The main outcome of the BD2Decide project is a platform that has been conceived as a modular and flexible Service Oriented Architecture (SOA) platform, following the loose coupling principle design.
The platform will consist of the following five main components:
The Patient Documentation System, which provides seamless and secure implementation of two main scenario of usage
- Assist the treating care providers taking decisions in their clinical context and environment.
- Assist researchers and epidemiologist in discovering knowledge through the Big Data large scale infrastructure.
A big data infrastructure for multisource data analysis and extraction, supported by a knowledge management system able to homogenize and mine existing know-how from research, clinical trials, HNSCC patient’s health data collected in hospitals and epidemiology data from cancer registries, smoking habits and alcohol consumption and co-medication. The system comprises three components:
- The Big Data infrastructure to manage a multi scale types of data coming from the PDS and the external data-sources.
- A Knowledge Management System (KMS) to semantically manage and merge the information related to patients, tumors, context and references and to enable the reasoning on the data.
- The analytical tools that investigate the Patient’s data collected in hospitals, from clinical evaluation of patient’s health status, risk factors, diagnostic imaging (CT,MRI) using both radiomics and image analysis techniques to automatically calculate tumor-related measures, to population-related data.
- A library of personalized risk scoring and prognostic models integrating existing multiscale and simulation algorithms for HNSCC prognosis prediction and therapy effectiveness evaluation. The developed IT system adopts reasoning techniques to automatically select from this library the most adapted models for the individual patient and cancer sub-type, includes time dimensions to refine prognosis prediction and to score the most relevant prognostic factors.
Imaging and Radiomics feature extraction tools are used:
- To extract in semi automatic way anatomical (functional imaging) and biomolecular (radiomics) features for tumor and lymph-nodes precise evaluation (IFE).
- For tumor phenotypization (RFT). AlgorINTs for feature extraction and features selection according to their relevance are implemented (RFE) and tumor .
A highly interactive Knowledge Assisted Patient Digital Representation and Visualization suite for:
- The "virtual" presentation of patient- , disease- and population-related data ("Digital Patient" concept) relying on to support knowledge and data sharing, to foster objectivity in the overall decisional process and to facilitate the access to second opinions wherever and whenever useful.
- The patient-physician communications and co-treatment: an Individualised Patient Decision Aid (IPDA) aimed at patient-physician co-decision on treatment, in which previously validated prediction models for survival and treatment-related toxicity will be combined and integrated.