Use of Mobile Phones for Exposure Assessment
Transcription
Use of Mobile Phones for Exposure Assessment
Department of Epidemiology and Public Health Use of Mobile Phones for Exposure Assessment Martin Röösli, PhD Swiss TPH Spring Symposium 8 May 2012 Many thanks to Oliver Lauer (ETHZ), Nicolas Maire and Harish Phuleria, who contributed slides to this talk. Content What is a mobile phone? HERMES: Use of mobile phones for radiofrequency electromagnetic field measurements Sensorscope/Climaps project NoiseTube: collecting noise data with mobile phones Basel, 08. 05. 2012 Martin Röösli 2 What does your smart phone? from Lane et al., IEEE Com Mag, 2010 Basel, 08. 05. 2012 Martin Röösli 3 And what may your mobile phone do in the future barometer, temperature, humidity sensors (Intel/University of Washington Mobile Sensing Platform) air quality and pollution (Honicky et al., 2008) blood pressure monitoring using the earphone (Poh et al., 2009) commercially interesting: combine low-level censor data with context and activity data. E.g. Sense Networks a U.S. start-up company which uses millions of GPS estimates from mobile phones to predict which subpopulation is interested in a specific type of event (Lane et al., 2010). Basel, 08. 05. 2012 Martin Röösli 4 Mobile phone sensing on different scales from Lane et al., IEEE Com Mag, 2010 Basel, 08. 05. 2012 Martin Röösli 5 Other conceptual issues passive vs active recording smart phone vs. smart sensors Basel, 08. 05. 2012 Martin Röösli 6 Benefits of using mobile phones for exposure assessment Two types of information is obtained: exposure and behaviour. Many data can be collected Continuous measurements -> directly sent to the server Apps can be developed to facilitate data collection (e.g. reminder, questionnaires) Appstores and similar can be used for apps distribution E.g. collect noise data, apply learning algorithms to identity classes of behaviour: Basel, 08. 05. 2012 Martin Röösli 7 HERMES: A Swiss TPH project Health Effects Related to Mobile phone usE in adolescentS Is radiofrequency electromagnetic fields (RF-EMF) from using mobile phones related to behaviour, symptoms and cognitive function Funding: SNF Cohort study in Central Switzerland: 1. baseline investigation in 2012 in 8th grade students 2. follow-up investigation in 2013 in the same students (9th grade) Use of mobile phone operator recorded data Modelling of environmental RF-EMF Collecting personal RF-EMF from 100 study participants Basel, 08. 05. 2012 Martin Röösli 8 Measurement Concept: combine exposimeter with smart phone Exposimeter 1. RF-EMF measurements Exposimeter Smartphone Smartphone 1. Position GPS Bluetooth Activity diary health 2. (Use of mobile phone) GPS Diary In te rn et Server Basel, 08. 05. 2012 Martin Röösli 9 9 Calibration 1. 2JMAS01 antenna 2. 824 MHz- 5.9 GHz Frequency Band Min. Field [V/m] Max. Field [V/m] GSM900TX 0.0008 1.4778 GSM900RX 0.0031 3.9388 GSM1800TX 0.0073 6.6912 GSM1800RX 0.0243 >12.5601 DECT 0.0037 >12.4465 UMTSTX 0.0053 6.7472 UMTSRX 0.0052 9.1096 ISM2.4 0.0117 >10.0520 Basel, 08. 05. 2012 Martin Röösli 10 10 Smartphone Requirements 1. Bluetooth Interface 2. GPS module 3. Open source OS Concept 1. Data Logging / Display Data storage Tracking GPS location Tracking transmission power (NF) Data distribution using 2G/3G backbone network 2. Active diary/questionnaire Basel, 08. 05. 2012 Martin Röösli 11 11 Test case-study Measurement location 1. Zürich Measurement settings 1. sampling period: 1.6 s 2. measurement duration: 15 min 3. linear antenna polarization Basel, 08. 05. 2012 Martin Röösli 12 12 Measurement Results GSM 1800rx Basel, 08. 05. 2012 Martin Röösli 13 13 Sensorscope/Climaps project: Real-time environmental monitoring • Partnership EPFL (lead) & SwissTPH (Harish Phuleria) • Real-time monitoring of various air pollutants and weather parameters in Lausanne •http://www.climaps.com/ Basel, 08. 05. 2012 Martin Röösli 14 Sensorscope/Climaps project: Sensor evaluation • Precision high in most sensors • Both sensitivity as well as specificity are still problems for some sensors, and accuracy still not enough to be used for community AQ monitoring • Most AQ sensors are still under development and not available commercially Basel, 08. 05. 2012 Martin Röösli 15 NoiseTube: citizen scientists for distributed sensing Participatory noise pollution monitoring using mobile phones. Maisonneuve et al., Information Polity, 15(1-2):51-71, Aug 2010. Basel, 08. 05. 2012 Martin Röösli 16 Noisetube architecture NoiseTube App (Android) NoiseTube server Upload data Store locally Android file system Basel, 08. 05. 2012 Martin Röösli 17 Noisetube limitations Experimental state after re-implementation for Android Calibration issues Basel Summer (filtered SLM data) 75 Smartphone, LAeq (dBA) 70 65 y = 0.71x + 5.1 60 2 R = 0.954 (w/o outlier) 55 y = 0.64x + 9.6 2 R = 0.867 (all data) 50 45 40 35 35 40 45 50 55 60 65 70 75 Sound Level Meter, LAeq (dBA) Basel, 08. 05. 2012 Martin Röösli 18 Challenges with mobile phone exposure data Technical issues Selection of participants: random sample difficult to achieve If interested in a specific population: not everybody will have a suitable smart phone Operating system of Iphone is not open Data quality Amount of data Post processing Basel, 08. 05. 2012 Martin Röösli 19 http://villevivante.ch/ Basel, 08. 05. 2012 Martin Röösli 20