Building a Human Digital Twin (HDTwin) Using Large Language Models for Cognitive Diagnosis: Algorithm Development and Validation
2024

Building a Human Digital Twin for Cognitive Diagnosis

Sample size: 124 publication 10 minutes Evidence: high

Author Information

Author(s): Mavragani Amaryllis, Rocha Heber L, Tziritas Nikos, Sprint Gina PhD, Schmitter-Edgecombe Maureen PhD, Cook Diane PhD

Primary Institution: Washington State University

Hypothesis

Can a human digital twin (HDTwin) effectively unify diverse cognitive health data to improve diagnostic accuracy?

Conclusion

HDTwin enhances the accuracy and explainability of cognitive diagnoses by integrating diverse data sources.

Supporting Evidence

  • HDTwin achieved a peak accuracy of 0.81, significantly outperforming baseline classifiers.
  • On average, HDTwin yielded accuracy=0.77, precision=0.88, and recall=0.63.
  • The approach shows promise for improving early detection and intervention strategies in cognitive health.
  • Participants were categorized as cognitively healthy older adults (60.5%) or older adults with MCI (39.5%).

Takeaway

This study created a digital twin that helps doctors understand and diagnose cognitive health better by using lots of different information about a person.

Methodology

HDTwin integrates cognitive health data from multiple sources and uses a large language model to predict cognitive diagnoses.

Potential Biases

Nondeterminism of LLMs may affect reproducibility of results.

Limitations

The study faced limitations such as missing data, reliance on LLM for response extraction, and a small sample size for training.

Participant Demographics

Participants were independent-living older adults, mean age 70.48 years, 71.78% female.

Statistical Information

P-Value

p<0.001

Statistical Significance

p<0.001

Digital Object Identifier (DOI)

10.2196/63866

Want to read the original?

Access the complete publication on the publisher's website

View Original Publication