USING TRANSFER LEARNING TO IMPROVE ALGORITHMIC DEMENTIA CLASSIFICATION ACROSS DIFFERENT RACIAL AND ETHNIC GROUPS
2024

Improving Dementia Classification Using Transfer Learning

Sample size: 9018 publication Evidence: moderate

Author Information

Author(s): Leist Anja, Glymour M Maria, Langa Kenneth, Kim Jung Hyun

Primary Institution: University of Luxembourg

Hypothesis

Can transfer learning improve the accuracy of dementia classification across different racial and ethnic groups?

Conclusion

Transfer learning can enhance the accuracy of dementia classification for underrepresented groups, leading to better public health research.

Supporting Evidence

  • The transfer-learned algorithm had higher accuracy than the best previously reported algorithm for non-Hispanic Black participants.
  • The transfer-learned algorithm improved model calibration for Hispanic participants.

Takeaway

This study shows that using a special technique called transfer learning can help make better guesses about dementia in different groups of people.

Methodology

The study used data from two sources: a large dataset for initial algorithm development and a smaller dataset for refining the model.

Participant Demographics

Participants included non-Hispanic Black and Hispanic individuals.

Digital Object Identifier (DOI)

10.1093/geroni/igae098.4347

Want to read the original?

Access the complete publication on the publisher's website

View Original Publication