Enhancing Dementia Classification for Diverse Groups: Transformer-Based Continuous Scoring of Clock Drawing Test
2024

Improving Dementia Classification with a New Clock Drawing Test Method

publication Evidence: moderate

Author Information

Author(s): Hu Mengyao, Murphey Yi, Qin Tian, Zahodne Laura, Gonzalez Richard, Freedman Vicki

Primary Institution: UTHealth at Houston, University of Michigan-Dearborn, University of Michigan

Hypothesis

Can a deep learning neural network improve the accuracy of dementia classification using the clock-drawing test?

Conclusion

The study found that continuous scoring from a deep learning model provides more precise thresholds for dementia classification than traditional ordinal scores.

Supporting Evidence

  • The clock-drawing test is a common tool for screening dementia.
  • Manual coding of the clock-drawing test can lead to errors and is time-consuming.
  • The deep learning model was trained on clock-drawing images to generate scores.
  • Continuous scores provided better thresholds for dementia classification than ordinal scores.
  • Demographic-specific thresholds were identified for different groups.

Takeaway

Researchers created a computer program to help doctors better understand if someone has dementia by looking at their clock drawings, and it works better for different types of people.

Methodology

Developed a deep learning neural network to automate clock-drawing test coding and compared its effectiveness to traditional scoring methods.

Participant Demographics

The study used a nationally representative sample of older adults, with specific thresholds identified for Black individuals, those with lower education, and those aged 90 or older.

Digital Object Identifier (DOI)

10.1093/geroni/igae098.4254

Want to read the original?

Access the complete publication on the publisher's website

View Original Publication