Parkinson’s disease (PD) is a progressive neurodegenerative disease that affects more than one million people

in North America. While symptomatic disease is diagnosed when patients develop the classic motor features,

many non-motor symptoms predate the motor symptoms, including constipation, sleep disturbances, poor sense

of smell, depression, and anxiety. This period of the disease has been termed the prodromal period. Although

medical treatments improve symptoms, there are currently no proven therapy to slow disease progression,

possibly due to failure to diagnose prior to irreversible brain damage. As a result, developing methods to

identify PD patients earlier, during the prodromal period, are critical to begin to find medications that modify

the inexorable disease progression. We previously developed a computer model that identifies people who are

going to develop PD, during the prodromal period, by using only medical diagnosis and procedure codes

obtained from the Center for Medicare and Medicaid Studies (CMS). This computer model used a diagnosis

system [International Classification of Disease version 9 (ICD9)] that was changed in 2015 to International

Classification of Disease version 10 (ICD10). This system has four times as many diagnoses so our computer

model must be redeveloped using this new system prior to being implemented. We will also use Medicare Part

D prescription drug data to identify medications associated with a lower risk of developing PD. This proposal

addresses the Parkinson’s Research Program focus area of using digital health technology, specifically digital

medical claims data, to develop a predictive model for early identification of PD. Forbes lists big data analytics

as the digital technology with the greatest impact on healthcare in 2019. As such, this research proposal

represents a timely and highly responsive application in response to the Department of Defense digital health

technology focus area. The studies outlined in this proposal are an important first step in developing a

prodromal PD cohort using an ICD10-based PD predictive model. Upon completion of these studies, the final

computer model we develop will be ready to be implemented using contemporary Medicare claims data and

will allow researchers to identify those Medicare recipients who are at highest risk of developing PD in the next

three years. Developing a computer model from digital medical claims data could potentially impact all patients

with PD. Early identification could result in early treatment when patients are beginning to experience disease

complications, such as falls, fractures, or traumatic brain injuries. This would have an immediate impact on

patients with PD. Clinical use of this type of computer model would have little risk to patients since they are all

Medicare age and are guaranteed insurance coverage. In addition, researchers have been trying to identify

people with PD during the prodromal disease phase, in order to study the effect of medications that could

potentially slow disease progression. So far, these efforts have only been modestly successful. This study has

the potential to be a major advance in these efforts. Finally, any medications we find that are associated with a

lower risk of developing PD may represent potential disease slowing medical therapies.