Date post: | 09-Jul-2015 |
Category: |
Health & Medicine |
Upload: | nikolaygrigoriev |
View: | 74 times |
Download: | 0 times |
From reactive to proactive May 2014
THE ACCURACY OF DIFFERENT ACTIVITY TRACKERS IN ESTIMATING STEPS TAKEN
RESEARCH PROJECT
MANY TRACKERS, FEW ENDPOINTS…
‣ JawBone Up ‣ FitBit Flex ‣ BodyMedia ‣ Withings ‣ GalaxyGear ‣ e.t.c
2
THE GOAL OF OUR RESEARCH AND THE COLLABORATION PROPOSAL IS TO EVALUATE AND COMPARE THE ACCURACY OF THE ACTIVITY TRACKERS CURRENTLY PRESENT ON THE MARKET.
There are many activity trackers on the market nowadays. However, there are very little sound scientific research providing a reference of the devices’ usage in relation to any health related goals.
Source: Berg Insight, October 2013
Total Shipments of wearable devices
Will reach
units by 201764 millions2013
2016
2015
2014
20122011
METHOD
3
Gero created a series of algorithms that can predict age and identify gender of a human being with 86% accuracy (after cross validation) through analysis of data streams gathered from activity trackers. We used data streams from FitBit in our experiment.
NOW WE WANT TO USE DATA STREAMS FROM OTHER DEVICES AND CHARACTERIZE DIFFERENT TRACKERS BY COMPARING THE PREDICTIVE ACCURACY OF OUR ALGORITHMS.
EXPERIMENTCollect data streams from different devices
4
STEPS
GERO GENDER AND AGE
IDENTIFICATION ALGORITHMS
ACCURACY CHECK
GENDER DEVICE%
DEVICE ACCURACY
Analyze data with Gero Gender and Age Identification Algorithms
Compare result with profile data
Calculate accuracy rate for each device
1
2
3
4
DATA REQUIREMENTS
5
Data streams: 1000 or more for each device
500 000 data points in a stream (a few weeks of data from a user)
Uniformly distributed Age and Gender
Representation of data - time series
1
2
3
4
No movement activity in stream - not less than 5%5
For more details, email to [email protected]
ABOUT GERO
Our team expertise, which include professionals in system biology, p h y s i c s , m a t h e m a t i c s , b i o -informatics, data science, IT and wearable electronics created a strong and dedicated unique formula capable of moving the industry forward.
GERO was founded to develop a new wearable 2.0 platform. We created a technology utilizing mathematical models and diagnostics tools that can identify risks of age-related diseases at early stages as well as biological age through analysis of everyday activity and other biological signals.
6
HOW IT WORKS The character of human movements depends on state and intricate balance of interactions between different sub-systems of a human body. We use our models to extract and characterize correlation properties of the tracker signals and relate the motor activity patterns with health conditions.
7
LOCOMOTOR SIGNAL
BEHAVIOR
HEALTH CONDITIONS Biological age Diabetes type 2 Depression Hypertension Parkinson's
BRAIN
ORGANS AND SYSTEMS
!HEALTH
CONDITIONS
COLLABORATION
8
WE WANT YOU TO TAKE PART IN OUR RESEARCH AND PUBLICATIONS
‣ Share a list of devices supported by your service
‣ Share with us data streams of selected devices
‣ Co-publish a press release officializing the collaboration
‣ Co-publish the results of research in a peer reviewed, PubMed indexed journal
‣ Take part in media campaign
The goal of our research and the collaboration proposal is to evaluate and compare the accuracy of activity trackers currently present on the market. This should help understand the relevance of the trackers readouts in relation to health-related endpoints.
NEXT STEPS COLLABORATION BETWEEN OUR COMPANIES
Provide any additional information.
9
Define the starting date of the collaboration.
Finalize the terms of collaboration.
Sign the contract certifying the collaboration.
1
2
3
4