Machine Learning in Parkinson’s Disease Diagnosis

The role of mitochondrial dysfunction in the onset of Parkinson's disease (PD) has been reported in recent studies. A machine-learning approach was used by researchers from the Luxembourg Institute of Health (LIH) to identify mitochondrial interactions as a novel biomarker for Classify patients with PD.

Since traditional research focusing on individual mitochondria has not provided satisfactory insights into PD pathogenesis, our pioneering work has taken a step forward by exploring the networks of interaction between these organelles," explained Feng He, PhD, group leader of the LIH Department of Infection and Immunity Immune Systems Biology Group."

This result was publish in the journal Nature partner Journals Systems Biology and Applications, under title 'Mitochondria interaction networks show altered topological patterns in Parkinson’s disease '.

Mitochondrial dysfunction is associated with Parkinson's pathogenesis. In PD patients, however, individual mitochondria-based analyses do not display a standardised function, the researchers wrote. Because mitochondria communicate with each other, we assume that in topological patterns of mitochondria interaction networks (MINs) there may be PD-related characteristics. Here we show that in colonic ganglia, both from healthy controls and PD patients, MINs developed nonclassical scale-free supernetworks; however in PD patients, altered network topological patterns were observed.

A large 700 Gigabyte dataset of three-dimensional mitochondrial images of colonic neurons, obtained from PD patients and healthy controls, and stem cell-derived dopaminergic neurons was analysed by the researchers. The researchers found that in PD patients, unique network structure features within MINs were altered.

These different topological patterns in MINs may mean that energy and information are possibly produced, shared, and distributed less competently in the neuronal mitochondria of PD patients relative to healthy controls, suggesting their connection to mitochondrial damage, deficiencies, and fragmentation typical of neurodegenerative disorders,” explained He.

When applying a machine learning approach to analyze these MIN characteristics, the researchers observed that the use of a combination of those network features alone allowed them to accurately distinguish between PD patients and healthy controls.

Our results illustrate the potential of using basic features of the mitochondrial network as novel biomarkers for the early diagnosis and classification of patients with PD, which could help to establish a new health index. As a next step, we will explore how our findings can give new perspectives for understanding various other mitochondrial dysregulation-characterized neurodegenerative diseases, such as Huntington's disease and Alzheimer's, making our work a true example of translational and transversal science,' added Rejko Krüger, professor and director of transversal translational medicine at LIH and co-author

In the application of advanced machine-learning techniques to unravel the dynamic network interactions of cellular organelles for disease stratification, this publication also constitutes a significant step forward. Indeed for our department and for LIH as a whole, data analytics and creative digital technology are a key focus area,' concluded Markus Ollert, director of the infection and immunity department and contributing author of the paper.