Designing an effective semantic fluency test for early MCI diagnosis with machine learning.

Area of Science:

• Neuroscience
• Cognitive Psychology
• Machine Learning

Background:

• Semantic fluency tests are crucial for early Mild Cognitive Impairment (MCI) detection.
• Speech and semantic memory impairments are early MCI symptoms.
• Existing tests can be time-consuming, necessitating more efficient diagnostic tools.

Purpose of the Study:

• To identify the minimal yet effective combination of semantic categories and variables for MCI diagnosis.
• To reduce the duration of semantic fluency testing batteries.
• To maintain or enhance diagnostic accuracy while improving efficiency.

Main Methods:

• Utilized machine learning algorithms.
• Analyzed a database of 423 assessments from 141 subjects over three time points.
• Categorized subjects into Healthy, stable MCI, and heterogeneous MCI groups.

Main Results:

• The most efficient combination included the 'animals' and 'clothes' semantic categories.
• Key variables identified were 'corrects', 'switching', 'clustering', and 'total clusters'.
• This combination effectively distinguished between diagnostic groups.

Conclusions:

• A reduced set of semantic fluency tests (animals, clothes categories; corrects, switching, clustering, total clusters variables) offers an efficient diagnostic approach.
• This optimized combination balances diagnostic capability with time efficiency.
• Ideal for large-scale screenings and clinical settings prioritizing speed.