AICCELERATE

Healthcare systems lack flexible AI solutions that allow hospitals to improve efficiency and the quality of patient care. Current solutions provide limited scalability and are confined to isolated applications. Scalable models that address data sharing, integration, privacy, and ethics are needed to ensure better adoption of AI in healthcare. The AICCELERATE project introduces an approach for scaling up AI-enabled digital solutions for different hospital use cases. AICCELERATE will develop partners’ existing digital solutions further to enable the development of a Smart Hospital Care Pathway (SHCP) Engine. This engine serves as a toolset for AI models and robotics to improve the quality of care and health outcomes. It will also enable lean management and effective decision-making. These tools are tested in three pilots that (will) provide feedback for improving the SHCP Engine: (i) patient flow management for ER and surgical units, (ii) digital care pathway for Parkinson’s disease, and (iii) paediatric service delivery. AICCELERATE provides an adaptable model for varied clinical use cases to enhance patient-centric digital care pathways and optimize patient flow management. Patient empowerment and evidence-based trust towards AI is a key part of the project.

One of the main objectives of AICCELERATE is to provide a data preparation pipeline which can be directly connected to the EHRs of hospital settings and extract datasets or prepare data for machine learning pipelines to use for preparing training or validation datasets or features for an instance (e.g., a Patient) for online prediction.

Reason for applying to HSbooster.eu services

The standardization of the methodology and definitions to build an interoperable data preparation pipeline for health research and ML use cases will enhance the reusability and scalability of AI and analytics research on health data performed in one setting to other settings. This also is very in line with the FAIR principles which have become very popular recently. Being part of the standardization process will provide great benefits for the AICCELERATE project to disseminate its results in Europe and beyond Europe and can provide exploitation opportunities. 

Main standardisation interest

AICCELERATE provides a methodology and declarative languages to define this pipeline in a declarative way with rich metadata where these definitions can be shared and executed in any hospital setting to prepare the same dataset (AICCELERATE also provides the tools to execute these definitions and the intended pipeline).

Within this language, specifications from HL7 FHIR standard (i.e., ‘FHIR resources and profiling’ to convert the custom EHR content into HL7 FHIR resources as a common data model for the use case, ‘FHIR REST API’ to access FHIR resources for feature and outcome variable extraction, ‘FHIR Path’ to process or navigate on data content) are used already. The methodology and these languages can be a good addition to the set of specifications of the HL7 FHIR standard.