This paper investigates the automated classication of oral food challenges (`allergy tests'). The electrocardiograms (ECG) of the subjects being tested for allergies were recorded via a wireless mote, and the QRS complexes were manually annotated. 18 features were extracted from the manually annotated QRS points. Principal component analysis dimensionality reduction and Gaussian mixture models were used to generate subject independent log likelihood plots, which were further classied with secondary subject independent and subject adaptive thresholding. The platform resulted in 87% accuracy of classication with 100% specicity. The algorithm presented can detect allergy up to 30 minutes sooner than the current state of the clinical art allergy detection (7minutes plus or minus 9). The allergy detection algorithm results in robust detection rejecting likelihood changes in three subject `outliers' correctly tagged as `pass.'