Virtual Coach App Suite-Quality of Life Technology Center - Carnegie Mellon University

iCoach App Suite


Virtual Coach Apps for better health at work or at play.


Carpal Tunnel Coach (Asim Smailagic)

The Carpal Tunnel Coach is a wearable computer platform for prompting and guiding heavy computer users and patients through a simple series of exercises that can prevent or alleviate the symptoms of carpal tunnel syndrome.

A comfortable eWatch sensor is worn on the user's wrist. The user sets a time interval, and the eWatch reminds them to take a break from office work to do the standard six-step, preventative exercise developed at the University of Oklahoma Orthopaedic & Reconstructive Research Foundation.  The users' wrist motion is logged by accelerometers in the eWatch. 

A K-nearest-neighbour Machine Learning algorithm programmed with training data detects when the user completes the Carpal Tunnel exercises and logs the data.  User activity reports can be downloaded for online or offline viewing to track exercise sessions completed or ignored as well as other data.

 

HeadCoach (Kevin Huang)

HeadCoach is a personal balance-exercise monitoring device.  The system is ideal for physical therapy patients prescribed Vestibulo-Ocular Reflex (VOR) exercises; but general users seeking to improve their balance or alleviate minor head, neck or eye strain can also benefit.

The device consists of an intelligent hat that monitors a user's balance exercise performance and then offers them feedback or guidance.  The current prototype features a baseball cap fitted with an iPhone 4 acceleromoter and gyroscope for a complete motion sensor/user interface package.  The simple hat plus iPhone form factor allows the user to easily don and remove the device anytime, anywhere. To start a session, the user puts the hat on and touches anywhere on the screen; after performing the balance exercise, the users taps the screen again to end the session.

The HeadCoach provides users and therapists with an automated record of direct data on exercise performance, including a history of completed sessions and a log of the performance quality, including frequency, duration and correctness. Data can be immediately sent to a clinical server or saved and uploaded at the next clinical visit. As the user's data accumulates, the HeadCoach can be used to better assist the user through correct performance of prescribed therapy routines.

 

Ergobuddy Coach (Asim Smailagic)

The Ergobuddy Coach detects a range of work-activity contexts to provide intelligent information and advice for preventing on-the-job injuries and promoting trained ergonomic work practices. The device is especially suited to workers facing frequent, physically strenuous tasks with high-mobility, such as package delivery specialists. 

Handheld and supportive wearable devices are used to infer the worker's activity context.  Example contexts include: sitting, standing, walking, running, lifting, carrying, sweeping/mopping, using stairs, using a ladder, or using a cart.  Data is gathered from five eWatch sensors worn at the arm, ankle, back, lanyard and wrist positions. 

Sensor Fusion algorithms combine local device contexts to a master device for decision-making and response feedback.  The fusion technique also creates system agility and resilience to failure so that all sensors need not be available at all times.  Fused sensor results can be used to target intelligent response and advice during specific work activities.

 

Stress Management Coach (Brian French)

The Stress Management Coach is an efficient electronic data collection device for administering, recording and scoring self-report assessments of mental and physical stress levels.

The device prompts the user to report perceived stress at random intervals to collect stress perception data from the user both before and after rest and stressor periods.  The process is known as Ecological Momentary Assessment (EMA). The self-reporting information quantifies the user's subjective experience of stress as well as associated contextual factors.  Results are fed through advanced machine learning algorithms for personalized stress inferencing that enables automatic stress detection and trusted response advice.

For physical symptoms, a multi-sensory platform provides continuous physiological measurement.   Six wireless sensors can be worn on the user's chest, including:  Galvanic Skin Response (GSR), ECG, triaxial accelerometer, skin and ambient temperature, and respiratory inductive plethysmograph (RIP).  When combined, these sensors provide readings of 14 unique physical stress indicators that can be personalized to the individual user.




Core Benefits:

  • Detects and monitors behavior
  • Provides smart alerts and reminders
  • Offers personalized feedback and direct guidance
  • Improves activity tracking and reporting

Targeted User Populations:

  • Workers across a wide spectrum of activity levels
  • General health-conscious consumers
  • Physical Therapy patients with specific, prescribed routines

"Building upon research in pervasive and context aware computing, Virtual Coaches are a whole new class of mobile applications that apply machine learning algorithms in real-time to determine what a user is doing, infer user intent, and to proactively anticipate user needs such as providing reminders to do a forgotten task, background on a current situation, encouragement, or detailed instructions."

 

-- Daniel P. Siewiorek, Carnegie Mellon University



About the Research:


Current work focuses on improving real-time reporting of feedback and coaching guidance across systems; experimentation with kinnect-based sensing as a coaching alternative; smartphone integration and cross-platform compatibility; and improvements to the wearable sensor interface designs for user comfort and fashion.


Learn More:

eWatch implementation of a carpal tunnel coach
Carpal tunnel coach with eWatch

headcoach app interface 1
HeadCoach App interface - start

headcoach app interface 2
HeadCoach App interface - history

headcoach app interface 3
HeadCoach App interace - session

Activity recognition sensors for ergo buddy coach
Ergo buddy coach activity recognition

Contacts:

Dan Siewiorek,
Lead Researcher
dps@cs.cmu.edu
(p): 412-268-2570


Asim Smailagic,
Lead Researcher
asim@cs.cmu.edu
(p): 412-268-7863


Kristen Sabol,
QoLT Communications and Media
ksabol@cs.cmu.edu
(p): 412-303-7396