AI4Lungs

Principal investigator: Viviana Mangiaterra 

Partners: InescTec (Coordinator), Fraunhofer, Technische Universitat Kaiserslautern, i3S - Instituto de Investigação e Inovação em Saúde da Universidade do Porto, KPMG Somekh Chaikin Partnership, EXUS AI Labs, YonaLink, Reichman University, Amsterdam University Medical Centres, Gaia Hospital, Mor Research, Rabin Medical Ctr, Oslo University Hospital, Val d’Hebron, Greek Ministry of Health, Future Needs, Institute for Digital Transformation in Healthcare (idigiT), Timelex, Bocconi University 

Sponsor: European Union's Horizon Europe Research and Innovation Action “Computational models for new patient stratification strategies” (HORIZON-HLTH-2022-TOOL-12-01-two-stage)  

Duration: 42 months starting from January 2024 

Abstract:

AI4Lungs aims to develop and validate novel, robust data-driven computational tools and computational models/algorithms to improve patient stratification in order to optimise diagnosis and treatment of infectious and noninfectious respiratory diseases. The artificial intelligence (AI)-based tools, designed to be streamlined into existing clinical pathways, will support decision making along the patient journey from early suspicion enabling personalised diagnosis and treatment planning to improve the standard of care and patient outcome. This cross-border research combines multiple data sources from clinical partners in five countries, data registries and open national/international databases including structured and unstructured data types e.g., lab data, genetics, comorbidities, medications, lifestyle, gender, text-based reports, raw imaging data and clinical results (x-ray, CT) as well as novel data from digital stethoscope (auscultation) and –omics (liquid biopsy). AI4Lungs applies a suite of methods including signal and image processing, machine learning, deep learning, natural language processing (NLP) and disease-specific computational modelling. The AI-based algorithms developed will identify imaging characteristics that are invisible to the human eye and draw hidden information from data. A patient digital twin (DT) will be developed, which can model individual patient performance and response to treatment even if they have not performed all the testing or have partial data. The results from the tools and models will be presented as a robust and accurate, easy-to-adapt decision support system (DSS), which will be transparent to patients and clinicians while promoting more accurate and timely diagnoses and recommending the most appropriate personalised therapies for patients with lung cancer, interstitial disease and inflammatory conditions (viral and bacterial pneumonia including COVID-19).