Principal Investigator: Dr Gabriel Gonzalez-Escamilla
University MC of the Johannes Gutenberg University (Germany)Tags: 41865, disease-prognosis, healthy-aging, Machine Learning, Multiple sclerosis, neurodegeneration
The most important challenges in clinical and biomedical research include the need to develop and apply tools for the effective integration, analysis and interpretation of complex data with the aim to identify testable hypothesis, and build accurate models of disease development and progression.
Recent advances in magnetic resonance imaging techniques allow the acquisition of increased amounts of data. With more and more data available, machine learning techniques, which are closely related to statistics, are becoming increasingly popular. Machine learning allows predictions based on sets of individual variables. The use of Machine Learning, more superficially deep learning, in Healthcare problems can be of great importance, mostly because it offers the opportunity of developing algorithms that identify complex patterns within large amounts of data, otherwise unachievable with standard methods.
Health maintenance is in fact a multifactorial phenomenon, determined by interactions of its factors, including genetic inheritance, internal physiological processes, personal behaviors, and the general external environment. Furthermore, neurodegenerative diseases are neurological illnesses that manifest as movement disorder, cognitive impairment, and/or psychiatric disturbance, attributed to neuronal cell death. Alzheimer’s disease (AD), Parkinson’s disease (PD), among others are the best known neurodegenerative diseases, but also demyelinating diseases, such as multiple sclerosis belong to this category. These illnesses occur worldwide representing a great challenge and effort for healthcare systems. We propose that by feeding a deep learning algorithm with several amount of data from the UK biobank the deep learning algorithm can learn how to recognize patterns associated and differentiated with certain phenotypes, such as disease conditions (e.g., multiple sclerosis). The definition of a framework for correct and opportune disease detection and surveillance would provide way for epidemiologic studies that facilitate health experts to deploy preventive measures and help healthcare administrators to make optimal decisions.
Further, such a context provides a novel way to capture individual differences within the general population as well as disease groups that relate to additional facets of various diseases or even predict future outcomes.