Supporting Alzheimer’s Residential Care

A Novel Indoor Localization System

Andrea Masciadri, Ilaria Carlini, Sara Comai and Fabio Salice

Department of Electronics, Information and Bioengineering Politecnico di Milano,

via Anzani 42, 22100, Como, Italy

Keywords:

Indoor Localization System, Android Mobile Application, iBeacon, Assistive Technology, Internet of Things.

Abstract:

This work illustrates a localization system speciﬁcally designed to be applied in “Il Paese Ritrovato”, a highly

innovative health-care facility for people affected by Alzheimer’s disease in Monza, Italy. Patients are provided

with an iBeacon bracelet broadcasting data packets that are collected through the use of a dense network of

devices acting as receiving antennas. The system evaluates the path-loss of the received signal and corrects

the computed position with a probabilistic approach to avoid wall-crossing. Localization data are merged with

information from other IoT devices such as smart sensors, appliances and expert annotations; the resulting

dataset will be extremely important to analyze behaviors, habits and social interactions among patients.

1 INTRODUCTION

The growing longevity and declined fertility rate of

the industrialized countries have been shifting the age

distribution of population towards increasingly elder

ages. This shift will continue to dangerously accel-

erate and, according to the United Nation forecast

(United Nations and Affairs, 2017), in 2030 the num-

ber of people older than 60 years old will grow to 1.4

billions worldwide. The projection for 2050 shows an

increment to more than double of the 2017 size of the

selected age group, reaching an increment of 116%,

surpassing the number of adolescent and youth aged.

A factor that has been proven to be largely corre-

lated to aging is the manifestation of dementia. Ap-

proximatively 3% percent of people between the ages

of 65 and 74, 19% between 75 and 84, and nearly

half of those over 85 years of age have dementia

(Umphred, 2012). Dementia is a syndrome charac-

terized by symptoms related to memory, problem-

solving, language impairments and other compro-

mised cognitive skills that affects the regular and ev-

eryday activities of a person.

Alzheimer’s disease is the most common form of

progressively disabling degenerative dementia with

onset predominantly in presenile age. It is estimated

that about 60-70% of dementia cases are due to this

condition, while 10-20% to vascular dementia (World

Health Organization, 2015). Numerous studies have

been conducted to identify a possible treatment for

Alzheimer’s disease, but currently no deﬁnitive thera-

pies has been found and due to the progressive course

of the disease, the management of patient’s needs be-

comes essential. The ultimate goal of the care pro-

vided by caregivers is to extend as much as possible

the maintenance of cognitive abilities, to decelerate

the progression of neuronal degeneration and, through

a series of speciﬁc daily activities, to encourage the

development of a natural brain compensation. Keep-

ing the patients active with daily activities is therefore

a prerequisite of the long term treatment centers. In

order to create a stimulating environment, that is also

able to accommodate people and make them feeling

at home, it is required to design a new type of health

care infrastructure.

The principal known methodologies used to esti-

mate a human position in an indoor localization sys-

tem can be divided into three categories (Farid et al.,

2013; Tariq et al., 2017): Proximity Detection, Trian-

gulation and Scene Analysis. The Proximity Detec-

tion method is the simplest in terms of implementa-

tion, it attributes the target’s position to the strongest

received signal (Zafari et al., 2017); the Triangula-

tion techniques uses the geometric properties to com-

pute the target location (Jacq et al., 2017), while the

Scene Analysis uses scene details computed from a

particular point of view to match patterns (Ter

an et al.,

2017). Since the localization system has to guaran-

tee the identiﬁcation of a patient in an environment,

but not its accurate position within it, the selected

272

Masciadri, A., Carlini, I., Comai, S. and Salice, F.

Supporting Alzheimer’s Residential Care - A Novel Indoor Localization System.

DOI: 10.5220/0006859502720278

In Proceedings of the 15th International Joint Conference on e-Business and Telecommunications (ICETE 2018) - Volume 1: DCNET, ICE-B, OPTICS, SIGMAP and WINSYS, pages 272-278

ISBN: 978-989-758-319-3

methodology for this work is based on Proximity De-

tection. The system has been modeled by reversing

the traditional paradigm of indoor localization: Blue-

tooth Beacons are the mobile devices inside the struc-

ture, while the location is determined through a sys-

tem of antennas arranged in both internal and external

environments.

The remainder of the paper is organized as fol-

lows: Section 2 introduces the requirements of an in-

door localization system specially targeted for a res-

idential care, Section 3 overviews the implemented

system, while Section 4 concludes with a brief sum-

mary and mentions future work.

2 REQUIREMENTS

“Il Paese Ritrovato” (Il Paese Ritrovato Project,

2018) is the ﬁrst village in Italy exclusively de-

signed by the La Meridiana cooperative (La Merid-

iana, 2018) as a pioneering care facility for people af-

fected by dementia, located in the city of Monza. As

a gated model village, “Il Paese Ritrovato” is a self-

contained community where patients are free to move

and interact with each other in wide open spaces,

and where structures have been speciﬁcally designed

to meet the patient’s needs, freeing them to the bur-

den of feeling slaves of the disease. The residence

consists of eight apartments with eight private rooms

each. Each apartment also has common rooms such

as the kitchen, the dining room and two living areas.

A square is located in the center of the village, where

residents can socialize visiting special buildings ded-

icated to them: a bar, a hairdresser, a cinema and a

church. La Meridiana has thought a new concept of

residence, where everything is designed to support the

lives of Alzheimer’s patients. Thanks to the collabo-

ration of many research institutions, the structure is

constantly looking for innovation with cutting-edge

technological systems, environmental and service de-

sign.

In order to preserve the independence of the pa-

tients without reducing their safety inside the struc-

ture, the number of caregivers to assist patients has

been redesigned to one every eight, which coincides

with one available caregiver for each apartment. Con-

sidering the dimension of the structure, the unconven-

tional freedom of the patients and the limited amount

of human resources, the activities of the caregivers

have to be simpliﬁed through the use of a system de-

veloped speciﬁcally for the supervision of the struc-

ture. The main need expressed by the structure is

to have a mobile application able to monitor the pa-

tients’ position, their meals and medicines. Further-

Figure 1: Push Notiﬁcations on the caregiver smartphone

rose if a patient cross virtual fences.

more, an efﬁcient localization system should keep the

caregivers updated about the state of the patients (see

Figure 1), allowing automatic generation of warn-

ings/alerts if patients approach/cross borders that are

not allowed. It is important to remember that the mon-

itored subjects are affected by a degenerative disease

that limits their cognitive abilities: the localization

system must not require any interaction by the users

and must be as non-invasive as possible.

Moreover, a constant monitoring may be useful to

obtain a complete analysis comprising medical data

resulting from clinical exams (e.g., electrical activity

of the heart, blood pressure, oxygen saturation), medi-

cal history and also an extremely accurate description

of the activities performed as a daily routine. This

type of monitoring is at the basis of behavioral drifts’

analysis (Veronese et al., 2018); changes in human be-

havior, such as changes in the attendance of environ-

ments or in the order of execution of daily activities

could be automatically recognized by the system and

reported to caregivers. A crucial aspect of this analy-

sis is the collection of data that must be carried out in

a non-invasive but constant way: a progression of the

disease may lie behind a change in the patient’s daily

routine.

3 LOCALIZATION SYSTEM

The main purpose of Assistive Technologies is to en-

sure the safety and well-being of patients, while trying

to lighten the workload of caregivers. The idea behind

this concept is to track the older person’s position con-

sidering routine activities and social interactions with

a minimal impact on his/her life.

The proposed system consists of an indoor local-

ization module based on iBeacon technology (IBea-

con, 2018), ﬁxed antennas, and a mobile application;

the entire system has been designed according to the

requirements of structures dedicated to the care of

people with dementia and Alzheimer’s disease.

Supporting Alzheimer’s Residential Care - A Novel Indoor Localization System

273

Figure 2: Architecture of the proposed indoor localization

system.

3.1 Architecture

The proposed localization system is speciﬁcally tar-

geted to operate with people affected by mild cog-

nitive disabilities such as dementia. Unfortunately,

most of the available localization systems require peo-

ple to use a device which is characterized by high

complexity. The following work moves all the com-

plexity to the system, requiring the users only to wear

a disposable bracelet. It is composed by three main

components (Figure 2):

• tags: iBeacon devices embedded into plastic

bracelets that will be worn every day by the pa-

tients of the structure;

• antennas: physical devices that receive the Blue-

tooth signals emitted by the iBeacons, and trans-

mit the information to the server;

• server: it computes the estimated position of pa-

tients given the data received from the antennas.

3.1.1 Tag

Each patient of the structure is equipped with a Blue-

tooth bracelet (iBeacon) and despite aware of being

localized while performing daily life activities, he/she

does not need to interact with the system.

Beacons are Bluetooth Low Energy (BLE) devices

that broadcast packets without the requirement of a

paired connection. The payload information meets

the standard speciﬁed by the iBeacon format and thus

contains - among others - a Universal Unique Iden-

tiﬁer (UUID), a Major and a Minor values to iden-

tify a speciﬁc device within a group, and the Broad-

casting power; a Beacon can be conﬁgured changing

its properties, as well as the transmission frequency.

Transmitting a packet with a period of 100ms (Apple

iBeacon preset), a Beacon has an expected battery life

up to 28 weeks; hence, a replace of the batteries every

year is necessary.

Antennas are responsible to intercept the beacon

messages broadcasted by the tags; the strength of the

received packet - Received Signal Strength Indicator

(RSSI) - is subsequently used by the system to esti-

mate the position of tags. However, even if the RSSI

value decays exponentially with the distance between

transmitter and receiver, this measure is strongly af-

fected by noise. Human body behave as an attenua-

tor of Bluetooth signals, affecting signal strength and

lowering accuracy; therefore the position of the bea-

con with respect to the antennas is fundamental for

the preservation of the signal. For this reason, partic-

ular attention was paid in the selection of the correct

position for the beacon, also considering the invasive-

ness on the individual. Finally, the choice fell on the

use of light-weight 3D-printable plastic bracelets with

an insert in which the beacon is positioned.

3.1.2 Antenna

The information advertised by iBeacons can be col-

lected only by other BLE enabled mobile devices such

as iPhones, Android devices, and also BLE enabled

Raspberry Pi, Odroid etc. (Aman et al., 2016). Fol-

lowing the structure needs, the fundamental require-

ment of the selected device is to have a BLE en-

able module in order to manage the collection of in-

coming information from the Beacons, and an Ether-

net adapter to connect with the server. Considering

the over mentioned requirements, the computational

power needed for the operations, and the cost, a Rasp-

berry Pi3 (Raspberry, 2018) has been chosen to act as

Antenna: this device is easily available on the mar-

ket and is certiﬁed, ensuring a good reliability. Thus,

a Raspberry Pi3 is positioned in the counter top of

each patient’s room and common spaces, receiving

and collecting Bluetooth information broadcasted by

the iBeacons.

The antennas continuously scan the Bluetooth

Beacons in the environment. Pre-processing is per-

formed on the RSSI data received from each tag, ap-

plying a median ﬁlter for noise reduction. The ﬁl-

tering window can be set by using a threshold deter-

mined during the installation and conﬁguration of the

antennas. The resulting data is periodically transmit-

ted to the server via an HTTP request containing the

following message:

<idA, UUID, major, minor, power, RSSI>

where idA corresponds to the antenna’s identiﬁer as-

signed by the server during the initial conﬁguration,

UUID, major, minor, and power describe the detected

beacon with its ﬁltered RSSI power strength.

The antennas are installed in the ceiling of rooms

and corridors, such that the entire surface of the struc-

ture is covered. The full coverage of the bluetooth

antennas is guaranteed both inside and outside the res-

idence. In order to reduce system wiring, Raspberry

WINSYS 2018 - International Conference on Wireless Networks and Mobile Systems

274

Pi is alimented through Power over Ethernet (POE)

allowing the appliance to be powered using the same

cable that connects it to the Ethernet data network.

More in details, each apartment is equipped with sev-

eral Raspberry Pi3 placed in the apartments both in

the private rooms, in the corridors and in the common

areas such as the kitchen or the living room. Regard-

ing the outdoors, each point of interest such as the

cinema, the bar or more simply the open-air areas are

mapped with devices.

During the installation of the system, antennas can

be conﬁgured in room or corridor or external mode.

The difference consists in the radius parameter, which

determines the sensitivity of the antenna. Increasing

the sensitivity of the antennas positioned in the cor-

ridor, for example, decreases the probability that the

patient is mistakenly positioned in the room if its ac-

tual position is in the corridor - as it was required by

the residence for security reasons.

3.1.3 Server

The server is responsible for receiving messages and

correctly saving them to the database. In order to

limit the possibility of wall-crossing in the localiza-

tion (e.g., the detection of a patient at the instant t

in his/her room, and at the instant t + 1 in the din-

ing room, without having been detected him/her in

the corridor that connects the two environments), the

localization algorithm has been extended to constrain

the admissible transitions between adjacent antennas.

To meet this need, the architecture has been modeled

as a graph whose nodes represent the antennas and the

arcs the walkable paths (see Figure 3). Two antennas

are therefore said to be adjacent if a user can freely

transit from one to the other.

Every second, the server computes the position of

all the tags. Considering a generic tag TAG

at time

instant t, the list of messages L

(t) received from an-

tennas within the last second is retrieved from the

database. Those elements with an RSSI value that

is lower than a threshold T

in are automatically re-

moved from L

(t). Then, the element with the higher

RSSI value, if it exists, is proposed as candidate posi-

tion CP

(t) to locate the tag. The actual position of

the tag FP

(t) is selected by evaluating CP

(t) and

(t − 1). This evaluation considers the probabil-

ity of each transition favoring (but not limiting) the

transitions between adjacent antennas. Algorithm 1

shows how to compute the position FP

(t) given the

candidate position CP

(t) of tag TAG

at time instant

t, its previous position FP

(t − 1) with its likelihood

P(FP

(t − 1)). The algorithm requires few threshold

values that are pre-initialized but can be tuned from

the server web interface:

• P

entry

: is the initial likelihood for a new localiza-

tion;

• P

delta

: is the value of likelihood that is

added/removed from the total whenever a

right/wrong candidate is proposed;

• P

min

: is the minimum value of likelihood that is

accepted. Under this level, jumping between non

adjacent antennas is allowed.

Algorithm 1: Localization algorithm.

1: if FP

(t − 1)notset then

2: //TAG

has not been localized before

3: FP

(t) ← CP

(t)

4: P(FP

(t)) ← P

entry

5: else

6: if CP

(t) = FP

(t − 1) then

7: //TAG

didn’t move from the last antenna

8: FP

(t) ← CP

(t)

9: P(FP

(t)) ← P(FP

(t − 1)) + P

delta

10: if CP

(t) 6= FP

(t − 1) then

11: if ad jacent(CP

(t), FP

(t − 1)) then

12: //TAG

moved to an adjacent antenna

13: FP

(t) ← CP

(t)

14: P(FP

(t)) ← P

entry

15: else

16: //TAG

is constrained

17: if P(FP

(t − 1)) > P

min

then

18: //TAG

cannot jump

19: FP

(t) ← FP

(t − 1)

20: P(FP

(t)) ← P(FP

(t − 1)) −

delta

21: else

22: //TAG

can jump

23: FP

(t) ← CP

(t)

24: P(FP

(t)) ← P(FP

(t − 1)) +

delta

3.2 Mobile Application

The mobile application is a primary support for the

caregivers. At the beginning of each work shift, the

caregiver logs in the application, selects his/her role

and immediately sees the position of the patients as-

signed to him. The main features of the application

are listed below:

• Notiﬁcation: caregivers receive a notiﬁcation if a

patient is approaching or crossing a physical or

virtual boundary for the patient (e.g., the exit of

the facility, the stairs). With this notiﬁcation mes-

sage, the caregiver can assist immediately and, if

necessary, can also request the intervention of an-

other colleague in support (see Figure 4).

Supporting Alzheimer’s Residential Care - A Novel Indoor Localization System

275

Figure 3: The map is modeled as a graph where nodes (rep-

resented as squares) stand for Antennas, and arcs (repre-

sented as dashed lines) identify the walkable paths. The

black spot indicates the position of a patient that is always

localized over a node.

Figure 4: The mobile application raises push notiﬁcations

to the caregiver whenever a patient violates a virtual fence

or another caregiver ask for help.

• Current position: the position of the patients is

visible in real time by consulting the “Map” sec-

tion. The caregiver can, at any time, view the po-

sition of the patients assigned to his shift (see Fig-

ure 5).

Figure 5: The mobile application allows caregivers to local-

ize the patients in the structure.

• Meals: selecting a patient, the caregiver can visu-

alize and annotate his/her personal data including

schedules, preferences, intolerances, etc.

• Medicines: each patient has a personal sec-

tion linked to medicines with the name of the

medicines to be administered, the schedule, and

the dosage. The caregiver marks in this section

the correct intake of medicines prescribed for the

patients.

The Mobile Application has been developed for

Android OS using Flutter framework (Google, 2018).

Caregivers are equipped with an Android smartphone

provided with the Mobile Application already in-

stalled.

3.3 Data

The gated village “Il Paese Ritrovato” has been inau-

gurated in February, 2018 and it will host 64 guests by

June, 2018. All the guests will be wearing the Blue-

tooth bracelet to allow their localization, while the

caregivers will be provided with a smart card to allow

also their localization inside the structure. Moreover,

the structure is technologically advanced: smart light

bulbs to simulate the sunrise, smart wardrobe to help

patients in dressing, smart beds to help people ﬁnding

WINSYS 2018 - International Conference on Wireless Networks and Mobile Systems

276

the way to toilet during the night, electric shutters and

several environmental sensors.

Once the village will be operative, it will become

a source of data from which to draw for further anal-

ysis such as daily habits and interactions, normal and

abnormal behaviors, etc.

Attendance of Environments. Using localization

data it is possible to determine how patients relate

to the environment; this is particularly important in

a village like “Il Paese Ritrovato” that is constantly

investing in the research of new strategies to to make

patients live better. These data constitute an important

source of knowledge to evaluate patients’ response to

the introduction of new experimental techniques.

Interaction. Alzheimer’s disease is characterized

by difﬁculties in language, in performing actions, in

perception (agnosia), and in the execution of complex

movements (apraxia); linguistic problems are mainly

characterized by an impoverishment in the vocabulary

and a decrease in ﬂuency, leading to a general deple-

tion of oral and written language. Caregivers usually

annotate relevant observed behaviors in order to track

the progression of the disease and to share these in-

formation with their colleague. The use of the data

collected in the structure allows continuous monitor-

ing of patient’s interaction, providing caregivers with

a basis of knowledge on which to base their observa-

tions.

Behavior. Many researchers are addressing the

problem of Activity Recognition and Behavioral

Changes leveraging Machine Learning Techniques on

data streams; their aim is to provide support in the

early diagnosis of chronic diseases and/or anomalies

(e.g., falls, strokes). The over mentioned dataset al-

lows the creation of behavioral path for the individual

deﬁned as the sequentiality of his/her activities during

the day with their duration and location. Alzheimer’s

is a form of degenerative dementia that becomes pro-

gressively disabling for the individual, therefore the

timing of the evolution of these events are not known.

The creation of this behavioral path allows the iden-

tiﬁcation of a growing discrepancy in the execution

of simple activities (e.g., having breakfast) and po-

tentially the delineation a different path of degener-

ation for each patient. In the intermediate phase of

the disease patients slowly are no longer able to per-

form daily activities, which entail a variation in the

daily routine of the patient. The caregiver needs to

know any variation in the schedule of activity of the

individual, in order to prevent possible abnormal sit-

uation.

4 CONCLUSIONS

A novel indoor localization system has been designed

to answer the needs of residential care for people af-

fected by Alzheimer’s disease. The application con-

text is “Il Paese Ritrovato”, a health-care facility

which is researching for innovative treatments to sup-

port the wellbeing of patients. Upon arrival in the

structure, patients are equipped with a bracelet con-

taining an iBeacon thanks to which they are moni-

tored during their stay. Data broadcasted by the iBea-

cons are collected through the use of Raspberry de-

vices acting as receiving antennas and analyzed with a

Web Server. The system evaluates the RSSI of the re-

ceived signal and corrects the computed position with

a probabilistic approach to avoid wall-crossing. In the

traditional approach, the beacons are positioned per-

manently while the BLE enabled device is in motion.

The system implemented involves a reverse approach:

the beacons become the moving devices, to be lo-

cated thanks to the antennas (BLE enabled devices)

arranged in the environments.

The ﬁrst contribution of the paper is the design

of an indoor localization system which is accepted

and used on people affected by Alzheimer disease.

Thus, caregivers have the possibility to monitor their

patients using a mobile application which is able to

show their position on the map and notify whenever a

patient is violating a virtual fence.

The second contribution of this work is the cre-

ation of a new dataset referring to the life of elderly af-

fected by Alzheimer’s disease in a controlled environ-

ment. This dataset will contain data referring to the

localization of people, their interaction with IoT de-

vices, their medicines and meals consumptions, their

activities and expert annotations. All the collected

data will be extremely important for analysis in the

ﬁeld of Behavioral Drift with the aim of identifying

- for example - what triggers an acceleration in the

progression of the disease.

This system is currently in an experimental state

at the structure built by the La Meridiana cooperative

which will open in June 2018.

ACKNOWLEDGEMENTS

Authors would like to thanks La Meridiana coopera-

tive and Fabrizio Danese who contributed to this work

with his Master Thesis in Computer Science at Po-

litecnico di Milano University.

Supporting Alzheimer’s Residential Care - A Novel Indoor Localization System

277

REFERENCES

Aman, M. S., Jiang, H., Quint, C., Yelamarthi, K., and Ab-

delgawad, A. (2016). Reliability evaluation of ibea-

con for micro-localization. In 2016 IEEE 7th Annual

Ubiquitous Computing, Electronics Mobile Communi-

cation Conference (UEMCON), pages 1–5.

Farid, Z., Nordin, R., and Ismail, M. (2013). Recent ad-

vances in wireless indoor localization techniques and

system. Journal of Computer Networks and Commu-

nications, 2013.

Google (2018). Flutter framework, https://ﬂutter.io.

IBeacon (2018). Apple informazioni sui servizi di localiz-

zazione, https://support.apple.com/en-gb/HT202880.

Il Paese Ritrovato Project (2018). Cooperativa la meridiana,

http://www.cooplameridiana.it/il-paese-ritrovato/.

Jacq, D., Chatonnay, P., Bloch, C., Canalda, P., and Spies,

F. (2017). Towards zero-conﬁguration for wi-ﬁ indoor

positioning system. In Indoor Positioning and Indoor

Navigation (IPIN), 2017 International Conference on,

pages 1–8. IEEE.

La Meridiana (2018). Cooperativa Sociale La Meridiana,

http://www.cooplameridiana.it.

Raspberry (2018). RaspberryPi 3 - Model B, https://www.

raspberrypi.org/products/raspberry-pi-3-model-b/.

Tariq, Z. B., Cheema, D. M., Kamran, M. Z., and Naqvi,

I. H. (2017). Non-gps positioning systems: A survey.

ACM Computing Surveys (CSUR), 50(4):57.

Ter

an, M., Aranda, J., Carrillo, H., Mendez, D., and Parra,

C. (2017). Iot-based system for indoor location us-

ing bluetooth low energy. In Communications and

Computing (COLCOM), 2017 IEEE Colombian Con-

ference on, pages 1–6. IEEE.

Umphred, D. (2012). Neurological rehabilitation. Elsevier

Health, London, 6th edition.

United Nations, D. o. E. and Affairs, S. (2017). World pop-

ulation ageing 2017. Technical report.

Veronese, F., Masciadri, A., Comai, S., Matteucci, M., and

Salice, F. (2018). Behavior drift detection based on

anomalies identiﬁcation in home living quantitative

indicators. Technologies, 6(1):16.

World Health Organization (2015). Dementia Fact sheet

N.362.

Zafari, F., Papapanagiotou, I., Devetsikiotis, M., and

Hacker, T. (2017). An ibeacon based proxim-

ity and indoor localization system. arXiv preprint

arXiv:1703.07876.

WINSYS 2018 - International Conference on Wireless Networks and Mobile Systems

278