News

Sage Bionetworks and MJFF announce the winners of the PD Digital Biomarker DREAM Challenge, the first to use remote sensor data to diagnose and track Parkinson’s disease.

January 17, 2018

Sage Bionetworks and MJFF announce the winners of the PD Digital Biomarker DREAM Challenge, the first to use remote sensor data to diagnose and track Parkinson’s disease.

Sage Bionetworks in collaboration with The Michael J. Fox Foundation announce winners in the DREAM Parkinson’s Disease Digital Biomarker Challenge.

The challenge is aimed at helping refine the use of mobile and wearable sensors to monitor health.

Seattle WA, January 17, 2018 Sage Bionetworks announced today the results of the Parkinson’s Disease Digital Biomarker (PDDM) DREAM challenge, an open crowd-sourced research project designed to benchmark the use of remote sensors to diagnose and track Parkinson’s disease (PD). The winners of this Challenge developed methods that are 38% better than previous models at detecting Parkinson’s disease from a simple walk and balance test, and can predict severity of different Parkinson’s symptoms 58% better than baseline models using wearable sensors. These methods can increase our ability to monitor diseases such as Parkinson’s disease outside of a clinical setting.

Over 440 data experts from six continents participated in the PDDB DREAM Challenge, which focused on developing features, or digital biomarkers, of Parkinson’s disease. This is the first in a series of open, crowd-sourced analytical projects sponsored by Sage Bionetworks, designed to help researchers identify ways to use smartphones and remote sensing devices to monitor health and disease.

The challenge was divided into two sub-challenges. In the first sub-challenge, participants used accelerometry and gyroscope data from mPower – a large mobile health study where over 15,000 individuals with PD or controls used their iPhones to, among other things, perform short walk and balance tests – to extract features that were used to predict whether the user had PD. In the second sub-challenge, participants extracted features for three different Parkinson’s symptoms from a study funded by The Michael J. Fox Foundation (MJFF) where patients performed tasks while wearing three wearable sensors. These features were used to predict clinician-assessed disease severity for tremor, dyskinesia and bradykinesia.

Winning Teams
The first sub-challenge was won by Yuanfang Guan and Marlena Duda from University of Michigan, Ann Arbor, who developed a deep learning convolutional neural network, one of the most advanced artificial intelligence techniques to extract features. Their features, when fed into a predictive model, were able to identify Parkinson’s patients 38% better than baseline models.

In the second sub-challenge, three different groups won for their features predicting the different Parkinson’s symptoms. Bálint Ármin Pataki from Eötvös Loránd University in Hungary took the honors in building features for tremor severity (11% improvement over baseline). Jennifer Schaff, Data Scientist at Elder Research, Inc., used statistical methods to derive features and feature selection to develop the top performing submission in predicting dyskinesia severity (59% improvement).  Team Vision from Columbia University consisting of Yuanjia Wang and Ming Sun used spectral decomposition to build features that outperformed all other teams in predicting bradykinesia (17% improvement).

Next Steps
Challenge winners and other participants have made their methods and source code publicly available and 85 of the participants will now move into a collaborative phase of the project. In this phase, participants will learn from each other and combine methods to both try to improve the value and interpret the clinical relevance of their features.

Parkinson’s Disease
An estimated five million people worldwide are living with Parkinson’s, a neurodegenerative disorder that can cause tremors, gait issues, speech problems, and interfere with memory. These symptoms can change with disease progression, medical treatment, and some lifestyle choices. The use of wearables and sensors has the potential to allow scientists to monitor disease fluctuations and progression to a much higher fidelity than current methods. Unfortunately, how to interpret and use such data, especially when collected in the home by the patient, is still an unsolved problem. New algorithms may eventually allow this data to supplement in-clinic measurements to help patients manage their disease.

“Increasing the accuracy of remote health monitoring will expand our ability to understand how our health is influenced by the context and choices of daily life,” said Paul Tarini, senior program officer at the Robert Wood Johnson Foundation. “It’s promising to see how challenges like DREAM are taking this work to the next level and helping us better understand how to process sensor data so that it is meaningful and actionable.”

About the Challenge
The project took an open, crowd-source approach to address the first step in analysis of sensor data – feature engineering, or the conversion of raw sensor data into analysis-ready data. Top performing teams used a mixture of signal processing and deep neural networks to predict disease state and disease severity.

“The proposed solutions were far outside the traditional techniques used in the field of actigraphy and many of the experts involved in organizing the challenge are reconsidering the way they interpret this kind of data,” said Larsson Omberg, VP of Systems Biology of Sage Bionetworks.

Funding for the challenge has been provided by the Robert Wood Johnson Foundation (RWJF) and The Michael J. Fox Foundation for Parkinson’s Research (MJFF).

For additional information about the PD Digital Biomarkers Challenge: https://www.synapse.org/DigitalBiomarkerChallenge

About DREAM Challenges
First conceived by IBM in 2006, DREAM Challenges have addressed objectives that range from predictive models for disease progression to developing models for cell signaling networks. Designed and run by a community of researchers, DREAM Challenges invite participants to propose solutions, fostering collaboration and building communities in the process. The DREAM Challenges community shares a vision of open collaboration to leverage the “wisdom of the crowd” to improve human health and sciences.

About Sage Bionetworks
Founded in 2009, Sage Bionetworks is a nonprofit biomedical research organization that promotes innovations in personalized medicine by enabling a community-based approach to scientific inquiries and discoveries. In pursuit of this mission, Sage Bionetworks has assembled an information commons for biomedicine supported by Synapse, an open compute space. The commons facilitates open research collaborations and innovative DREAM Challenges; it also empowers citizens and patients to share data and partner with researchers through Sage’s BRIDGE platform (https://developer.sagebridge.org/).

About The Michael J. Fox Foundation for Parkinson’s Research
As the world’s largest nonprofit funder of Parkinson’s research, The Michael J. Fox Foundation is dedicated to accelerating a cure for Parkinson’s disease and improved therapies for those living with the condition today. The Foundation pursues its goals through an aggressively funded, highly targeted research program coupled with active global engagement of scientists, Parkinson’s patients, business leaders, clinical trial participants, donors and volunteers. In addition to funding more than $750 million in research to date, the Foundation has fundamentally altered the trajectory of progress toward a cure. Operating at the hub of worldwide Parkinson’s research, the Foundation forges groundbreaking collaborations with industry leaders, academic scientists and government research funders; increases the flow of participants into Parkinson’s disease clinical trials with its online tool, Fox Trial Finder; promotes Parkinson’s awareness through high-profile advocacy, events and outreach; and coordinates the grassroots involvement of thousands of Team Fox members around the world. For more information, visit us on the WebFacebookTwitterLinkedIn and Pinterest.