MD2K hosting Career Day on Oct. 25 — Find out more.

Empowering individuals to improve health through the
real-time analysis of high-frequency mobile sensor data

The Center of Excellence for Mobile Sensor Data-to-Knowledge (MD2K) has shown that the key to improving your health can be the cellphone in your hand or the smartwatch on your wrist. That is because MD2K has developed the tools that make it possible to take complex, high-frequency mobile sensor data and turn it into something more meaningful than a heart rate or step count. 

MD2K has harnessed the sensors commonly found in smartphones and fitness trackers, along with creating some new ones, to find indicators of health states that provide insight into conditions such as drug addiction, smoking, heart failure and obesity.

Behavioral Biomarkers

biomarker validation

To identify and validate biomarkers, MD2K researchers have correlated sensor data with self-reports known as Ecological Momentary Assessments (EMAs) to identify biomarkers within the sensor data of stress, craving, smoking, drug use, and eating. These biomarkers are exposed within the millions of bits of sensor data by using computational models that identify combinations within the sensor data. 

For example, a wrist sensor might identify a hand movement to the mouth that could mean anything. But, when that information is combined with respiratory sensor that shows inhalation and exhalation, the moment a puff is taken becomes clear. 

Likewise, researchers have developed a way to identify episodes of craving, stress, eating, and cocaine use from within the mobile sensor data deluge.

jitai3

mHealth Interventions

When biomarkers indicate a health problem, such as smoking lapse or drug use, MD2K’s novel software platform, generates a just-in-time adaptive intervention, which is sent to the person’s phone. It may be something as simple as a prompt to engage in a one-minute stress reduction exercise or as complex as a call from a counselor.

PhoneSensors

Big Data Analytics

The quest to identify health behaviors from mobile sensor data has led to methodological advances that enable researchers to visualize patterns in the mobile sensor data and even predict the risk of a future adverse health event.

Mobile Sensor Big Data Software

Software1As one group of MD2K researchers was discovering and developing biomarkers and methods to identify them, another team went about the work of building a platform to give mHealth researchers a powerful means to collect, store, analyze and interpret the data coming from these high-frequency mobile sensors.

MD2K designed the revolutionary mCerebrum platform from the ground up. The platform spans more than 23 apps that combine to create an ecosystem that allows MD2K to concurrently run field studies at more than 10 sites, processing and storing an expected 106,806 person-days or 4.7 trillion data points.

mCerebrum has a big-data cloud counterpart, Cerebral Cortex. Cerebral Cortex runs on Apache Spark, which has been improved with three capabilities developed by MD2K scientists. BigDatalog, Titian, and BigDebug provide a declarative language and an interactive, real-time debugging framework with data provenance with the goal of better supporting MD2K data processing workflows. 

Lock blueBehavioral Privacy

MD2K has also addressed the issue of privacy for the data of participants in mHealth field studies with the development of mSieve​ , a ​privacy framework that uses a new behavioral privacy network based on differential privacy.

Real-life Deployments

MD2K has begun the validation and discovery process through scientific studies. At present, 10 studies are underway at locations around the country. These studies are applying the MD2K biomarkers and methods to study smoking, eating, stress, congestive heart failure, oral health, cocaine use and behavior change.

 

 

 

Copyright © 2017 MD2K. MD2K is supported by the National Institutes of Health Big Data to Knowledge Initiative (Grant #1U54EB020404)

Team: Cornell Tech, GA Tech, Harvard, U. Memphis, Northwestern, Ohio State, Open mHealth, UCLA, UCSD, UCSF, UMass, U. Michigan, Utah, WVU