FROM TWINS TO DIGITAL TWINS: LLM-DRIVEN COGNITIVE PERFORMANCE PREDICTION
2024

Using Digital Twins to Predict Cognitive Performance

Sample size: 1040 publication Evidence: moderate

Author Information

Author(s): Zheng Anqing, Gustavson Daniel, Corley Robin, Reynolds Chandra

Primary Institution: University of Colorado Boulder

Hypothesis

Can Large Language Models accurately simulate cognitive performance by comparing predictions with actual performance of siblings?

Conclusion

The study suggests that LLM-based simulations may outperform traditional methods in predicting cognitive performance.

Supporting Evidence

  • The study uses a dataset of 1,040 individuals from 505 sibling pairs.
  • Preliminary analyses show strong cognitive performance correlations among monozygotic pairs.
  • The approach aims to develop personalized strategies in cognitive health management.

Takeaway

This study looks at how digital twins, which are like virtual versions of people, can help predict how well someone thinks and learns.

Methodology

The study uses the CATSLife dataset and intraclass correlation coefficients to evaluate cognitive performance similarities.

Participant Demographics

Individuals aged 28-49 from 505 sibling pairs, including twins and non-twin pairs.

Digital Object Identifier (DOI)

10.1093/geroni/igae098.4197

Want to read the original?

Access the complete publication on the publisher's website

View Original Publication