When Parkinson's Meets AI: Models for Disease Progression Expected

The Michael J. Fox Foundation partners with IBM to analyze "treasure trove" of data.

There's a new weapon in the arsenal to fight Parkinson's Disease: artificial intelligence (AI) and machine learning. If the showdown between technology and the debilitating neurological condition goes well, researchers will better understand disease progression, and perhaps create a blueprint for more effective treatments that traditional forms of medical research would take much more time to unveil. Such outcomes would also likely reduce associated healthcare costs and the impact on health systems.

This week IBM Research Healthcare and Life Sciences announced a partnership with the Michael J. Fox Foundation, which includes a grant for an undisclosed amount from the New York City-based foundation, as well as access to data the Foundation has collected for years. The key to progress lies in analyzing this data, known as the Parkinson’s Progression Marker Initiative (PPMI). That capability dovetails with work IBM has been doing in the area of neurodegenerative diseases.

"We have published quite extensively in this area—not just in Parkinson's disease—but also in Alzheimer's, as well as Huntington's," says Jianying Hu, PhD, global science leader, AI for healthcare at IBM Research. Representatives from the two organizations began exploring whether similar methodologies could be applied to the PPMI data. "That's how this collaboration came about," says Hu. 

A Costly Disease
 

According to the Parkinson's Foundation, nearly one million will be living with Parkinson's Disease in the U.S. by 2020, and approximately 60,000 Americans are diagnosed with it each year. It also has a huge impact on the healthcare system.

• A study published by Movement Disorders in 2013, indicates the national economic burden of Parkinson's Disease exceeded $14.4 billion or approximately $22,800 per patient in 2010.
   
• The prevalence of the disease is expected to more than double by 2040, according to an article about the research cited above, published by the Michael J. Fox Foundation.
    
• In a separate study, also published by Movement Disorders in 2013, "a treatment that could slow Parkinson's progression by 50% would yield a 35% reduction in excess costs, representing a dramatic reduction in cost of care spread over a longer expected survival," the Michael J. Fox Foundation reported.

AI and Machine Learning Enhance Data Analysis
 

“We’ve made huge leaps in the last two decades to better understand Parkinson’s disease, but more research is needed to illuminate pathways to better treatments and ultimately, a cure," said Mark Frasier, PhD, senior vice president of research programs for the Michael J. Fox Foundation, via email to HealthLeaders. Parkinson's is a highly heterogeneous disease, he explained, with varied clinical presentations and associated pathology involving numerous genetic mutations.

"The Michael J. Fox Foundation has invested heavily in the collection of comprehensive, standardized data from patients and healthy volunteers," said Frasier. This PPMI data includes imaging scans, biosample analysis, and patient-reported outcomes.

"Today our challenge is to analyze this treasure-trove of complex imaging, clinical, and molecular data using traditional statistical approaches," said Frasier. "AI and machine learning technologies can help scientists identify trends and models across Parkinson's data. These frameworks can help researchers design more efficient and accurate clinical studies and drug trials, speeding discovery and bringing new personalized treatments to patient hands faster.”

Two Phases of Research
 

Initially, researchers will examine the data to produce a  disease progression model to better assess stages of the disease "not just based on symptoms [patients] have today, but based on the longitudinal progression that we have absorbed over time," says Hu.

"From that, our model will also be able to help predict what kind of progression pathway this patient is likely going to take in the next phase of the disease," she says. This work has the potential to enable better assessments plus the ability to track disease progression to better manage patients. The team expects that phase of research to be completed by the end of the year.

The second phase will involve enriching the model and looking at potential clinical and therapeutic use cases and applications.

"The goal is to create models that can identify distinct stages of Parkinson’s and provide predictions for future disease progression,” said Frasier.

Research May Lead to Identification of Sub-Groups, Biomarkers, and More
 

In a blog on this topic, published by Soumya Ghosh, PhD, a research scientist at IBM Research, provides further insights:

• "Because the models provide a quantitative description of progression, they may allow us to discover sub-groups of [Parkinson's Disease] patient profiles that share a common progression pathway through the disease, and may also help us in the identification of biomarkers that are predictive of progression."
   
• "More broadly, insights arising from an improved understanding of [Parkinson's Disease] progression may have the potential to transform the care of Parkinsonian patients. For instance, accurate staging can aid in the recruitment of subjects to clinical trials for new drugs, while the discovery of [Parkinson's Disease] sub-groups can help inform more personalized treatments, and may hopefully improve quality of life and outcomes."