LLM-Based Human Digital Twins for Research and Decision Making
- Partner
- Research focus
- Period
- Feb 2025 – PresentOngoing
- Focus
- Human Digital Twins
Key outcomeTwinlyses, a working two-sided research prototype: a contributor app and a research workspace
Context
A research line on human digital twins that represent specific, consenting individuals: grounded in their own evidence, governed by their consent, and honest about what they cannot know. It began with SozioMimic, continued with MindsetTwin, and is built out as a working prototype called Twinlyses.
Inputs
- Qualitative interviews, diary studies, and longitudinal surveys
- Voluntarily contributed preferences, values, and reflections
- Consent and purpose settings defined by each contributor
Approach
- SozioMimic: representing individuals from their own qualitative data rather than generic personas
- MindsetTwin: modelling reasoning, memory, and consistency across repeated interactions
- Twinlyses prototype: consent-governed twins that show their evidence and confidence, and abstain when evidence is insufficient
Outcomes
- Twinlyses, a working two-sided research prototype: a contributor app and a research workspace
- Study results that state model confidence, evidence limits, and a suggested next validation step