Subcortical but not cortical activation-trajectory parameters classify BDNF genotype above chance in an exploratory decision-tree analysis¶

Claim · All claims
Evidence · 1 study (2 entries) · associational r1–r2 · q2
Currency · newest study 2024 · not yet checked against its sources · not reviewed by a person

Subclaims¶

q2 i? Decision-tree classification of genotype from subcortical parcel parameters was significantly better than chance (median = 0.82), but cortical-parcel classification was not (median = 0.67). → Lena S. Geiger 2024 q2 i? Cortical-parcel classification of genotype was not significantly better than chance. → Lena S. Geiger 2024 (2)

Evidence¶

Lena S. Geiger 2024¶

Lena S. Geiger, Torsten Wüstenberg, Zhenxiang Zang, Mirjam Melzer, Stephanie H. Witt, Marcella Rietschel, Markus M. Nöthen, Stefan Herms, Franziska Degenhardt, Andreas Meyer-Lindenberg & Carolin Moessnang. (2024). Longitudinal markers of cognitive procedural learning in fronto-striatal circuits and putative effects of a BDNF plasticity-related variant. npj Science of Learning. https://doi.org/10.1038/s41539-024-00282-2

q2 · i? · associational · r1

Exploratory decision-tree classification (2-fold random permutations cross-validation, 999 passes) on exponential-model feature triples from 11 subcortical parcels; accuracy median = 0.82 with 90% CI 0.63/0.92, described as "signific a n t l yb e t t e rt h a nchance level". No effect size printed.

"For the subcortical parcels, classification accuracy most heavily relied on differences in the interceptand amplitudeparameters and was signific a n t l yb e t t e rt h a nchance level (median = 0.82, lower/upper 90%CI boundary = 0.63/0.92)."

Lena S. Geiger 2024 (2)¶

Lena S. Geiger, Torsten Wüstenberg, Zhenxiang Zang, Mirjam Melzer, Stephanie H. Witt, Marcella Rietschel, Markus M. Nöthen, Stefan Herms, Franziska Degenhardt, Andreas Meyer-Lindenberg & Carolin Moessnang. (2024). Longitudinal markers of cognitive procedural learning in fronto-striatal circuits and putative effects of a BDNF plasticity-related variant. npj Science of Learning. https://doi.org/10.1038/s41539-024-00282-2

q2 · i? · associational · r2

The same decision-tree procedure on 12 cortical parcels yielded median accuracy 0.67 (90% CI 0.42/0.92) and "was not signi ficantly better than chance level". Equivalence was not tested; no effect size printed.

"For the cortical parcels, classi fication accuracy mainly relied on the tau and amplitude parameters and was not signi ficantly better than chance level (median = 0.67, lower/upper 90%CI boundary = 0.42/0.92; see Fig.1cf o raschematic depiction and Supplementary Fig. 1 for more details on the classification results)."

Discussion¶

Related Claims¶

• In the practiced condition, 23 brain parcels showed genotype differences in exponential model fit, mostly favoring the Met group and concentrated in frontal and striatal regions — related
• BDNF Val66Met genotype is associated with behavioral learning curves: Met-allele carriers show higher learning gain and, per the authors, less efficient early learning — related
• Repeated procedural training produces exponential decay of brain activation in a distributed set of regions including fronto-striatal circuits, paralleling exponential improvement in task performance — related
• Decision trees achieve competitive accuracy with black-box models (LR, RF, SVM), which performed slightly better but lack interpretability — related
• Group-average neural patterns in 160 of 300 cortical parcels contained decodable information about the three mechanical categories of the stimuli — related