Applications of Machine Learning in Pulmonology: Tools for Clinical Decision Making

The general objective of this research project is to explore how artificial intelligence (AI) for the analysis of large and complex clinical data, can improve our respiratory practice. Driven by our successful development of software for automated pulmonary function interpretation, we will investigate a number of new potential applications. We will explore and develop novel AI-based models for phenotyping, diagnostic and therapeutic decision support. To improve our understanding of the heterogeneous lung disease COPD, we will use large clinical datasets to computationally extract phenotypes that can aid in predicting therapeutic benefits of specific interventions. Diagnostic decision support tools will be developed to support multidisciplinary board discussions for pulmonary hypertension which has a difficult diagnosis. The data from randomized controlled trials in COPD will be addressed to develop mathematical models to predict individual treatment response or effect. By using the state-of-the-art methods and developing novel models on unique datasets, this project will be at the foreground of modern and personalized medicine.