• Funding: National co-funding procedure for Bulgarian teams in approved COST Actions
  • Contract: КП-06-КОСТ/25/12.12.2025 г.
  • Principal Investigator: Assoc. Prof. Yordan Yordanov, PhD
  • Scientific field: Medical sciences
  • COST Action: CA24165 — Network for Cardiovascular Pharmacogenomics and Precision Medicine (CardioPharmaGENET)
  • Host organization: Medical University of Sofia
  • Project title: Pharmacogenomics applications of large language models and assessment of their impact on accessibility and readiness for integration in personalized cardiovascular pharmacotherapy
  • Duration: 24 months
  • Total budget: 50,000 BGN

Contact: yyordanov@pharmfac.mu-sofia.bg


Project summary

Pharmacogenomics (PGx) can improve therapy choice and dosing based on the patient’s genetic profile. This project tests whether chat-based interfaces powered by large language models (LLMs) can improve clinicians’ readiness to interpret and apply PGx information in cardiovascular pharmacotherapy.

A prototype intelligent agent will be developed and evaluated using anonymized example genomic data from public repositories. The project also includes regulatory, ethical, and socio-economic analyses to support safe and practical deployment.


Expected results

  1. Prototype agent integrating:
    • PGx interpretation software (e.g., PharmCAT or alternatives)
    • Open-access LLM models (e.g., biomedical models from Hugging Face)
  2. Documented internal and external testing results
  3. Evidence-based recommendations for implementation (including ethics, regulation, and pharmacoeconomics)
  4. Scientific outputs (publications / conference reports)

Work plan and stages (high level)

Stage 1 — Internal prototyping and controlled testing (months 1–12)

  • Tool and guideline review; selection of PGx resources
  • Infrastructure setup
  • Agent development + internal testing + improvements
  • Ethics and safety analysis based on pilot results

Stage 2 — External testing and impact analysis (months 12–24)

  • Methodology for external evaluation (questionnaires / interviews)
  • External testing with physicians/pharmacists
  • Socio-economic and policy impact analysis
  • Final contextual analysis and dissemination activities

Achieved results (to be updated)

  • Stage 1: In progress
  • Stage 2: Planned

(As you generate outputs, add them below with dates and links.)


Publications (Article 90 requirements)

Add publications here with links to journal pages and/or open-access copies.

  • (planned) Publication 1 — TBD
  • (planned) Publication 2 — TBD

Open data / resources

Add links to publicly accessible datasets and resources (where applicable).

  • Example genomic data: public repositories (e.g., 1000 Genomes) — link(s) to be added
  • Software / code repository — link(s) to be added

Intellectual property (if applicable)

  • Patent / IP filing: Not applicable / To be assessed

Potential applications

  • Support for clinicians to interpret PGx results faster and more consistently
  • Improved accessibility to precision pharmacotherapy decision support
  • Policy and implementation guidance for safe use of LLM-assisted PGx tools