Data Science Platform

Introduction

The Data Science Platform is a cross-functional platform designed to support research, healthcare management, and clinical decision-making using big data and artificial intelligence (AI). The platform integrates information from outpatient and inpatient care for the population of the Valencia Clínico-Malvarrosa Health Department, which serves approximately 350,000 residents. Through predictive analytics techniques, AI models, and the development of telemedicine interfaces, the platform makes it possible to identify inefficiencies in operational and clinical workflows, optimize care processes, and promote the implementation of 5P Medicine (predictive, preventive, personalized, population-based, and participatory).

The Data Science Platform is based on the INTEGRA-CLOUD Platform, developed at INCLIVA with funding awarded directly to the institution by the Ministry of Economic Affairs and Digital Transformation through the UNICO R&D Cloud Program, as part of the European Recovery, Transformation, and Resilience Plan in 2023.

Services

• Scientific and Technical Consulting:
  • Consulting on study protocol initiation, administrative procedures, and CEIm.
  • Data protection consulting.
  • We offer guidance on selecting appropriate analysis tools and artificial intelligence algorithms for processing health data.

• Procedures with Health Authorities: operational management and processing of permits, licenses, and agreements with public agencies and health authorities for the use of clinical data in research projects.

• Data Analysis and Processing:
  • Processing of clinical data from its collection on the platform through its conversion into homogeneous, organized structures suitable for analysis. This includes data cleaning techniques, outlier removal, imputation of missing data, and standardization of formats to facilitate interoperability with other databases and analytical tools.
  • Access to structured and validated datasets using interoperability standards such as HL7 FHIR or OMOP to facilitate integration with other systems and databases.
  • Data analysis from the Platform through the creation and monitoring of patient cohorts: definition, structuring, and maintenance of patient cohorts for epidemiological and clinical studies. We ensure data traceability, longitudinal patient follow-up, and the application of consistent inclusion and exclusion criteria. We offer the ability to link clinical data with information from other health registries to expand the analysis.
  • Development of artificial intelligence models for secondary uses.
  • Collaboration on and/or development of clinical trials.

• Participation in national and European projects requiring the use of the database.
  • Training and education.

Equipment

On-premises cluster designed to create a data lake capable of storing and processing data from the Regional Ministry of Health’s corporate applications. This infrastructure is connected to the Arterias Network (the Regional Ministry of Health’s corporate network):

Two high-performance servers, each equipped with:

• Processors: 48 cores / 96 processing threads.
• Memory: 2 TB of RAM.
• SSD storage in a RAID1 configuration for the hypervisor and base software.

Each server includes an NVIDIA GPU. Ampere A30 architecture with support for CUDA and GPU.

This cluster is connected to three storage arrays, each designed to meet different specific requirements:

• SAN Storage: 18 TB on NVMe drives, configured as iSCSI volumes. Designed to ensure high availability and maximum performance.
• NAS Storage:
  • High-performance, horizontally scalable array: 216 TB combining SSDs and HDDs. It supports secondary disks, NFS, S3, and Data Lake capabilities.
  • Conventional array: 95 TB primarily intended for backups, providing reliable storage optimized for backup and recovery tasks.

In addition, connected to Rediris (the Spanish academic and research network), there are two machines capable of hosting applications and analysis environments outside the Arterias Network.

Each machine has: 24 cores/48 threads, 1.5 TB of RAM, 24 TB of HDD (in RAID6) for moderate-speed bulk storage, and 5 TB of SSD (in RAID5) for system disks and fast access

This entire infrastructure is equipped with the highest security measures and is highly scalable and robust thanks to the high redundancy of its most critical components.

These resources are complemented and can be enhanced by the ability to use the HPC computing resources shared by the foundation’s platforms and research groups, which are managed by INCLIVA’s IT Unit.