Development of a Machine Learning Prediction Model for the Detection of Meniere's Disease from Cerumen Chemical Profiles

点击翻译按钮获取中文摘要

ABSTRACT/PROJECT SUMMARY Ménière’s disease is a chronic, incurable vestibular disorder that produces a recurring set of symptoms as a result of abnormally large amounts of endolymph in the inner ear. Manifestations of the disease include recurrent episodes of vertigo, tinnitus, imbalance, nausea and/or vomiting, a feeling of fullness or pressure in the ear, and fluctuating, progressive low-frequency hearing loss. Diagnosis is difficult because other neurological conditions present some of the same symptoms. Thus, Ménière’s disease diagnosis, which is challenging, imprecise, and time consuming, involves the painstaking process of excluding other diseases with overlapping symptoms. Because it has no known chemical or radiographic markers, diagnosis is based on the observation of a clinical compendium of symptoms, and misdiagnosis is fairly common. If chemical markers of Ménière’s and other relevant neurological disorders could be determined, more rapid and accurate diagnosis could be achieved based on assessment of the presence (or absence) of these relevant compounds. It is hypothesized here that the chemical profile of cerumen can serve as a reporter of the presence of Ménière’s disease and other neurological disorders with overlapping symptoms, and that knowledge of these differential profiles can be leveraged to accurately and rapidly reveal the presence of Ménière’s disease. This hypothesis will be investigated through pursuit of the following specific aims: Specific Aim I: Collection and determination of the mass spectral chemical signatures of cerumen from healthy donors, Ménière’s disease patients, and patients diagnosed with other neurotological disorders with overlapping symptoms. Specific Aim II: Development of machine learning prediction models that enable accurate determination of the presence of Ménière’s disease and/or other neurotological disorders from cerumen chemical profiles, and reveal the presence of the subset of compounds that are important for the ability to distinguish Ménière’s disease samples from others. Specific Aim III: Structural characterization of compounds revealed by the machine learning prediction model(s) developed in Specific Aim II, to be associated with Ménière’s disease. The results of this work will reveal whether there is a correlation between the lipid profile of earwax and the presence of particular disease states. Structural information will be acquired on the molecules that are responsible for the differences in healthy and Ménière’s disease patients. The information revealed would provide the opportunity for development of a potential non-invasive method for the rapid diagnosis of Ménière’s disease.