Noninvasive assessment of pulmonary edema using machine learning

Polina Golland with Ruizhi Liao, Steven Horng, and Seth Berkowitz

Noninvasive assessment of pulmonary edema using machine learning

Heart failure is the number one cause of hospitalization in the United States, with high readmission and mortality rates. Effective treatment for acute heart failure depends on the accurate measurement of fluid overload in the lungs, known as pulmonary edema, but this is challenging and costly. This team uses machine learning algorithms to automatically assess the severity of pulmonary edema from chest X-ray images. Combined with other clinical measurements, the project’s unique fluid status visualization will provide accurate, noninvasive, longitudinal tracking of pulmonary edema, and of patients’ response to treatment. This visualization will enable physicians to deliver better targeted therapies.

Read about the team's research in MIT News: https://news.mit.edu/2020/anticipating-heart-failure-machine-learning-1001

Read the team's papers:

Watch Ray Liao present Non-Invasive Assessment of Pulmonary Edema Using Machine Learning at IdeaStream 2021: