Indie-Pocket

COVID-19 contact-tracing via enhancement – “Indie-Pocket”

The “Indie-Pocket” project successfully participated in the LauzHack COVID-19 hackathon

Contact-tracing allows individuals to be alerted if they have encountered affected people and prevent the spread of SARS-CoV-2. The European consortium PEPP-PT is developing a distributed approach to do so while preserving the privacy of its users. Using Bluetooth technology, the distance with other people is only computed and stored in the local phone memory, therefore not transmitted to third-party companies or governments.

The accuracy of the contact tracing measure depends on how the user carries their phone: in the front or back pocket, in the hands or in a purse.

The accuracy of the contact tracing measure depends on how the user carries their phone: in the front or back pocket, in the hands or in a purse.

Indie-Pocket is able to detect where you carry your phone using only its integrated accelerometer sensor. Leveraging on machine learning and previously collected data from voluntary participants, the algorithm can tell with 95% accuracy in which pocket your phone is located. This can be used by contact-tracing systems such as the PEPP-PT, DP-3T or the Google-Apple joint initiative, to enhance the precision of their measurement.

The privacy preserving collection of health-/ medical data through intelligent systems is part of the application areas the EPFL Center for Intelligent Systems is working on.

Romain Baud

BIOROB-REHAssist

Doctoral assistant

Linus Gasser

C4DT

Computer Scientist

Julien Pache

BIOROB-REHAssist

ETS/HES Engineer

Tristan Vouga

BIOROB-REHAssist

Doctoral assistant