Barcelona Supercomputing Center (BSC) participates in the AI-SPRINT project contributing its experience on the programming and parallelization of applications on distributed infrastructures. The work will be organized in two main contributions, the deployment of COMPSs on the edge devices considered by the use cases and the implementation of the AI applications to be executed across distributed heterogeneous infrastructures (on premise, edge and public clouds). In particular, AI-SPRINT will benefit from the recent developments on the adaptation of COMPSs for Fog to Cloud platforms and will extend it to support the execution of serverless functions as a service.
On the other side, COMPSs will be adopted to develop AI and big data applications in support of the use cases also leveraging the recent enhancements to develop workflows that combine HPC compute engines with High Performance Data Analytics (HPDA) and machine learning methods. These ML implementations are available through the dislib library that is also part of the Fujitsu-BSC collaboration.
The technology developed within the project will be put to test by BSC on a personalized healthcare use case that will focus on privacy and security, much needed in healthcare scenarios since the information to be exchanged and processed involves medical data about patients.
More specifically, an automated system for personalized stroke risk assessment and prevention will be developed by using continuous, non-invasive monitoring of heart activity. The process will gather heart parameters collected from a wearable device, patient’s lifestyle information and biochemical blood indicators from a mobile application. All data will be anonymized, processed and used to train AI models cooperatively by local edge servers and cloud. At the same time, it will provide personalized notifications, alerts, and recommendations for stroke prevention.
“AI-SPRINT defines a novel framework for the design and operation of AI applications in computing continua leveraging the COMPSs programming framework and supporting AI applications development by enabling the seamless design and partition of AI applications among the plethora of cloud-based solutions and AI-based sensor devices. Moreover it will generate impacts bringing together different European industrial end-users while and making available the software tools through a marketplace for AI start-ups, SMEs, system integrators, and European cloud providers” states Daniele Lezzi, Senior Researcher in the Computer Sciences department - Workflows and Distributed Computing at BSC.
COMPSs is a task-based programming model known for notably improving the performance of large-scale applications by automatically parallelizing their execution. The COMPSs runtime has been recently extended within BSC projects: CLASS and ELASTIC to manage distribution, parallelism and heterogeneity in the edge resources transparently to the application programmer and to handle data regardless of persistency by supporting a single and unified data model. COMPSs is the base of the Design Tools of the project and it will support developers to easily compose AI/ML applications also leveraging the dislib library, helping end users to deal with big datasets on distributed resources and providing automatic parallelization of the code.
About Personalized Healthcare:
AI-SPRINT applications will pave the way for an effective framework for personalized AI models preventing risks coupled with a lifestyle modification modification programme benefiting people aged between 40 and 80, improving and extending human lives. The project addresses the United Nations strategic development goals SDG3 (Good Health and Well Being) through the personalized healthcare pilot.
AI-SPRINT will tackle the skill shortage and considerably reduce steep learning curves in the development of AI software on edge ecosystems through OSS (Operations Support System). The project addresses the following United Nations strategic development goals (SDGs): SDG8 (Decent Work and Economic Growth) enabling novel AI applications running in computing continua, SDG9 (Industry, Innovation and Infrastructure) by fostering innovation in the maintenance and inspection use case and contributing OSS and SDG12 (Ensure sustainable consumption and production patterns) through farming 4.0 pilot.
For further information, visit AI-SPRINT website: https://www.ai-sprint-project.eu/
This article was first published on 2 February by BSC.