Table 1. Classification modeling using the 261 MF biosignature list.

RF, random forest; LASSO, least absolute shrinkage and selection operator; MF; molecular feature.

Classification Model	Test-Set Sample Group	Number of Data Files*	RF (261 MFs)		LASSO (38/82 MFs±)
			Number Correctly Predicted	% Classification Accuracy	Number Correctly Predicted	% Classification Accuracy
1: Two-Way Model	Early Lyme Disease	60	58	97	59	98
	STARI	38	34	89	34	89
2: Three-Way Model	Early Lyme Disease	60	51	85	51	85
	STARI	38	35	92	35	92
	Healthy Controls	40	38	95	37	93

^*Samples were analyzed in duplicate by LC-MS.

^±A total of 38 MFs were selected by the LASSO model for two-way modeling and 82 MFs were selected by the LASSO for three-way modeling.