Modelling to predict Parkinson’s
Validation and calibration of the PREDIGT score to predict the incidence rate of Parkinson’s disease
People with Parkinson’s disease rely on neurologists with clinical experience to diagnose their illness, often based largely on motor symptoms like slowness, stiffness, tremors and difficulty walking. There are no objective medical tests to identify the disease early, or to predict who might develop it.
That could soon change, thanks in part to Juan Li‘s work at the Ottawa Hospital Research Institute. Li, a statistician, is testing the ability of a mathematical model called PREDIGT to differentiate between those with Parkinson’s versus those without it. She’ll then assess the model’s ability to predict who might develop it in the future.
PREDIGT – developed by Dr. Michael Schlossmacher, a neurologist and researcher at the Institute – considers five factors, including: a person’s exposure to environmental risks; genetic susceptibility (i.e., particular genes linked to Parkinson’s, or a family history of Parkinson’s); chronic tissue changes, such as inflammation and depression; gender; and age. Li and her colleagues think these factors explain the onset and potential spread of the disease.
So far, the model can successfully identify people with Parkinson’s compared to people without the disease, using two carefully studied sets of data.
“Surprisingly, in our first tests the model is as good as a neurologist in accurately diagnosing Parkinson’s, but we need more work,” Li says.
Ultimately, Li and her colleagues want to develop a questionnaire and an online calculator tool that practitioners could use to determine the likelihood someone will develop Parkinson’s during adulthood. If the model works, it could identify people before healthy persons develop any motor symptoms associated with the disease.
“Surprisingly, in our first tests the model is as good as a neurologist in accurately diagnosing Parkinson’s.”
Before they can do that, however, Li will assess the model’s accuracy and validity using a large set of data to see whether PREDIGT can also predict when a person who is at high risk of the illness will actually develop it.
“Prediction is more important and interesting to us,” she says. “If we can tell you that you are at high risk or we can predict you may develop Parkinson’s disease in five or 10 years, that’s when prevention could come into play by modifying concrete risk factors in the future.”
Li hopes using this tool could help people in the early stages of Parkinson’s or those at higher risk of the disease access treatments being developed. For example, subjects with a high PREDIGT score might be considered suitable candidates for participation in clinical trials.
Li, who grew up in Langfang, China, began working in the Parkinson’s field after obtaining a PhD in petroleum engineering. She missed the applied mathematics field that was her first love during her undergraduate years and wanted to work on something meaningful – which led her to the Parkinson’s work in Ottawa.
While working with Dr. Schlossmacher, she met people with Parkinson’s and their families.
“A lot of people have a lot of knowledge about this disease and interest in our model – so that made me feel really happy. This is the perfect chance for me to do something that I really care about.”