SVM classification of fNIRS activation patterns distinguishes RD with 60–90% accuracy, with left MTG channels the most consistent discriminative features¶

Claim · All claims
Evidence · 1 study (2 entries) · associational r2 · q2
Currency · newest study 2024 · not yet checked against its sources · not reviewed by a person

Subclaims¶

q2 i? LOOCV SVM classifiers using 48-channel fNIRS beta values achieved 60–90% accuracy classifying RD in children, adults, or all participants across speech tasks, with five classifications significantly above permutation-test chance. → Yang Fu 2024 q2 i? Left MTG channels appeared in the optimum feature set of all five significant classifications, and left premotor cortex, left SMA and left IFG had high weights in three classifications. → Yang Fu 2024 (2)

Evidence¶

Yang Fu 2024¶

Yang Fu, Xiaohui Yan, Jiaqi Mao, Haibin Su & Fan Cao. (2024). Abnormal brain activation during speech perception and production in children and adults with reading difficulty. npj Science of Learning. https://doi.org/10.1038/s41539-024-00266-2

q2 · i? · associational · r2

Machine-learning analysis with RBF-SVM, recursive feature elimination, and LOOCV on fNIRS beta values (1 × 48 feature vectors) from 87 participants; permutation testing (1000 iterations) evaluated significance, with printed p-values such as 0.002–0.031 for the significant classifications. No effect size is printed, so impact is null.

"Using LOOCV, the SVM classi fier yielded an accuracy ranging from 60% to 90% when classifying children, adults , or all participants across the two fNIRS tasks (Table 2)."

Yang Fu 2024 (2)¶

Yang Fu, Xiaohui Yan, Jiaqi Mao, Haibin Su & Fan Cao. (2024). Abnormal brain activation during speech perception and production in children and adults with reading difficulty. npj Science of Learning. https://doi.org/10.1038/s41539-024-00266-2

q2 · i? · associational · r2

Feature-weight analysis across cross-validation folds for the five classifications exceeding permutation tests. The quote reports left MTG channels appearing in all five classifications and left premotor, SMA, and IFG in three; channel 22 is anterior MTG and channel 24 posterior MTG.

"channels located in the left MTG were the most consistent region with high weights, because they appeared in allfive clas- sifications, and the left premotor cortex, left SMA, and the left IFG had high w e i g h t si n t h r e e c l a s s ifications."

Discussion¶

Related Claims¶

• RD readers show reduced deactivation of left MTG (channel 22) during Spanish compared to Chinese speech production — related
• Feedback types elicit distinct fNIRS activation patterns: frontopolar and middle temporal gyrus higher under metacognitive feedback, supramarginal gyrus under affective feedback, DLPFC under neutral feedback — related
• Familiarity recognition is represented by a distributed multivariate spatial-spectral pattern rather than localized single-feature shifts, necessitating multivariate machine learning — a broader claim this one bears on