Vicomtech participates in the european project Lucia, an iniatiave born with the idea of paving the way for better management and treatment of lung cancer

Lung cancer is a major health concern and the cancer with the most mortality in the EU (388,000 deaths in 2018). Smoking is the most common risk factor, responsible for about 85% of cancers. Delayed diagnosis, due to difficulties in early detection prevents more successful interventions at early stages. Ongoing trials are already showing the potential benefit of Low-dose Computerized Tomography Screening, mainly targeting heavy smokers. Yet, risk models are required for screening candidate selection that improve standard inclusion criteria for screening.

In this context Vicomtech leads the development of AI-based Risk Models in collaboration with other partners in the LUCIA Horizon Europe Project as well as the development of the project data space and virtual research environment. LUCIA is a project for the cancer mission aimed at the understanding of lung cancer risk factors and lung cancer development, as well as the development of new technologies and tools for lung cancer screening and early detection, leveraging precision medicine approaches. LUCIA considers a wide range of diagnostic and population health data sources such as cohort study data, electronic health records, geolocated open data, genomics, transcriptomics and epigenomics data, LDCT imaging, digital pathology and volatomics technology (biomarkers of breath and skin headspace) among others.

Classical statistical risk models are typically based on long-term cohort studies with extensive risk factor data collection. Despite their potential benefit as compared with inclusion criteria for screening trials, as colleagues from DKFZ demonstrated (Bhardwaj et al 2022, Lung Cancer), they show limitations with personalized prediction. However, AI risk models have the promise of improving risk assessment with better predictive accuracy, leveraging multiple data sources and diagnostic technologies in a stratified manner. Beyond traditional risk factor data, we are currently developing, together with LUCIA clinical partners, lung cancer risk models based on large-scale electronic health record data, using latest deep learning technology, which has the potential, as shown in research studies (Yeh et al 2021, J Med Internet Res) to provide accurate personalized predictions (AUC~0.9) in short time frames. Additionally, we are developing risk models based on LDCT imaging data either solely or by combination with other risk factor data, as well as studying the potential malignancy of suspicious lung nodules by AI-enabled image analysis. Additionally, we will study the potential of inclusion of other molecular data sources that will be collected in the extensive LUCIA prospective studies. Our goal is improving current approaches to lung cancer early detection and future implementation of screening programmes with AI-enabled personalized prevention approaches.

LUCIA project has received funding from the European’s Union’s Horizon Europe research and innovation programme (Cancer Mission calls), under grant agreement No. 101096473.