2012 ISBNPA Conference, Austin, TX - Novel Technologies Workshop

Concepts...

• Residential trap / Local trap (Cummins, Chaix)
• Principles of space-time geography (Hagerstrand)
• Frequency
• Spatial polygamy (Matthews)
• Activity space
• Network of usual places (Flams)
• Trips/routes

Lack of interactive mapping tool to collect detailed information on activity locations, spatial behavior and perceived spaces

Meanwhile, increasing 'map-literacy'

Part I: An interactive map-based questionnaire...

One of the aims of a contextual expology is to define the exposure area for measurement

Development of a

GPS Protocol Management Tool (GPMT)

Issues

1. Burden of multiple sensors

2. Battery life (GPS)

3. Ethical

4. Complexity of device manipulation (e.g. cellphone)

5. Real-time data transmission

6. Feedback

Output information can be used for exposure assessment and as a basis for further data collection...

Thank you!

Assessing activity places

in neighborhood and health research:

Contributions of surveys of regular destinations

and GPS tracking

Q & A

2012 ISBNPA Conference Workshop

Emerging Technologies for Assessing Environmental Influences on Physical Activity:

A Demonstration of Methods and Applications

SPHERELAB.org

Basile Chaix, INSERM

Yan Kestens, Montreal University

Principle

Application:

VERITAS RECORD

Introduction

1. Series of questions about 'where'

2. Answers provided through interactive map

French Cohort in Paris Region

N=7,290; 30-79 yo; 1915 neigbourhoods

ENVIRONMENT

HEALTH

Use of VERITAS to document perceived neighbourhood limits and network of regular destinations including:

Residence, work, food stores, press/tabaco store, bank, hairdresser, mass transit, sports, cultural, community activities, leisure, friend/family visit

MOBILITY

HOME

3. Can document points, lines, polygons

Active living

Mobility data

Spatio-behavioural

indicators

Environmental

exposure

• Residential history
• Workplace history
• Activity space anchor points
• Perceived spaces
• Barriers
• Routes

GPS trackers

(Life-space) questionnaires

Mobility surveys

Time use surveys

Preliminary results

Standard deviation ellipse

A = 3.4 km²

Major to minor axis ratio: 2.2

Convex hull

A = 6.3 km²

P = 17.0 km

A novel interactive map-based

questionnaire for spatial data

collection in health studies:

VERITAS

Ongoing, 40,200 reported activity places

for 2,790 participants

Characterization of spatial behavior

Mobility-based measures of exposure may increase stratification according to:

• true environmental exposures
• potential confounders also influencing behavior

Shortest path network area

(estimated routes)

Buffers around activity places

Contextual expology:

-> SELECTIVE DAILY MOBILITY BIAS

Assessment of environmental exposures

Major risk with activity place surveys or GPS tracking

• a step forward for exposure assessment
• a step backward for causality assessment?

• (sub)discipline interested in the spatio-temporal configuration of exposure

= the "where" and "when" of exposure

Filtering based on activities:

related / unrelated to the outcome

Both with destination surveys and GPS tracking, it is critical to collect information on the activities practiced at the different places for filtering of exposure measures.

Part II: The tribulations of GPS tracking...

Sensors

GSM tower

COMPONENT 1: Improvement of wearable device, real-time transmission

End users

Acquisition server

COMPONENT 3: Web applications, prompted recall, counseling

COMPONENT 2: Data processing, GPS algorithms, Accelerometer algorithms

Outputs /

Applications

Algorithms

GIS

Web server

Three main challenges:

2

1

3

Recruitment and management

Data treatment and enrichment

Sensors and tools for data collection

Challenge...

From raw GPS files, extract:

• Activity locations
• Trips/Routes
• Transportation modes

Issues

SPHERELAB GPS ArcToolbox

1. Complex and technical instructions to deliver

2. Need to collect operational information

Development of a novel multisensor device

The management app' allows for:

A kernel-based activity location extraction algorithm

• the standardized presentation of the study
• data collection

Extraction of trips between activity locations

Compilation of frequency, length of stays and trips

Transportation mode detection

GPS – SiRFStar IV

Tri-axial accelerometer

Possible integration of multiple datastreams

• Ultra High Sensitivity :147dBm (Cold Start Acquisition), 163 dBm (Tracking)
• Low power consumption: 56/68 mW @ 1.8 V
• Accuracy: Position: 2.5 m
• NMEA protocol, switchable to OSP binary

• 12 bits sampling
• +/- 2g, 4g, 8g et 16g range

Memory

3. Management of participants at

different stages of the study

4. Management of a pool of devices

ANT+ module

GPRS

GPRS module

• Second generation single chip ANT solution
• Reception and transmission on 8 channels
• (2,4GHz) 0 dBm max

• GSM network connection
• 4 bands : (850, 900, 1800, 1900) MHz
• SIM card configuration

Long lasting battery

• 3200 mAh

Participant and device managers

Common public health approach to GPS tracking

Clothing store

- From 4:31 to à 4:36 pm

- Transport: walking

- Frequency: 0

Mobility Web Mapping app'

(for the RECORD GPS Study)

• No assessment of activity places and activities
• Accelerometry data used without assessment of transportation modes

Consumer panel

- From 3:01 to 4:20 pm

- Transport: bus 47

- Frequency: once per month

• Survey of activities, visit frequencies, and transportation modes over 7 days
• Delete fictive activity places
• Report unobserved or unidentified activity places
• Identification activity places also in VERITAS
• Aim: use of the app in near-real time mode with over-the-air GPS data transmission

The transportation approach to GPS tracking

Clothing store (Phildar)

- From 4:56 to 5:22 pm

- Transport: metro

- Frequency: twice per year

• Identification of activity places
• Assessment / confirmation of activities and transportation modes

Pharmacy (Place Monge)

- From 5:33 to 6:11 pm

- Transport: walking

- Frequency: twice per year

Supermarket (Monoprix)

- From 6:20 to 6:45 pm

- Transport: walking

- Frequency: once per month

Import the transportation approach in public health

GSM Network

5. Organize actions of follow-up

(phone calls, SMS, face-to-face meetings)

HR

monitor

Residence

- From 0:00 am to 2:10 pm

- Transport: walking

- From 7:21 à 11:59 pm

- Transport : bus 27

• Determinants of active transportation
• Address selective daily mobility biases

Blood

pressure

Central Unit

GPS STUDY (ONGOING)

FINAL OUTPUT

48

Accele-

rometer

Glucose

monitor

GPS

GPRS

48%

Accelerometer

1. Timetable with

activity places

and trips

2. Corrected

geographic file

of the trajectory

Task schedule manager

36

Memory

ANT Module

~1 day

# of participants who carried a device:

Acceptance/completion rate (May 21 2012):

# of prompted recall surveys completed:

Processing time for survey per participant:

% of visits identified by the algorithm confirmed through survey:

# of visits added during the survey (missed by the GPS / the algorithm):

After processing, for each participant:

• # of visits to activity places
• # of corresponding unique places

96%

Galvanic skin

response

Other

8.5

...integrating:

• cleaned GPS data (activity places, trips)
• corrections and additions from the survey
• attribute data (activity type, etc.)

Acquisition server

40.1

19.3

Concluding remarks

Improved consideration of place in health research through use of wearable sensors opens great possibilities while posing new challenges

Data collection

Novel tools to address each phase of the 'Sensor data collection triptic':

Data transformation

Data usage

Contact:

chaix@u707.jussieu.fr

yan.kestens@umontreal.ca

SPHERELAB.org

record-study.org

spherelab.org