Israeli neuroscientists win prestigious data competition for Parkinson’s research
The team from Bar-Ilan University was one of four winners and the only winning group selected from outside the US
By ILANIT CHERNICK
An Israeli research team was selected as one of four winners in a data competition designed to test new methods of predicting the severity of Parkinson’s disease patients in their homes.
The Israeli team of neuroscientists from Bar-Ilan University (BIU), was the only winner from outside the US and competed against 43 teams including groups from Harvard Medical School, the University of Rochester Medical Center, and the University of Michigan, among others.
The DREAM Challenge was co-sponsored by Michael J. Fox Foundation for Parkinson’s Research (MJFF) and Sage Bionetworks BEAT-PD (Biomarker and Endpoint Assessment to Track Parkinson’s Disease).
Speaking to IsraelNewsStand, team member Prof. Izhar Bar-Gad, who heads up the university’s Gonda Brain Research Center’s Neural Interfaces Lab, explained that in the BEAT-PD challenge, they analyzed the accelerometer data changes for example in hand movements that were recorded “using smartwatches to predict the subjective perception of Parkinson’s disease patients regarding different symptoms (tremor, dyskinesia and medication effect).”
He said that the research “consisted of generating machine learning algorithms based on neuroscience to predict the perception of these patients and the score of each group in the challenge was simply based on the difference between the predictions of the algorithms and the actual report of the patient.”
Bar-Gad said that the challenge “was unique” because of two main factors.
“The data was acquired from patients behaving freely during their everyday lives in their natural environment (home, work, etc.) rather than performing specific tasks in the clinic,” he explained, adding that the second reason was that “the rating of the symptoms was performed subjectively by each patient rather than (semi) objectively by a clinician or another expert in the field.”
According to Bar-Gad, this type of information is far more reflective of the real problem of predicting the patient’s state “and thus provides an important step in generating automatic and semi-automatic tools for aiding the treatment of Parkinson’s disease patients.”
Asked about what made this challenge different to the usual way of addressing a problem in the scientific world, Bar-Gad said that usually, in the normal scientific process, hypotheses are tested in prolonged, well controlled, experiments and studies in which the researcher and his supporting team function by themselves for a long time.
“In this challenge we had dozens of teams from around the world trying diverse methods of addressing the research question and severe time limits leading to a very dynamic and competitive setting,” he continued. “We were excited to see that such an approach to research yields such results for many of the teams and of course overjoyed to see that our results were competitive with those from larger, more established, universities and companies from around the world.”
Addressing what’s next for the team, he said that they are looking forward to the next stage of the challenge, which is the community phase.
“The top teams from around the world [will] collaborate to generate a single, better, predictor combining the results of multiple teams providing, hopefully, hope for patients suffering from the disorder,” Bar-Gad said.
The BIU team, who nicknamed themselves “HaProzdor,” included Ayala Matzner, Yuval El-Hanany and Bar-Gad, who are all from the Neural Interfaces Lab at BIU’s Gonda (Goldschmied) Multidisciplinary Brain Research Center.
Asked about their work in the lab, Bar-Gad pointed out that their research is typically based on the data that is collected from patients suffering from different neurological and psychiatric disorders and from experimental animal models of these disorders.
“We are using many of the key concepts from the challenge in two of our existing projects: the first, which is “predicting the expression of motor tics in Tourette syndrome patients using video and kinematic sensors,” adding that this is a study led by Yocheved Loewenstern in collaboration with clinicians from Schneider hospital led by Dr. Noa Ben-Aroya.
The second is studying the neurophysiological correlation of free behavior in experimental models using similar sensors, which is a study led by Orel Tahary.
“This bidirectional interaction between research projects performed on self-collected data and ones performed on collaboratively-collected data are crucial for advancing scientific and clinical research,” Bar-Gad concluded.
Senior Vice President, Research Programs at the MJFF Mark Frasier said the “foundation has supported research into sensors and other digital tools for Parkinson’s for many years” and that the “BEAT-PD projects are unlocking the potential of data collected by digital devices to help people with Parkinson’s, their physicians, and researchers.
“Now more than ever, we understand the critical importance of remote monitoring for the safe and effective delivery of healthcare and the progress of clinical research,” Fraiser added.
Forty-three teams participated in the Challenge with data hosted by the BRAIN Commons, a scalable cloud based platform for computational discovery designed for the brain health community. The winners will share a $25,000 prize.