Published On: October 2, 2026

Secondment: Christine Kakalou

 

 

Institution of origin

Institute of Applied Biosciences (INAB), Centre for Research & Technology Hellas (CERTH)

Host institution

Nostrum Biodiscovery

Initial objective

Explore the use of Large Language Models to simulate patient–healthcare professional interactions and investigate how psychological characteristics can be represented and evaluated in these simulations.

Results achieved

During her secondment at Nostrum Biodiscovery in July 2026, Christina Asimina Kakalou contributed to a three-team collaboration between Nostrum Biodiscovery, the Barcelona Supercomputing Center (BSC) and the INAB|CERTH PsyLab team. The work built on simulation activities led by BSC, while bringing together technical, healthcare and psychological expertise from across the three teams.

The work focused on simulated patient profiles based on established health psychology frameworks, including illness perceptions and beliefs about medicines, and on testing whether these predefined characteristics were reflected in LLM-generated patient–healthcare professional conversations.

The results showed that the simulations could reproduce several of the intended profile differences, while also highlighting cases where characteristics were less clearly or consistently expressed. The work also explored more specialised evaluation metrics derived from the simulated interactions, including indicators related to medication adherence risk, to examine whether the simulations could support assessment beyond general profile consistency.

Overall, the secondment helped establish a shared methodology for combining technical simulation with psychological validation and more domain-specific behavioural indicators. This creates a basis for testing the framework in more complex contexts where patient and healthcare professional perspectives may differ in meaningful ways, including future work on women’s health scenarios.

 

Visit to MareNostrum 5 during the stay at Nostrum Biodiscovery