HOME ›   OUR ROLE & IMPACT ›  FoxFeed Blog

FoxFeed Blog

Winning Data Challenge Entry Seeks to Use ‘Machine Learning Approach’ to Improve the Monitoring of Parkinson’s Disease

Posted by  Nate Herpich, April 24, 2013

LIONsolver team members, from left, Dr. Roberto Battiti, Drake Pruitt, Dr. Mauro Brunato

LIONsolver team members, from left, Dr. Roberto Battiti, Drake Pruitt, Dr. Mauro Brunato

In February, The Michael J. Fox Foundation launched a $10,000 Data Challenge to spur ideas for using patient data, collected using smartphones, in Parkinson’s disease (PD) monitoring and treatment.

This week, the Foundation announced the winner of the prize — a team from LIONsolver, Inc., who have developed a “machine learning” approach to tackling the problem. Here, we speak with LIONsolver CEO Drake Pruitt to learn more about what “machine learning” is, his team’s motivation for entering the competition, and their hopes for developing practical applications for Parkinson’s patients that use smartphone-collected data moving forward.

MJFF: Why enter this particular challenge?  What is your interest in Parkinson’s disease?

Drake Pruitt: Our team is very concerned about the quality and affordability of healthcare in the U.S. and across the world. As problem solvers, we continually look for ways to attack these challenges. When we learned of this competition, we educated ourselves about Parkinson's, and a new understanding of the hardship it places on patients motivated us to enter. We felt confident that a ‘machine learning and optimization’ approach (what we call LION) would be the best way to solve the problem.

We're also big fans of Kaggle.com (who hosted the Challenge) and how they aggregate really hard problems for talented teams.

MJFF:  What is a ‘machine learning approach’?

DP: Machine learning is a branch of artificial intelligence which allows computers to build powerful models and improve their predictions by learning from the data itself. The results automatically improve as more data becomes available.

MJFF: How could machine learning lead to better diagnosis and monitoring of PD?

DP: Our team believes that training computers to evaluate and respond to data from mobile devices can improve the feedback between doctors and patients with Parkinson's. For example, a doctor could gauge patient’s symptoms such as tremors via smartphone data, rather than relying on written diaries or interviews with patient caregivers. We think this can improve patient well-being, while also reducing the long term costs of care delivery.

We see this opportunity as part of an overall trend in healthcare toward applying forecasting and prediction to health record and wellness data, in order to help doctors and their patients achieve healthier lives with manageable healthcare costs. In short: More and more mobile devices are linking with monitoring services to analyze a growing amount of data. This analysis will provide a unique opportunity to take better care of patients, and to teach patients to take better care of themselves.

MJFF: Discuss the potential of your project. What can you expect to accomplish with this particular dataset, and how might that inform future research projects?

DP: This project proved the feasibility and value of gathering mobile data for monitoring Parkinson's. It also validated the power of our LIONsolver platform for machine learning and optimization.

As a next step, we'll be looking at how to deliver Web services (ways to communicate between devices over the Internet) that make the data actionable and helpful to physicians and patients alike.

MJFF: In your entry, you mention the potential for an ‘intelligent mobile application’ to aid in the monitoring of Parkinson’s disease. What does this mean, and how might such an application work?

DP: Mobile devices and low-cost sensors are creating a way for patients and clinicians to stay more closely connected and informed. This can lead to more proactive care and improved communications. The key lies in building intelligent systems that make sense of the data and present it in a way that is relevant to doctors and patients alike. We're very excited about the potential here.

MJFF: You mentioned that you learned a lot about Parkinson’s disease prior to submitting your project. How did you go about doing this?

DP: Dr. Michele Tagliati, director of the Movement Disorders Program in the Neurology Department at Cedars Sinai, and his team were invaluable to us in this challenge. As clinicians and researchers, they were able to communicate the symptomatology of Parkinson's and the needs of clinicians in a way that allowed our data scientists to easily translate Parkinson’s management goals into mathematic analysis models on our LIONsolver platform. From there, the software did all of the work for the challenge, except to write up the results and our Kaggle submission.

MJFF: What does it mean to your group to win this award?

DP: Our team has many years of experience in solving complex data science problems. However, winning this competition which included so many well-qualified teams, and doing so in service to a great organization like The Michael J. Fox Foundation, has been extremely rewarding. We're thankful for the opportunity to show what we can do.

TAGS: Research News, Parkinson's Data Challenge, Data Sharing, Foundation News

 

 

EXISTING MEMBER?

SIGN IN

Close

New to MJFF?  Register Now

 

 

 

  REMEMBER ME

 

Forgot username/password?

 

  Applying for a grant? Sign in here.

EMAIL TO A FRIEND

Close