GENAIHEALTH: Generative AI for better healthcare decisions

In the context of the project, we will design and develop a unified architecture supporting virtual assistants for health professionals and patients, with a focus on clinical decision support and medication management. Multimodal AI models will be trained onto health data and then put into use for 3 distinct use cases. Their assessment will prove their utility in currently unmet needs.

Objectives

1. Technical

• Develop an AI-driven medication management system leveraging Natural Language Processing (NLP) to analyze authoritative medical documents and patient records.

• Utilize distillation learning to enhance model efficiency over multiple modalities while ensuring accuracy and reliability.

• Implement Retrieval-Augmented Generation (RAG) techniques using PubMed and other trusted medical knowledge bases to enhance AI-generated recommendations.

• Develop methods for AI-generated clinical insights to be interpretable, transparent, and aligned with ethical healthcare principles.

• Develop an easy-to-use patient-facing application integrating voice-enabled chatbot support and medication management functionalities.

2. Standardization and cross-country adoption

• Engage stakeholders, including regulatory bodies and Health Technology Assessment (HTA) agencies, to align AI solutions with existing legal and ethical frameworks.

• Establish methodologies to ensure cross-border applicability and acceptance of AI-powered virtual assistants in different healthcare systems.

3. Demonstration and validation

• Implement and assess the solution in at least three real-world healthcare use cases, demonstrating its clinical utility and effectiveness. 

• One use case will focus on geriatrics, addressing challenges such as polypharmacy, multimorbidity, and cognitive impairments, ensuring AI solutions improve care coordination and patient safety.

• The other use cases will target relevant medical fields, such as, but not limited to, psychiatrics, cancer, etc.

Expected Results

For HCPs: improved decisions, enabled by AI-generated recommendations and faster access to patient-specific factual information and relevant literature.

For Patients: Improved safety and better clinical outcomes, improved adherence to treatments, AI-driven support tailored to individual patient needs.

For Healthcare Systems: Improved care coordination, reduced workload, enhanced patient outcomes, supporting healthcare sustainability and equitable access.

For Research and Academia: Established methodologies for integrating AI tools in healthcare, supporting regulatory approval, trust, and large-scale adoption.

Ergobyte’s Competencies

• AI-ready knowledge base of 30.000 SmPCs (medicine marketing authorizations)

• Clinical decision support expertise, using both deterministic and stochastic methods

• Access to 70.000 HCPs through Galinos.gr, 42M page views in 2024

• Team of domain experts: medicine, pharmacology, biology

Known Partners

• White Research (BE)

• CERTH ITI & ΙΝΑΒ - Research institutes (GR)

• HealThink (GR)

Required Expertise

• Development of patient-facing applications

• Pilot / demonstration sites for additional use cases

• GenAI expertise with modalities other than text