Boston University, UNSW researchers develop AI tool to predict Parkinson's disease

Researchers from the University of New South Wales (UNSW) and Boston University have developed an artificial intelligence (AI) model capable of identifying Parkinson's disease up to 15 years before it is clinically diagnosed. Parkinson's disease is a rapidly increasing neurological disorder worldwide, characterized by the degeneration of brain nerve cells and accompanied by symptoms such as body tremors, slow movement, stiffness and balance problems.

In addition to these well-known physical and cognitive signs, there are other symptoms, such as depression, sleep disturbances, constipation, and loss of smell that can manifest many years before the more recognizable indications of the disease. Currently, there are no definitive tests available for diagnosing Parkinson's disease, and medical professionals rely on symptoms, physical examinations and medical history to make a diagnosis.

To address this challenge, the UNSW and Boston University team trained their AI model using blood samples obtained from a study involving 41,000 participants, who were originally being investigated for cancer and nutrition. The model was trained to detect Parkinson's disease based on the data generated from this study, providing a potentially groundbreaking approach to early detection.


The AI model was provided with blood data from 39 individuals who eventually developed Parkinson's disease, as well as an equal number of control patients chosen randomly. By analyzing metabolites in the blood samples, the AI achieved a remarkable accuracy of 96% in identifying those individuals who would later be diagnosed with Parkinson's disease, and this detection could occur up to 15 years before the disease is clinically diagnosed.

In comparison, current clinical methods for diagnosing Parkinson's disease typically yield a range of accuracy between 65% and 93%, highlighting the limitations of existing approaches.

Metabolites are produced by the body as a result of the breakdown of various substances, including food and drugs. Variations in the concentrations of specific metabolites can indicate the presence of a disease or even resistance to it.

The researchers from UNSW have identified a significant finding in their study. They identified a particular chemical, likely a triterpenoid, as a crucial metabolite that could potentially prevent Parkinson's disease. They observed that the abundance of this triterpenoid was lower in the blood of individuals, who later developed Parkinson's disease compared to those, who did not develop the ailment.

Additionally, the research revealed a link between a synthetic chemical called a polyfluorinated alkyl substance and an increased risk of Parkinson's disease. This chemical was found at higher levels in individuals, who were later diagnosed with Parkinson's.

While the results are promising, the UNSW team acknowledges the need for further research to validate these findings, preferably involving larger populations. However, they are optimistic that this breakthrough can lead to earlier detection of Parkinson's disease, ultimately improving the quality of life for patients and potentially reducing healthcare costs. As a step towards that goal, they have already made the tool publicly available.

FAQs

1. What does AI stand for?
   AI stands for Artificial Intelligence.
2. Where have the researchers uploaded AI tool to recognize Parkinson's disease?
   On Github.