Association analysis of food allergens

Abstract

Food allergy patients are known to present with allergic reactions to multiple allergens, but extrapolating these associations is difficult. Data mining, a procedure that analyzes characteristic combinations among large amounts of information, is often used to analyze and predict consumer purchasing behaviour. We applied this technique to the extrapolation of food allergen associations in allergy patients. We sent 1510 families our 'Questionnaire survey for the prevention of food allergies'. Responses noting 6549 allergens came from 878 families with 1383 patients, including 402 with anaphylaxis. Some results of the survey have already been published and here we presented the results of our association analysis of combinations of food allergens. Egg, milk, wheat, peanuts, and buckwheat are the most common food allergens. The most common simultaneous combinations of these allergens were 'egg-milk', 'egg-wheat', and 'milk-wheat'. The occurrence probability of a combination (i.e. one person suffering from a certain allergen also suffers from another) is called 'confidence'. Confidence was higher for 'chicken-egg', 'abalone-salmon eggs', and 'matsutake mushroom-milk'. As well, the combinations of 'crab-shrimp', 'squid-shrimp', and 'squid-crab' also indicated higher values in a statistical examination of the occurrence probabilities of these allergen combinations (Z-score). From the results of the association analysis, we speculated that some food allergens, such as abalone, orange, salmon, chicken, pork, matsutake mushroom, peach and apple did not independently induce food allergies. We also found that combinations, such as 'crab-shrimp', 'squid-shrimp', 'squid-crab', 'chicken-beef', and 'salmon-mackerel' had strong associations.