Evaluating Spirometric Trends in Cystic Fibrosis Patients

摘要

In this research, we applied both supervised and un-supervised machine learning methodologies to spirometric data from patients with cystic fibrosis (CF). We developed an ensemble of neural networks to evaluate the severity of chronic CF within an individual, given the appropriate clinical input data, and a series of reference equations to describe the CF patient's pulmonary function at different ages, heights, and sex groups in order to determine longitudinal spirometric trends. The neural networks were able to be eighty-eight percent accurate when evaluating chronic disease severity and our regression analysis revealed several trends, such as in females with CF, obstruction and functional airflow movement within the lungs generally tends to deteriorate at an accelerated rate compared to males with CF. Our findings have the potential to serve as useful reference tools to physicians in the diagnosis and treatment of cystic fibrosis.