2023/05/30 | Research | Artificial Intelligence
Over 200 diagnoses exist for the umbrella term that represent interstitial lung diseases (ILDs). These are currently determined by an expert board based on CT imaging, patient data, pulmonary function tests, and histology. Artificial intelligence (AI) methods could help aid in accurate ILD diagnosis and predict prognosis and progression in a holistic system. In a review paper, researchers from the AI in Health and Nutrition Lab and the Department of Radiology of the Inselspital have compared current approaches discussed in literature, identifying potential gaps and areas requiring further research as well as the most promising results.
Prompted by the urgent need for diagnostic support during the COVID-19 pandemic, AI methods assisting clinicians in ILD diagnosis (computer-aided diagnosis CAD) have risen exponentially during the past years. Albeit transferable to ILDs, these methods can still not provide information on risk factors for progression in individual patients.
In the area of disease progression, a combined analysis of imaging and clinical data, lung function tests, and laboratory values is required. The review paper compared regression methods, proportional hazard models, generative models, and unsupervised learning approaches. Authors conclude that AI systems are able to classify suspected ILD cases and assess disease severity showing great potential as assisting tools to be integrated into radiologist workstreams to help for a quicker and more accurate diagnosis.
Link to the study
AI in Health and Nutrition