Imagine one day, your breathing became consistently labored and shallow. Months later you were finally diagnosed with pulmonary fibrosis, a disorder with no known cause and no known cure, created by scarring of the lungs.
Any patient in this situation would want to know the severity of their condition. Unfortunately, the outcomes of pulmonary fibrosis can range from long-term stability to rapid deterioration, and doctors are unable to tell where a patient falls on this spectrum.
With the help of a non-profit called Open Source Imaging Consortium (OSIC), my team is working on a machine learning project to mitigate this problem. The aim of our project is to predict the patient’s severity of decline in lung function based on CT scans of their lungs. OSIC provides anonymized patient data and CT scans so that we can train the model. Our project will allow patients and their families to better understand their situation when they are first diagnosed with pulmonary fibrosis.
Our team experimented with different approaches and evaluated our results based on a modified version of Laplace Log Likelihood, which is a useful evaluation metric in medical applications. Our predictions have been steadily improving. Hopefully, we’ll hear about the positive impact of our project from OSIC soon!
