How AI can predict Parkinson’s disease up to 15 years before diagnosis based on blood analysis

Parkinson’s disease is a debilitating neurological condition characterized by symptoms like slow movement, tremors, muscle stiffness, and reduced balance. With the world’s fastest growing neurological disease, an AI tool has been developed that can predict Parkinson’s with up to 96% accuracy up to 15 years before clinical diagnosis by analyzing the chemicals in blood. While promoting early diagnosis, the AI tool also uncovered chemicals that were strongly linked to a correct prediction. At UNSW, in collaboration with Boston University, researchers built the AI tool that can analyze mass spectrometry datasets from blood samples. The Spanish European Prospective Investigation into Cancer and Nutrition (EPIC) study, which involved over 41,000 participants, was used to train the AI model. Metabolites were identified, which are chemicals produced or used in the body as it digests and breaks down substances. Triterpenoid is a key metabolite that can prevent Parkinson’s, found to be lower in those who had developed the disease, while a polyfluorinated alkyl substance was linked to an increased risk of the disease. Every year, Australians with Parkinson’s spend over $14,000 in out-of-pocket medical costs.