Semantic verbal fluency assessment using computational analysis in the Czech language

The semantic verbal fluency (SVF) task is a key method in neuropsychological assessment. A decline in SVF performance is common in mild cognitive impairment (MCI) and Alzheimer's disease (AD). While total word count is most often assessed, qualitative analysis can provide additional insights. However, traditional qualitative methods yield mixed results and have limitations. This study aims to assess the diagnostic potential of the computational method in the Czech context for patients with MCI and compare the novel approach with the traditional qualitative analysis. We analyzed the number of switches (NOS) and mean cluster size (MCS) in the vegetable and animal categories. In the animal category, the traditional and computational approaches showed poor diagnostic value for MCS. For NOS, the traditional approach was poor, while the computational approach showed fair diagnostic value. In the vegetable category, MCS had poor diagnostic value in both methods. For NOS, both the traditional (AUC = 0.761) and computational (AUC = 0.708) approaches showed fair diagnostic value. No significant differences were observed between the computational and traditional approaches across the measured indexes. Despite limitations such as group imbalances and difficulties handling multi-word phrases, computational analysis appears to be a promising tool for MCI assessment in Czech.