A machine-learning algorithm for predicting brain age using Rey-Osterrieth complex figure tests of healthy...

Area of Science:

• Neuroscience
• Artificial Intelligence
• Gerontology

Background:

• The Rey-Osterrieth Complex Figure Test (RCFT) is a standard tool for assessing cognitive function and constructional ability in neuropsychology.
• Age is a primary factor influencing cognitive performance and brain aging.

Purpose of the Study:

• To investigate a supervised machine learning (ML) algorithm for predicting brain age gap using RCFT drawings.
• To explore the potential of ML-based brain age gap prediction for early dementia detection in healthy elderly individuals.

Main Methods:

• Collected RCFT drawings and demographic data (age, gender, education) from 1,970 healthy Korean adults (ages 45-90).
• Trained a Convolutional Neural Network (CNN) regression model using RCFT copies, recall scores, and education level.
• Utilized Keras with Tensorflow for model development.

Main Results:

• The CNN regression model achieved a Mean Absolute Error (MAE) of 7.2 years and a Root Mean Squared Error (RMSE) of 8.9 years in predicting brain age gap.
• Performance was evaluated on a test dataset of 300 healthy subjects.

Conclusions:

• The developed ML model demonstrates potential as a biomarker for individual brain aging.
• This approach offers a cost-effective method for the early detection of dementia.