Table 2.

Comparison of models relating exposome and PFN topography to cognitive functioning across domains. We trained linear ridge regression models to predict three domains of cognition (General Cognition, Executive Function, and Learning/Memory). The first model type (“Exp-Factor”) used only a participant’s general exposome score, while the second model type (“PFN Topography”) used each participant’s multivariate pattern of PFN topography. The third model type (“Exp-Factor + PFN Topography”) used both exposome scores and PFN topography. Correlations between true cognitive performance and model-generated cognitive performance (r) were significant for all model types. Model comparison using the Akaike Information Criterion (AIC) and Bayesian Information Criterion (BIC) reveals that the Exp-Factor model is the most parsimonious.

	Discovery				Replication
Prediction Accuracy	r	p	AIC	BIC	r	p	AIC	BIC
General Cognition
Exp-Factor	0.42	2.65 ×10−148	-248.8198	-8.0250	0.46	1.17 ×10−179	-454.2790	-7.8809
PFN Topography	0.41	3.05 ×10−146	2.0198 ×106	8.2493 ×106	0.45	3.85 ×10−174	2.0196 ×106	8.2267 ×106
Exp-Factor + PFN Topography	0.44	2.44 ×10−166	2.0196 ×106	8.2491 ×106	0.48	3.30 ×10−194	2.0195 ×106	8.2270 ×106
Executive Function
Exp-Factor	0.11	8.86 ×10−11	1217.1946	-8.8570	0.14	7.30 ×10−16	1035.4933	-8.7450
PFN Topography	0.17	1.37 ×10−23	2.0210 ×106	8.2493 ×106	0.16	5.48 ×10−22	2.0209 ×106	8.2267 ×106
Exp-Factor + PFN Topography	0.17	4.41 ×10−24	2.0210 ×106	8.2491 ×106	0.17	8.54 ×10−23	2.0209 ×106	8.2270 ×106
Learning/Memory
Exp-Factor	0.25	1.35 ×10−50	470.3910	-8.4332	0.27	6.20 ×10−57	386.2874	-8.3685
PFN Topography	0.27	2.06 ×10−61	2.0204 ×106	8.2493 ×106	0.27	2.91 ×10−57	2.0204 ×106	8.2267 ×106
Exp-Factor + PFN Topography	0.28	3.49 ×10−66	2.0203 ×106	8.2491 ×106	0.28	4.92 ×10−63	2.0203 ×106	8.2270 ×106