Concept and Approach

ONCORELIEF aims to deliver a framework that consists of three main sub-systems:

  1. a back-end data platform where data are securely collected from heterogeneous sources, anonymised, annotated and stored, etc.,
  2. an AI engine built on top of the back-end platform, which consumes and analyses data, extracts important features, produces meaningful AI models and updates them accordingly, produces correlations, etc.,
  3. a downloadable application (ONCORELIEF Guardian Angel) available for portable devices, which will be connected with the ONCORELIEF platform and with patients’ sensing devices.

It runs locally and uses models produced by the AI engine to extract insights on the patient life and condition and make suggestions. The GA may optionally send data back both for research-related reasons (another source of big data) and for computation offloading reasons (AI engine not powerful enough on the mobile phone). These models are then downloaded to personal, portable devices in order to locally analyse individual data and to generate QoL (Quality of Life) indices, recommendations and warnings for the patients using the application.


The Guarding Angel application will contain a local, decentralised AI engine, regularly updated by the central ONCORELIEF platform AI engine, where all the big data are processed and analysed. These updates will be also related to the specific user’s needs, condition and preferences, as the embedded AI engine will be designed to initially contain a minimum set of features and functions. In this way, the application (via its local embedded analytics platform) will be personalised while maintaining an affordable size for portable devices. This architecture enables decentralisation for privacy-respect, data-security, ethical and user-friendliness purposes. The app interface will allow patients to have full access to all their medical and health-related data, to input the ones needed to complete their medical profiles and to set-up or confirm whom of their related ones they would like to be notified in case of emergency. If they wish so, the app can be used as a bridging point in between patients and health professional(s), as the latter will be able (only upon the patient’s confirmation or request) to get regularly updated with indicative indexes or information regarding the patient’s progress.

The back-end data platform is informed by the anonymised data from patients, healthcare professionals, caretakers, health institutions and organisations and also from existing local databases. All data is transferred, stored and processed securely and in accordance with the GDPR regulations. The proposed architectural approach for detaching the design, development, execution and exploitation of HPC empowered big data analysis processes is depicted in the following figure. This approach aims at ensuring interoperability among all involved components and stakeholders, emphasising on security, privacy respect, but also user friendliness and customisation. In order to achieve this, ONCORELIEF consortium aims at designing and delivering 1) standardised interfaces such as smart algorithms and AI elements of self-learning so that reusability of core system resources is supported and 2) appropriate data models and communication protocols, allowing heterogeneous and sensitive information to be exchanged between the various components in a harmonic and secure way.

In a nutshell, big data coming from various sources and stakeholders are made available through HPC powered repositories. Prior to the check-in and storage data pass through a data quality check pipeline in order to address the veracity and timeliness challenges associated with big data. From the moment that data are collected in the ONCORELIEF platform, quality checks are performed to discover inconsistencies and other anomalies in the data and eventually ensure their integrity and completeness, to be followed by a number of steps associated with data cleansing starting from filtering and ending with the normalisation stage. The next step is the selection of an analysis template among a set of available templates, each one of which represents a specific algorithm with the associated software and execution endpoint and will provide to the platform the flexibility to adjust the relevant configuration parameters, including input, executing parameters and parameters associated with networking and computing resources constraints, as well as output parameters. Upon the execution of an analysis template, the outcome could constitute the input for other analysis template. The output of the analysis template execution will be a session object that contains all output values in memory.

After input datasets have been selected, the advanced analytics on top of big data need to be executed. In the case of ONCORELIEF, advanced analytics algorithms will be entailed, providing the platform with the ability to visually explore the different kinds of data, while discovering and addressing new patterns. Machine learning and predictive modelling techniques will be updated in order to be able to manage the predictive life cycle of data preparation, exploration and analysis, for achieving better deployment and monitoring. 

The designed analytics workflows will be then executed on HPC and Big Data frameworks in the central ONCORELIEF AI Engine. The execution results will be then generating and informing accordingly the AI models and algorithms that derive from the patterns identified in the previous stage of the execution. These models will contain all the necessary information for the exploitation of the patients’ data, as these result from their sensing and personal devices.
Along with the ONCORELIEF Guarding Angel, the required AI algorithms will be also downloaded so that the user’s data can be locally processed and analysed using the AI Engine’s data models. The users’ data will then be processed and become anonymised (upon user’s consensus) and will be uploaded to the platform’s core nutshell in a secure way.

This section describes the five (5) main technology components of ONCORELIEF, which are also portrayed in the overall architectural view featuring all interaction points in the picture bellow. In particular, 

  1. The ONCORELIEF Guardian Angel. an application, compatible with portable devices operating with iOS and Android that will be built upon the AI models generated by the ONCORELIEF platform. 
  2. The ONCORELIEF Back-end and data platform. This is where all the data is collected, processed and utilized to deliver the services of the ONCORELIEF platform.
  3. The ONCORELIEF Big Data and AI Engine. This resides on top of the back-end platform and exploits the insight derived from the analysis of the data as well as it is responsible for the empathic interaction with the users of the ONCORELIEF Guardian Angel.
  4. The ONCORELIEF wellbeing and QoL Index. The unique number which is used to monitor the wellbeing status and progress of the different users.
  5. The ONCORELIEF Sensing Framework. The unobtrusive and pervasive, wearable and embedded sensing ecosystem of the patient for data collection.