Big Data and models for personalized Head and Neck Cancer Decision Support
BD2Decide is an H2020 project that aims to develop a more precise prognostic prediction than the ones currently used in head and neck squamous cell carcinomas (HNSCC). Currently, the treatment decision method is based on TNM system, which considers few risk factors and results in late diagnosis of relevant cases in advanced stages. A more precise prognostic prediction would allow implementing the first-line treatment maximising the therapeutic result.
A multi-centric clinical study has been applied with more than 1.000 patients to validate the system. BD2Decide has built the largest HNSCC dataset with 1537 Stage III-IV HNSCC patients with loco-regionally advanced treated with curative intent. Upon this dataset, an Ontology has been created, providing a formal description of all dataset. All data relevant to the patient have been modelled, containing the demographic and clinical information. The Ontology was enriched with annotations to provide semantic meaning to the data model and external ontologies have been mapped to extend the meaning of these annotations.
BD2Decide introduces a Decision Support System (DSS) that combines existing data, personalised prognostic models, big data analysis techniques and representation technologies for personalised prognosis and treatment decision-making for HNSCC. BD2Decide is thought to help clinicians decide in these challenging situations as it produces not only information from evidence medicine but also prognostic information derived by the embedded models. Two main representation tools have been implemented: Clinical DSS for patients’ data exploration and facilitating the collaboration among the clinicians in treatment decision-making, and the Visual Analytics Tool for population data analysis for research purposes. Significant scientific results have been obtained in BD2Decide as regards the HPV testing in Oral Cavity region vs Oropharynx, and the identification of two relevant transcriptomics signatures. Both works are under publication.
The main role of LST has been the technical management, the design and implementation of the visual analytics tool and the development of the ontology. Other activities where LST has been involved are: system architecture definition, data quality assurance, knowledge management system and the overall system validation.
The consortium comprises 12 partners (4 research institutes, 3 SMEs and 5 cancer clinics) from 6 different European and associated countries.
