Comparison of RCF Scoring System to Clinical Decision for the Rey Complex Figure Using Machine-Learning Algorithm.

Area of Science:

• Neuroscience
• Artificial Intelligence
• Medical Diagnostics

Background:

• Interpreting the Rey Complex Figure (RCF) relies on both scoring systems and clinical judgment.
• Clinical interpretation of RCF may lack accuracy in diagnosing mild cognitive impairment (MCI) or dementia compared to standardized scoring.
• A machine learning approach was employed to objectively compare these diagnostic methods.

Purpose of the Study:

• To evaluate the diagnostic accuracy of the RCF scoring system versus clinician-based interpretation for identifying cognitive impairment.
• To utilize machine learning to compare the efficacy of RCF scoring against clinical decision-making in detecting MCI and mild dementia.

Main Methods:

• A dataset of 2,232 subjects with neuropsychological assessments was utilized.
• Two datasets were created: normal vs. abnormal, and normal vs. MCI vs. mild dementia.
• Convolutional neural networks were trained, and receiver operating characteristic curves analyzed for model performance.

Main Results:

• Machine learning models achieved 96% accuracy (normal vs. abnormal) and 88% accuracy (normal vs. MCI vs. mild dementia).
• The model demonstrated 85% sensitivity (AUC 0.847) for detecting abnormalities and 78% sensitivity (AUC 0.778) for MCI/mild dementia detection.

Conclusions:

• The RCF scoring system shows potential for more accurate cognitive impairment diagnosis than clinical judgment alone.
• Standardized RCF scoring offers a more reliable method for distinguishing between normal cognition and various stages of cognitive decline.