Detecting mild cognitive impairment by applying integrated random forest to finger tapping.

Area of Science:

• Neurology
• Biomedical Engineering
• Gerontology

Background:

• Early detection of dementia is vital for maintaining quality of life and managing healthcare costs in aging populations.
• Mild cognitive impairment (MCI) is a precursor to dementia, necessitating effective screening tools.
• Current screening methods may lack simplicity or accuracy for widespread use.

Purpose of the Study:

• To develop a simple and accurate screening test for mild cognitive impairment (MCI).
• To utilize finger-tapping measurement as a novel diagnostic tool for MCI.
• To enhance classification accuracy for MCI detection using an improved machine learning approach.

Main Methods:

• Extracted 248 features from finger-tapping waveforms of 182 MCI patients and 352 healthy controls.
• Applied five conventional classification methods and an improved Random Forest (RF) method (Integrated RF).
• The Integrated RF model combined MCI vs. normal and Alzheimer's disease vs. normal classification models.

Main Results:

• The proposed Integrated RF method achieved the highest discrimination accuracy.
• Achieved an F1-score of 0.795, with a recall of 0.778 and precision of 0.814.
• Demonstrated superior performance compared to conventional classification methods.

Conclusions:

• Finger-tapping measurement shows significant potential as a highly accurate screening tool for MCI.
• This method offers a simple, non-invasive approach for early detection of cognitive decline.
• Further validation could lead to widespread clinical application for dementia prevention strategies.