twin used to see impact

of the medicine given to an

individual.

Heart-twin Monitor heart of an

individual.

The related work highlights areas of digital twins’

implementation, within healthcare, but establishes

that no existing research, in particular, creates a

digital twin of the user for health promotion and risk

prevention. That also considers the states of the user;

both internal and contextual. This establishes that

digital twins using wearables as a source can provide

new insights into user-level decision-making. The

goal would be to empower the user with more

personalization and timely support.

3 DIGITAL TWIN FOR HEALTH

PROMOTION OF A USER

A digital twin is defined as "a digital twin is a virtual

model to accurately reflect a physical object" (IBM,

digital twin 2022). The digital replica is a virtual

representation of a physical asset. The virtual copy is

a combination of all the elements, associated with that

asset. Digital twin connects real and virtual worlds

by binding them with real-time data. A change in a

physical asset can be seen on the digital twin

instantly. The digital twin is not a 3D model of a

physical object, but it includes all the parameters

where the asset exists and also considers risks. An

illustration is a digital twin of a car: that can collect

data from sensors attached to the real car. It includes

all the information about the car e.g., mechanical, and

electric parts. Also, the environment where the car

exists, weather, temperature, and traffic.

Information flow from the digital twin is a 2-way

process. First, the data is collected continuously from

the sensors. This is the real-time data, stored and fed

to the virtual twin. This is a one-way flow; the second

part is where the collected data models are formed

that can be predictive and preventive. These models

can then provide useful information back to the real

asset.

When it comes to health promotion with digital

twins it is noteworthy to consider health promotion as

the: "process of health promotion is more than the

absence of disease; it is a resource that allows people

to realize their aspirations, satisfy their needs and to

cope with the environment to live a long, productive,

and fruitful life." (CDC, 2018). This establishes that

user empowerment can significantly impact health,

and digital twins can allow personalization. A user is

part of a dynamic environment that is very

impressionistic, meaning it affects someone

differently, for instance, a higher pollen count can be

a bad situation for someone who is allergic to pollen.

This information is drawn from the user profile.

A digital twin is usually mixed with simulation,

although it allows simulation it is not just a

simulation. The difference is that a simulation is just

a process that does not need 2-way communication,

and does not need real-time data. The digital twin can

run multiple simulations but also considers real-time

data. The simulation focuses on a process to simulate

an outcome, it depreciates the need for states.

Digital twins for the health promotion of a user

must include all the parameters that can have an

impact on the user. The wearables will be the source

of health information in real time. The

communication to the user is characterized as 2-way.

This will be achieved by an intervention system that

allows timely intervention to the user and gets

feedback from the user.

To understand the parameters of a user, let us

consider the context of the user. The context is:

"the interrelated conditions in which something

exists or occurs" (Merriam-webster). A study

(Sulaiman et al., 2022) explained user context as

anything that can impact the user positively or

Fit-Twin: A Digital Twin of a User with Wearables and Context as Input for Health Promotion

463

negatively. Table 2 shows the context of the user with

some examples. Many different circumstances e.g.,

environment, and surroundings, can contribute to the

context. A user’s context is based on the location and

conditions where the user exists. Bad air quality can

have a bad impact and thus a part of the context of the

user. Context is dynamic with ever-changing

surroundings. A system must be adaptive to cope with

the dynamic changes in context.

Table 2: Context of the user.

Environment/Surroundings

Weather,

air pollution,

threats, pollen count,

and outbreak

Another parameter is the user characteristics

which is the combination of behavior and daily

patterns. Table 3 shows the parameters of user

characteristics. The behaviors and daily patterns are

very unique, different users can act and distinctly

react to anything. For behavior change, it is important

to consider multiple factors considering the user.

When it comes to daily patterns, they depend on user

preferences, for instance, someone works during the

night and sleeps during the day.

Table 3: User characteristics.

Daily patterns Behavior

Step counts,

activity,

sleep

and other

Lifestyle,

schedule,

and preferences

Finally, the user profile includes the physical and

other attributes of the user along with all the vitals.

This information is collected from the user explicitly

and implicitly. Table 4 presents the user-profile

example. The user profile also combines factors that

are the core component of the digital twins. The

physical and mental states and the current state of the

user are also formulated from the profile.

Table 4: User profile.

Vitals Heart rate, SpO2,

temperature, step-count,

activity, sleep, and

readiness score

Physical attributes Body mass index (BMI)

Family history Disease history

Location Actual location of the user

The next step after the formation of user

parameters is to look for sources and resources to

collect data from. Table 5 presents the sources

available for collecting data. The table also provides

extensive information for each of the parameters of

the user.

Table 5: User parameters with sources for input to the user-

twin.

Sources Parameters Actual source

Wearables Daily patterns,

activity data,

health

information,

real-time

sudden changes

Wearable

devices: Fitbit

APIs Weather, air

quality, Pollen,

warning

YR.no or other

metrological

source

Sensors Humidity,

temperature,

noise, rust, air-

quality

Netatmo or a

user-made

project

Figure 1: A digital twin of the user with two modules.

Figure 1 presents the digital twin of a user with

two modules. The information collected and

combined can create a digital twin for the user. The

continuous data from the wearables and the context is

significantly important to have real-time changes in

user internal and contextual states. This information

is then used for feeding the digital twin and collecting

for modelling.

4 WEARABLES AS A DATA

SOURCE FOR INPUT TO THE

FIT-TWIN ALONG WITH APIS

FOR CONTEXT INPUT

After determining the input of the digital twin, it is

specified that it requires continuous data from

multiple sources. Wearables are pivotal for

integrating user contextual data and personalized data

Context

Wearables

HEALTHINF 2023 - 16th International Conference on Health Informatics

464

as input. The paper presents a Fit-twin that uses

wearables as the input source. Wearables include

devices that users can wear, for instance, fitness

trackers and smartwatches (Loucks et al., 2021). They

are combined with different sensors that are designed

to collect data. These devices enable continuous

measurement of different body signals and

parameters. Wearables are divided into two main

types (Min Wu, 2021):

─ For health professionals

─ For consumers

In our use case of digital twins, the focus will be

on consumer-level wearables. These are further

divided into three categories.

Wearable Fitness Trackers. These are wrist activity

trackers that provide a sensor to collect information.

The common activity trackers are Fitbit (Fitbit, 2022),

Polar (Polar, 2022), Garmin (Garmin, 2022) and

others. They have overlapping features, Table 6

provides a comparison between some fitness trackers

available.

Table 6: Wearables fitness trackers.

Fitbit

charge 5

Polar

Gramin

Vivo

Oura ring Mi band

Activity

tracking,

GPS,

Continuo

us heart

rate,

breathing

rate,

Stress

managem

ent score,

SpO2,

Skin

temperatu

re, sleep

and

inactivity

Activity,

Sleep, heart

rate

measureme

nt and

inactivity

Sensors:

acceleromet

er, optical

heart rate

monitor

Activity,

Sleep, heart

rate

measureme

nt and

inactivity

Sensors:

acceleromet

er,

pedometer

Activity,

Steps,

Inactivity

time, heart

rate, body

temperature

and sleep

Sensors:

acceleromet

er, optical

heart rate

monitor

Activity

tracking,

GPS,

Continuo

us heart

rate,

Stress

managem

ent score,

SpO2,

sleep and

inactivity

time

Smart Health Watches. These are smartwatches that

are also capable of continuous measurement of body

signals and patterns. An example is the Apple Watch,

and Samsung watches. The new pixel watch is

another new addition to this category.

Wearable Biosensors. Wearable biosensors are used

for the continuous monitoring of different parameters

more accurately. A common illustration is the

continuous glucose monitoring sensor by Abbott.

Libre-2 sensors provide continuous monitoring and

alarms for blood glucose levels.

Figure 2: Wearable sensors (Rodrigues et al., 2018).

Figure 2 provides different sensors, their type, and

use-case. It also presents details on where the user can

wear these devices.

Figure 3: Holistic view of the user.

An example of a holistic view is shown in Figure

3. This holistic view of the user is provided by the

wearables. It is the first step towards digital twin

formation. The holistic view provides multiple

parameters about the user. It includes user daily

patterns, current state, data from sensors, and real-

time vitals.

A comparison in Table 7 is drawn between two

wearable devices for our use case. The first is a Fitbit

fitness tracker, and the second is the Oura ring 3. The

comparison furnishes details about the parameters

and features of both.

The head-to-head comparison shows that both

trackers are useful for collecting data from the user.

They can provide in-depth real-time data from

sensors. The data collected from these can be used as

input for our digital twin.

The next step is to compare the APIs of both

devices. The Fitbit API (Fitbit-API, 2022) provides

endpoints to integrate and collect data. The collected

data can be stored and processed to be used for input

to the digital twin. Oura ring also provides Oura API

V1 (Oura-API). The outcome of this comparison

shows that Oura ring is an excellent choice when it

comes to sleep tracking, but Fitbit leads the way as a

Fit-Twin: A Digital Twin of a User with Wearables and Context as Input for Health Promotion

465

multipurpose tracker. The Fitbit API is well-

established and documented.

Table 7: Comparison between Fitbit and Oura ring.

Fitbit features Oura ring

-Wrist

-Continuous heart rate

-SpO2 in sleep

-Activity tracker

-The time, duration, type

and intensity

-Active minutes

-Readiness score

-No temperature sensor

-Location

-Ring

-Continuous heart rate

-SpO2 in sleep

-Activity tracker

-The time, duration, type

and intensity

-Only with MET for

activity tracking

-Readiness score

-Temperature sensor

-Location

Integration to Google Fit is still a work in

progress. Although Google acquired Fitbit some time

ago. There is no direct official syncing between the

platforms. A third-party (HealthSync) application is

used for syncing these together.

So, with this extensive comparison between

different wearables. It was concluded that Fitbit

would be an ideal choice for our use case. The data

from Fitbit would be fed to the digital twin. Fitbit

profile can also establish the user-profile setting

required.

Choosing the wearable is a starting point for our

digital twin. The next step would be to collect data

from the context. In this use case, YR.no is used as a

source. The metrological API provides the following

features.

─ YR.no (YR, 2022) as a Context Source. Weather

data, Met warning, Air-quality, Pollen, Storm, Ice

mapping, and ocean forecast.

Another choice would be to use indoor-outdoor

weather stations. These can be built using a

microcontroller and sensors. An example of an indoor

context station is given below.

─ Sensors with microcontroller:

 Arduino Uno R3 (Arduino): A microcontroller

 MQ-135 (MQ135): A gas sensor

 BME 280 (Bosch): Temperature, humidity, and

pressure sensor

 PM2.5 (Adafruit): Dust and particulate matter

Sensor

Figure 4: Sources and resources for the user-twin.

After combining the sources and resources,

Figure 4 presents sources and resources to input for

the digital twin of a user. Fitbit as a wearable can

provide real-time data and collect daily patterns. The

Fitbit API can also provide user-profile information.

The contextual data can be collected from YR.no as a

resource. This data is based on the location of the

user. Other indoor sensors can be used for collecting

indoor contextual data.

5 FIT-TWIN- A DIGITAL TWIN

FOR USER (FITBIT + AZURE +

CONTEXT)

After the identification of sources for input to the Fit-

twin. The next step is to implement each of the

components. Implementation components for the Fit-

twin are listed and briefly explained.

Fitbit Charge 5. It is a fitness tracker that is worn on

the wrist. It provides continuous measurement of

activity, heart rate, calories burned, active minutes,

breathing rate, and sleep. The collected information is

displayed on-screen or can be synced to the mobile

device. Fitbit charge 5 possesses multiple sensors

like an optical heart rate monitor, a 3-axis

accelerometer, built-in GPS + GLONASS, and red

and infrared sensors.

Fitbit Web API. Fitbit provides a public API for the

integration and collection of data. The data collected

is stored and then retrieved using the Web API. The

endpoints are accessible for the retrieval of data.

Examples of some end-point scopes are activity,

sleep, heart rate, location, and oxygen saturation. The

retrieved data is then stored and fed to the digital twin.

HEALTHINF 2023 - 16th International Conference on Health Informatics

466

YR.no. YR provides a well-documented API for

metrological data. To integrate the context data for

our digital twin YR is a good source. The available

end-points are weather, warnings, air quality,

turbulence, icemap and ocean map. This data

incorporates current, daily, monthly, and yearly

information on temperature, precipitation, and wind.

5.1 Implementation Logic

When it comes to implementation the goal is to have

two digital twins combined with a relationship to

implement a one-user Fit-twin. The first step would

be, to understand how the user digital twin is created.

The user digital twin is a digital replica of an actual

user. It replicates and is in sync with the real user by

input stream of real-time data from the wearables.

Figure 5 demonstrates how the input works for the

user twin. The implementation is based on Azure

digital twins (Azure-DT). Azure furnishes a complete

implementation framework for creating digital twins.

Figure 5: Fit-twin with input stream.

In this proof-of-concept, two digital twins will be

implemented and connected to create a Fit-twin. The

first step is to create a model of both twins, define the

relationship, and connect this model to an actual

source to demonstrate a working Fit-twin.

5.1.1 Creating the Model

Azure provides Digital Twin Definition Language

(DTDL), which uses JSON-LD, a method to encode

linked data to create a model of a twin. This model is

a combination of the interface that requires four key

components, based on our use case. The key segments

are presented with the two models, User-model and

the context model.

 User Model

─ Properties: User profile: ID, Physical attributes,

location

─ Telemetry: Heart rate, SpO2, Activity, Sleep

─ Components: When we want to combine models,

in this case, we will add the context model as part

of it.

─ Relationships: It defines the user model

relationship with the context.

 Context model

─ Properties: Location of the user

─ Telemetry: Weather, air quality, temperature,

warnings

─ Components: Add the user model as part of the

context model

─ Relationships: It provides information to the user

model.

The actual implementation using Visual

Studio Code (VS). The code below is the pseudo code

of both models.

User-Model: JSON.

"@id": "User-twin;1",

"@type": "Interface",

"displayName": "User-twin",

"contents": [

{

"@type": "Telemetry",

"name": "BMI",

"schema": "double"

{

"@type": "Property",

"name": "Heart_rate",

"writable": true,

"schema": "double"

Context-model: JSON.

{

"@type": "Telemetry",

"name": "Location",

"schema": "double"

Relationship between both:

{

"@type":

"Relationship",

"name": "contains",

"target": "context-

twin"

Fit-Twin: A Digital Twin of a User with Wearables and Context as Input for Health Promotion

467

5.1.2 Creating an Instance of the Model on

Azure

After creating the models, the next step is to upload

this model to Azure and define the access roles. After

creating an instance, the client applications can

directly connect to the digital twin instance. Since we

are not using any client application for this proof-of-

concept, we will be using a sample Azure twin

explorer application (DTe).

5.1.3 Using Azure Twin Explorer, Creating

Fit-Twin

Azure twin explorer is used to create twins and

connect them to the azure instance. The Fit-twin is

connected to the wearables and context twins. The

implementation shows any change in the user data or

context data in the real world is imitated by the Fit-

twin.

─ Both twins are presented before and after

connecting to the real source.

Figure 6: Fit-twin-user-before on the left, and after

connecting to the real source.

Figure 7: Fit-twin-context- before on the left, and after

connecting to the real source.

The client applications can be used, for instance,

the Fitbit dashboard in Figure 8 provides a good-

looking interface for the twin.

The implementation shows that after connecting

to the real source, the user twin will collect and

present any change in real-time shown in figure 6.

The number of steps taken by the user and other data

points will be updated when connected.

This also allows connecting to the real source for

context model. Context twin when connected to the

API will present real-time data as shown in figure 7.

Figure 8: Client- application example for digital twin.

5.1.4 Architecture Diagram of the Fit-Twin

Figure 9: Architecture diagram of the Fit-Twin.

The figure 9 presents the architecture diagram that

combines two twins (user and context) into one

model. It also presents that each of the models has

some properties. The one-way communication from

the user to fit-twin is implemented in this proof-of-

concept. The two-way communication requires

predictive capabilities for interventions.

6 DISCUSSIONS

Fit-twin provides more personalization. Fit-twin

combines the user model and the context model. The

communication from the sensors to the Fit-twin is

considered one-way. The next step is to apply

algorithms to the collected data and simulate it.

An example is to simulate the Fit-twin for a walk,

whether it is safe for the user to walk between points

A and B. The risk prevention mechanism will follow

the timely intervention back to the user with

information about possible risks.

The two-way communication for Fit-twin

requires AI modelling, which will consider all the

parameters and provide predictive analytics to predict

the situation e.g., A real user can ask whether an

activity performed will be beneficial for their health?

HEALTHINF 2023 - 16th International Conference on Health Informatics

468

Fit-twin can then combine all the parameters and

attributes associated with the user to provide an

answer.

The risk-free modelling will allow the simulation

to be applied on the Fit-twin first before applying it to

the actual user. More information about the user

collected and applied to the Fit-twin will help achieve

precision. The goal is to promote health and prevent

risks.

Challenge is with real-time sudden changes,

modelling with real-time data is challenging. Another

challenge is to either use a generic model or a

personalized model.

Fit-twin will also help generate more data

(collecting from multiple parameters) that can be used

for modelling, to provide timely interventions back to

the user.

7 CONCLUSIONS AND FUTURE

WORK

Health promotion is to enable users to take control of

their health. This increase in control contributes to

user empowerment. Wearables along with the context

of the user allow personalization and precision.

Another innovation that reinforces user

empowerment is a digital twin. A digital twin is not

just a virtual replica of an asset, it also combines all

the properties of the asset e.g., context and state.

Digital twins allow 2-way communication between a

resource. The availability of wearables and contextual

APIs that can provide real-time data highlights the

need for creating user-digital twins. In this paper, we

developed a digital twin of a user "Fit-twin". The Fit-

twin is connected to the real user with wearables and

context API. The Fit-twin is created using Azure,

Fitbit charge 5, and a local metrological resource for

context. The outcome is a Fit-twin that mimics the

properties of an actual user. The change in context

and state of the user can be seen on the Fit-twin. The

provided solution only provides one-way

communication for now but provides placeholders to

add predictive capabilities for intervention

mechanisms.

In future, the Fit-twin will allow Just-in-time

interventions generated based on the collected data

from multiple parameters of the user. The

intervention mechanism will depend on prediction

capabilities of AI model to provide the right support

to the user at the right time for health promotion or

risk prevention.

REFERENCES

CDC. (2018, October 31). Well-being concepts. Centers

for Disease Control and Prevention. Retrieved

November 15, 2022, from https://www.cdc.gov/

hrqol/wellbeing.htm

WHO. (n.d.). Health promotion. World Health

Organization. Retrieved November 17, 2022, from

https://www.who.int/health-topics/health-promotion#tab

=tab_1

CDC - Global health. (2022, November 9). CDC - Global

Health. Centers for Disease Control and Prevention.

Retrieved November 17, 2022, from

https://www.cdc.gov/globalhealth/index.html#global-

health-issue

Alexis Wise, E. M. I. (2020, January 9). Transforming

health: Shifting from reactive to proactive and

Predictive Care. MaRS Discovery District. Retrieved

November 17, 2022, from https://www.marsdd

.com/news/transforming-health-shifting-from-reactive-

to-proactive-and-predictive-care/.

Argyres, D., Hung, A., Kennedy, K., Pérez, L., & Tolub,

G. (2022, July 27). Digital Health: An opportunity to

advance health equity. McKinsey & Company.

Retrieved November 17, 2022, from https://www.

mckinsey.com/industries/life-sciences/our-insights/di

gital-health-an-opportunity-to-advance-health-equity

Deloitte (2021). Key contacts Nico Kleyn Managing

Partner- Predicting the future of healthcare and Life

Sciences in 2025. Deloitte Switzerland. Retrieved

November 17, 2022, from https://www2.deloitte.

com/ch/en/pages/life-sciences-and-healthcare/articles/

predicting-the-future-of-healthcare-and-life-sciences-in

-2025.html.

Sulaiman, M., Håkansson, A., & Karlsen, R. (2021). AI-

enabled proactive mhealth: A Review. ICT for Health,

Accessibility and Wellbeing, 94–108. https://doi.org/

10.1007/978-3-030-94209-0_9

Huber, M., van Vliet, M., Giezenberg, M., Winkens, B.,

Heerkens, Y., Dagnelie, P. C., & Knottnerus, J. A.

(2016). Towards a ‘patient-centred’ operationalisation

of the new dynamic concept of Health: A Mixed

Methods Study. BMJ Open, 6(1). https://doi.org/

10.1136/bmjopen-2015-010091

Marwaha, J. S., Landman, A. B., Brat, G. A., Dunn, T., &

Gordon, W. J. (2022). Deploying Digital Health Tools

within large, complex health systems: Key

Considerations for adoption and implementation. Npj

Digital Medicine, 5(1). https://doi.org/10.1038/s41746-

022-00557-1

Loucks, J., Stewart, D., Bucaille, A., & Crossan, G. (2021,

November 30). Wearable Technology in health care:

Getting better all the time. Deloitte Insights. Retrieved

November 17, 2022, from https://www2.deloitte.com/

us/en/insights/industry/

Min Wu, P. D. and J. L. (2021, April 2). Wearable

technology applications in Healthcare: A literature

review. HIMSS. Retrieved November 17, 2022, from

https://www.himss.org/resources/wearable-technology

-applications-healthcare-literature-

Fit-Twin: A Digital Twin of a User with Wearables and Context as Input for Health Promotion

469

What is Industry 4.0 and how does it work? IBM. (n.d.).

Retrieved November 17, 2022, from https://

www.ibm.com/topics/industry-4-0

Li, J., & Carayon, P. (2021). Health care 4.0: A Vision for

smart and connected health care. IISE Transactions on

Healthcare Systems Engineering, 1–10. https://

doi.org/10.1080/24725579.2021.1884627

Agarwal, R. (2021). Predictive analysis in health care system

using AI. Artificial Intelligence in Healthcare, 117–131.

https://doi.org/10.1007/978-981-16-6265-2_8

Grieves, M. (n.d.). (PDF) origins of the Digital Twin

Concept . Retrieved November 17, 2022, from

10.13140/RG.2.2.26367.61609

Håkansson, A., & Hartung, R. (2014). An infrastructure for

individualised and intelligent decision-making and

negotiation in Cyber-physical systems. Procedia

Computer Science, 35, 822–831. https://doi.org/

10.1016/j.procs.2014.08.248

Hilary, M. (2021, August 10). Can Digital Twins lead to a

dramatic improvement in healthcare? - IBA group.

Blog. Retrieved November 17, 2022, from

https://blog.ibagroupit.com/2021/08/can-digital-twins-

lead-to-a-dramatic-improvement-in-

Coorey, G., Figtree, G. A., Fletcher, D. F., & Snelson, V. J.

(2022). The health digital twin to tackle cardiovascular

disease—a review of an emerging interdisciplinary

field. Npj Digital Medicine, 5(1). https://

doi.org/10.1038/s41746-022-00640-7

Dan Bland (2018). Hospital Ward Management and

improvement using Digital Twin. Challenge Advisory

Retrieved November 17, 2022, from https://

www.challenge.org/insights/digital-twin-hospital-

management/.

The rise of the digital twin: How healthcare can benefit -

blog - philips. (n.d.). Retrieved November 17,

2022, from https://www.philips.com/a-w/about/news/

archive/blogs/innovation-matters/

My health. Babylon Health. (n.d.). Retrieved November 17,

2022, from https://www.babylonhealth.com/en-us/

what-we-offer/healthcheck

Merriam-Webster. (n.d.). Context definition & meaning.

Merriam-Webster. Retrieved November 17, 2022, from

https://www.merriam-webster.com/dictionary/context.

Sulaiman, M., Håkansson, A., & Karlsen, R. (2022). A

framework for AI-enabled proactive mHealth with

automated decision-making for a user’s context.

Proceedings of the 15th International Joint Conference

on Biomedical Engineering Systems and Technologies.

https://doi.org/10.5220/0010843200003123

Fitbit. Fitbit Official Site for Activity Trackers & More.

(n.d.). Retrieved January 7, 2023, from https://www.

fitbit.com/

Polar. Heart Rate Monitors, activity trackers and bike

computers | Polar Global. (n.d.). Retrieved January 7,

2023, from https://polar.com/welcome

Garmin . Garmin International | Home. (n.d.). Retrieved

January 7, 2023, from https://www.garmin.com/.

Fitbit Api. Fitbit API-Web-Api. (n.d.). Retrieved January 7,

2023, from https://dev.fitbit.com/build/reference/web-

api/

Oura API. Oura API (2.0) documentation. (n.d.). Retrieved

January 7, 2023, from https://cloud.ouraring.com/

v2/docs

Health sync . Google Play. (n.d.). Retrieved January

7, 2023, from https://play.google.com/store/apps/

details?id=nl.appyhapps.healthsync&hl=en&gl=US

YR. Yr-Norwegian Meteorological Institute. (n.d.).

Retrieved January 7, 2023, from https://www.yr.no/en

Arduino R3. Arduino Documentation | Arduino

Documentation. (n.d.). Retrieved January 7, 2023, from

https://docs.arduino.cc/hardware/uno-rev3

MQ135 . Winsen. (n.d.). Retrieved January 7, 2023, from

https://www.winsen-sensor.com/sensors/voc-

sensor/mq135.html

Bosch. Bosch Sensortec. (n.d.). Retrieved January 7, 2023,

from https://www.bosch-sensortec.com/products/

environmental-sensors/humidity-sensors-bme280/

Adafruit. Adafruit Learning System. (n.d.). Retrieved

January 7, 2023, from https://learn.adafruit.com/pm25-

air-quality-sensor

Azure. Digital Twins – Modeling and Simulations |

Microsoft Azure. (n.d.). Retrieved January 7,

2023, from https://azure.microsoft.com/en-us/

products/digital-twins/

VS. Microsoft Visual Studio. (2021, November 3).

Retrieved January 7, 2023, from https://code.

visualstudio.com/

DTe. GitHub-Azure-samples/digital-twins-explorer. (n.d.).

Retrieved January 7, 2023, from https://github.com/

Azure-Samples/digital-twins-explorer

Rodrigues, J. J., De Rezende Segundo, D. B., Junqueira, H.

A., Sabino, M. H., Prince, R. M., Al-Muhtadi, J., & De

Albuquerque, V. H. (2018). Enabling technologies for

the internet of health things. IEEE Access, 6, 13129–

13141. https://doi.org/10.1109/access.2017.2789329

HEALTHINF 2023 - 16th International Conference on Health Informatics

470