A Graphical User Interface for an Automatic Rest-activity Cycle

Detection and Dichotomy Index Computation

Racha Soubra

, Aly Chkeir

, Mohammad O. Diab

, Majd Abdallah

and Jacques Duchene

University of Technology of Troyes, France

Rafik Hariri University, Meshref, Lebanon

Keywords: Circadian Rhythm, Rest-activity Cycle, Dichotomy Index, Graphical User Interface (GUI).

Abstract: The success of chemotherapy treatment is achieved based on “Chronotherapy”: the concept of administering

the correct drug at a precise time based on the circadian rhythm study. This paper aims to detect the

rest/activity cycle and automatically calculate the dichotomy index (I<O), as both parameters have been

proved to be reliable indices of the circadian rhythm. First, the DARC “Détection Automatique du Rythme

Circadien” algorithm is used to segment the rest-activity phases automatically. Then, a Graphical User

Interface (GUI) is used to calculate easily the I<O across several days of records and smooth the analysis. The

outcome of this study provides an easy-to-use GUI that minimizes patients’ intervention, facilitates user

involvement, and reduces the time required for analysis.

1 INTRODUCTION

The endogenicity of the circadian rhythm persists in

constant environment in microorganisms, plants and

in all kinds of animal species including man (Levi

2006). These endogenous rhythms manage daily

events such as sleep, activity, hormonal secretion,

cellular proliferation and metabolism (Levi 2006).

Scientists have shown that the rest-activity cycle is a

reliable marker of the circadian system function.

locomotor activity dependably reflects the circadian

clock function in several animal species like rodents

and man. It is evident that researchers focus their

work on examining the circadian rest-activity rhythm

disturbances among patients with several diseases

mainly cancer. For instance, Mormont, M. and Levi,

F. documented the link between the rest-activity

rhythm and the welfare of cancer patient (Mormont

and Levi 2003). In their study, it was clear that the

circadian rhythm tends to be lost in the rapidly

growing or advanced stage of tumors. Similarly, in

Levin, R. et al research study, the rest-activity

circadian function differs significantly between

patients with advanced non-small-cell lung cancer

and control subjects (Levin, Daehler et al. 2005).

Outcomes have shown that patients suffer poor sleep

quality and high levels of fatigue. In addition,

Mormont, M.et al found that patients with poor

circadian rhythmicity had a 5-fold higher risk of

dying within 2 years than the patients with a better

circadian rhythm (Mormont, Waterhouse et al. 2000).

Consequently, the rest-activity rhythm is a

determinant of quality of life. Its level of disturbance

can be set as a reference for anti-cancer efficacy and

tolerability. Therefore, the rest-activity rhythm can

provide additional prognostic information regarding

patients’ response to treatment and maximum

survival potential since it reflects tumor burden and

patient general conditions (Mormont, Waterhouse et

al. 2000, Mormont and Levi 2003, Rich, Innominato

et al. 2005).

Moreover, the circadian or rest-activity disruption

in patient with cancer is the result of chronic illness

development and worsening of pre-existing

conditions (Ortiz-Tudela, Martinez-Nicolas et al.

2010). However, these disruptions can resulted from

a “wrongly timed or excessively dosed

chemotherapy”. Scrully, C. et al proved that the

circadian timing of anti-cancer medications improves

treatment tolerability up to fivefold and doubles the

efficacy in experimental and clinical studies (Scully,

Karaboué et al. 2011). Therefore, the time-qualified

chemotherapy, known as by “Chronotherapy”, is an

essential element for cancer treatment. It guarantees

improved cancer-treatment and optimizing the

development of new anticancer or supportive agents

(Mormont, Waterhouse et al. 2000, Levi 2001).

340

Soubra, R., Chkeir, A., Diab, M., Abdallah, M. and Duchene, J.

A Graphical User Interface for an Automatic Rest-activity Cycle Detection and Dichotomy Index Computation.

DOI: 10.5220/0007395803400344

In Proceedings of the 12th International Joint Conference on Biomedical Engineering Systems and Technologies (BIOSTEC 2019), pages 340-344

ISBN: 978-989-758-353-7

Personalized cancer chronotherapeutics encourage

the cell division cycle and the pharmacology

pathways to improve patients’ quality of life and

survival (Lévi, Filipski et al. 2007). Thus, the

circadian rhythm needs to be explored on large scale,

and circadian biomarkers should be calculated to

estimate the incidence of cancer-associated circadian-

system alterations.

The most effective parameter that can correlates

with the quality of life is the dichotomy index (I<O)

(Mormont, Waterhouse et al. 2000, Innominato,

Focan et al. 2009, Natale, Innominato et al. 2015).

This latter represents the percentage of the activity

counts measured when the patient is in bed that are

inferior to the median of the activity counts measured

when the patient is out of bed. This index can

theoretically vary between 0 and 100%, where high

I<O reflects a marked rest/activity rhythm.

In order to record rest-activity cycle, the majority

of recent studies used wrist actigraphy; a wearable

device used to measure the activity motors. On the

other hand, various techniques were used to calculate

the I<O. For instance, in Mormont, M. et al study, the

calculation was done manually where each patient

had kept a diary for times of rising and retiring during

the diagnosis (Mormont, Waterhouse et al. 2000).

Scrully, C. et al and Ortiz-Tudela, E. et al have used

square and mean waveform techniques respectively

which resulted as a poor biomarkers (Innominato,

Focan et al. 2009, Ortiz-Tudela, Martinez-Nicolas et

al. 2010). In Ortiz-Tudela, E. et al study, patients

were requested to give an informed consent and to

complete a sleep and feeding log during the days of

recording (Ortiz‐Tudela, Iurisci et al. 2014). Finally,

some patients were demanded to push an event-

marker button on the wearable device to mark

occurrences of time in and out of bed such as Natale,

V. et al research (Natale, Innominato et al. 2015).

In this study, we aim to detect the rest- activity

cycle automatically and calculate the I<O while

minimizing the intervention of patients and

smoothing the interference of physicians. After data

acquisition, I<O was calculated automatically based

on DARC algorithm. Then, a graphical user interface

(GUI) was performed to detect and calculate

automatically rest-activity cycle and I<O.

2 METHODOLOGY

2.1 Database

Our study is based on 9 control subjects (5 females

and 4 males) aged 40 ± 10.6 years. After receiving a

detailed description of the objectives and

requirements of the study, the participants wore the

infrared sensor “Movisens GmbH - move II”. The

move II sensor consists of a tri-axial acceleration

sensor (adxl345, Analog Devices; range: ±8 g;

sampling rate: 64 Hz; resolution: 12 bit) and a

temperature sensor (MLX90615 high resolution

16bit ADC; resolution of 0.02°C). This sensor was

patched onto the participants’ upper right anterior

thoracic areas by means of a hypoallergenic patch

for a minimum of three consecutive days. It only

weighs 32 g, and measures 5.0 x 3.6 x1.7 cm3. The

recorded data is saved on a memory chip inside the

sensor and transferred to a server via the General

Packet Radio Service (GPRS). Three signals were

available:

 Zero Crossing Mode (ZCM) signal:

representing the human activity in function of

time, with 1 record per minute

 Body Position: representing the human body

slope with respect to the vertical x-axis, with 1

record per minute

 Body Temperature: representing the human

body temperature, with 1 record every 5

minutes

2.2 Rest/Activity Cycle Detection

In this study, the automatic detection of rest/activity

cycle is achieved based on the “Détection

Automatique du Rythme Circadien” (DARC)

algorithm (Chkeir et al. 2017). Six phases summarize

our work, and for confidential reasons, it will be

discussed generally in a brief way.

First, as we have one record of body temperature

each 5 minutes and one record per minute for each of

body position and ZCM signals, the Polynomial

Cubic Spline method is used to interpolate the

temperature signal, so we get an equal number of

records between signals. The interpolated

temperature signal intervenes as a reference to check

if the sensor is worn or not. The algorithm will

directly eliminate the body position and ZCM records

when sensor is not worn. In case the sensor is worn

upside down, the algorithm will correct the Body

Position signal: when X is greater than 90, the value

will be replaced by 180-X.

Subsequently, all outlier points that could

appear in the signals will be eliminated based on

the median filter techniques. After that, the method

cited in the DARC Brevet automatically operates

A Graphical User Interface for an Automatic Rest-activity Cycle Detection and Dichotomy Index Computation

341

to separate between rest/activity phases for both

signals.

After detecting the indices for each of the rest

and activity phases, the ZCM and body position

signals are converted into binary signals, where 1

represents the activity cycle and 0 represents the rest

cycle. The “AND” logical operator is applied to

combine the two binary signals, so new indices can

state accurately the starting and ending point of each

phase.

2.3 Graphical User Interface

Graphical user interface, abbreviated by GUI, is a

software interface that works at the point of contact

between a computer and its user. It allows the user to

interact with computers through graphical elements,

such as dialog boxes, pointing devices, push-buttons,

menus, and scroll bars. By selecting one of these

visual elements, either by mouse, pen or other

selection from menu, the user can manipulate what is

on screen and control commands to run a program

without writing a text characters.

The main reason from creating a GUI is to make

things simple for the program end-users. In order to

accomplish a task, users do not have to edit the

command line interface, create a script command, or

even understand the details of how tasks are

performed. Applications that are based on GUI are

simple to learn and run.

Figure 1: Segmentation of Rest/Activity cycle and

application of I<O formula.

Accordingly, in order to automatically calculate

the I<O and facilitate the intervention of users

(physicians, clinicians, researchers…), we aimed to

perform an easy-to-use GUI. The goals of this GUI

were to: (i) Detect the rest/activity cycle either

automatically through the DARC algorithm or

manually by user involvement and (ii) Calculate

automatically the I<O while applying the following

formula:

I<O = (1 – NB_L/NB_C) × 100 (1)

Where NB_L represents the number of activities

recorded during the day and NB_C

represents the

number of activities recorded during the rest period

that are greater than the median of NB_L.

3 GUI RESULTS

In this section, we demonstrate the GUI usage shown

in Figure 2.

First, users import the Patient Excel File (xls,

xlsm, xlm…).

Then, they could choose between ‘Manually’ and

a ‘Graph’ radio-button to select the starting time of

the day. If the ‘Manually’ option is chosen, a Time

Panel will appear on the screen. Users select the

appropriate time and then press ‘Set’. If the ‘Graph’

option is chosen, the select cursor will be available

and ZCM signal will be plotted. So, users select a

starting point from the graph. Time will be displayed

on the screen with an option of Reset.

At this time, users will have the opportunity to edit

‘threshold’ and ‘filter’ values as convenient.

To segment rest-activity cycle, users can choose

an ‘automatic’ and/or ‘manual’ detection

methodology. If ‘Automatic Detection’ is chosen,

users can press ‘Activity/Position’ to plot signals

segmentation and ‘Average’ buttons to get I<O

results. Once the ‘Activity/Position’ button is

pressed, ZCM and position signals will be plotted in

graphs 1 & 2 respectively showing the rest/activity

phases. ‘Next’ and ‘Previous’ buttons are available to

show the plots across each day. If ‘Manual Detection’

is chosen, a ‘help window’ will be displayed asking

the users to select the intervals of sleep across each

day plotted separately in graph 1. If users select the

sleep intervals wrongly, a ‘warning window’ will be

displayed requesting the repetition of the manual

detection procedure. If both ‘automatic’ and ‘manual’

detection techniques are chosen, the ‘Auto/Manual

Activity’ button can be pressed to plot the difference

of segmentation between both techniques and for

each day.

HEALTHINF 2019 - 12th International Conference on Health Informatics

342

Figure 2: Graphical User Interface.

After the rest-activity detection, users can press

the ‘Average’ button to examine the I<O results of

each day coordinated in a table. Additionally, they

can select the number of days (X) to calculate the

I<O average of each X consecutive days.

To save the plotted figures and results, users

have the option of selecting a folder and the excel

filename. For each day, figures and I<O results will

be plotted on a separate sheet titled by the day

number. Moreover, the I<O table will be copied to

‘I<O Results’ sheet.

A ‘Restart’ button is available to repeat the

process. Once this button is clicked, a message box

will appear to ensure the restarting process.

Similarity for the ‘Exit’ button, once it is clicked, a

message box will appear to ensure the closing.

In order to validate our method, the student-test

and mean squared error (MSE) measurements were

evaluated. High correlation was found between I<O

calculated automatically and those calculated

manually, giving an R of 0.95 and MSE of 0.022.

4 DISCUSSION

The I<O, a reliable marker of rest/activity cycle, has

been used frequently in order to identify the optimal

time of the chemotherapy. In addition, it was

demonstrated that this parameter can give

physicians supplementary information for about the

patients’ quality of life. Therefore, an automatic

detection of the rest and activity phases can

minimize the patient’s intervention, reduce the time

required for I<O calculation and encourage more

related studies. Additionally, a GUI can facilitate

the users’ intervention and reduce the time of

analysis.

In the first part of our study, patients intervened

only by wearing the Movisens. Data was directly

saved on a memory chip and transferred to the

server via GPRS. The I<O of each subject was

calculated automatically. Then, a comparison of

I<O results was performed.

Next, a GUI was performed. Several options and

advantages encourage its usage. For instance, the

user has the option of segmenting rest-activity cycle

automatically and/or manually. Accordingly, a

comparative study can be performed between the

two techniques. Moreover, several parameters are

set as “variable data” such as I<O threshold,

temperature reference value, median filter and

others. Therefore, these factors can be changed

based on the user analysis and depending on each

subject. Then, segmented graphs, stating each of the

rest and activity phases, can be plotted for each

recorded day. Therefore, physicians can see the

relation between rest-activity cycle and I<O results

for both manual and automatic detections and save

outcomes on Excel File. Last but not least, a clear

demonstration is available with the GUI to smooth

A Graphical User Interface for an Automatic Rest-activity Cycle Detection and Dichotomy Index Computation

343

the user interface and explain other supplementary

options.

5 CONCLUSION

In this paper, we provide physicians and researchers

with a new technique to detect the rest-activity cycle

and calculate the dichotomy index automatically.

This work provides an easy-to-use graphical user

interface. It incorporates our listed purposes with

several analysis options that vary according to the

user requirements for an enhanced I<O study and an

improved chronotherapy. Physicians can detect the

optimal time for a chemotherapy treatment so that a

better quality of life and less disturbance of

circadian rhythms can be achieved.

REFERENCES

Chkeir A., Duchêne J., Hewson D., Lévi F., Beau J.,

Method for Determining Automatically the

Dichotomy Index I<O of an Individual. Pub. No.

WO2017/103203 22-6-2017

Innominato, P. F., C. Focan, T. Gorlia, T. Moreau, C.

Garufi, J. Waterhouse, S. Giacchetti, B. Coudert, S.

Iacobelli and D. Genet 2009. "Circadian rhythm in

rest and activity: a biological correlate of quality of

life and a predictor of survival in patients with

metastatic colorectal cancer." Cancer research 69(11):

4700-4707.

Levi, F. 2001. "Circadian chronotherapy for human

cancers." The lancet oncology 2(5): 307-315.

Levi, F. 2006. The circadian timing system, a coordinator

of life processes. implications for the rhythmic

delivery of cancer therapeutics. Engineering in

Medicine and Biology Society, 2006. EMBS'06. 28th

Annual International Conference of the IEEE, IEEE.

Lévi, F., E. Filipski, I. Iurisci, X. Li and P. Innominato

2007. Cross-talks between circadian timing system

and cell division cycle determine cancer biology and

therapeutics. Cold Spring Harbor symposia on

quantitative biology, Cold Spring Harbor Laboratory

Press.

Levin, R., M. Daehler, J. Grutsch, J. Quiton, C. Lis, C.

Peterson, D. Gupta, K. Watson, D. Layer and S. Huff-

Adams 2005. "Circadian function in patients with

advanced non-small-cell lung cancer." British journal

of cancer 93(11): 1202.

Mormont, M.-C., J. Waterhouse, P. Bleuzen, S.

Giacchetti, A. Jami, A. Bogdan, J. Lellouch, J.-L.

Misset, Y. Touitou and F. Lévi 2000. "Marked 24-h

rest/activity rhythms are associated with better quality

of life, better response, and longer survival in patients

with metastatic colorectal cancer and good

performance status." Clinical Cancer Research 6(8):

3038-3045.

Mormont, M. C. and F. Levi 2003. "Cancer

chronotherapy: principles, applications, and

perspectives." Cancer: Interdisciplinary International

Journal of the American Cancer Society 97(1): 155-

169.

Natale, V., P. F. Innominato, M. Boreggiani, L. Tonetti,

M. Filardi, A. Parganiha, M. Fabbri, M. Martoni and

F. Lévi 2015. "The difference between in bed and out

of bed activity as a behavioral marker of cancer

patients: A comparative actigraphic study."

Chronobiology international 32(7): 925-933.

Ortiz-Tudela, E., A. Martinez-Nicolas, M. Campos, M. Á.

Rol and J. A. Madrid 2010. "A new integrated

variable based on thermometry, actimetry and body

position (TAP) to evaluate circadian system status in

humans." PLoS computational biology 6(11):

e1000996.

OrtizTudela, E., I. Iurisci, J. Beau, A. Karaboue, T.

Moreau, M. A. Rol, J. A. Madrid, F. Lévi and P. F.

Innominato 2014. "The circadian restactivity

rhythm, a potential safety pharmacology endpoint of

cancer chemotherapy." International journal of cancer

134(11): 2717-2725.

Rich, T., P. F. Innominato, J. Boerner, M. C. Mormont, S.

Iacobelli, B. Baron, C. Jasmin and F. Lévi 2005.

"Elevated serum cytokines correlated with altered

behavior, serum cortisol rhythm, and dampened 24-

hour rest-activity patterns in patients with metastatic

colorectal cancer." Clinical cancer research 11(5):

1757-1764.

Scully, C. G., A. Karaboué, W.-M. Liu, J. Meyer, P. F.

Innominato, K. H. Chon, A. M. Gorbach and F. Lévi

2011. "Skin surface temperature rhythms as potential

circadian biomarkers for personalized

chronotherapeutics in cancer patients." Interface

Focus 1(1): 48-60.

HEALTHINF 2019 - 12th International Conference on Health Informatics

344