Crowdsourced Identification of Possible Allergy-Associated Factors: Automated Hypothesis Generation and Validation Using Crowdsourcing Services

Background Hypothesis generation is an essential task for clinical research, and it can require years of research experience to formulate a meaningful hypothesis. Recent studies have endeavored to apply crowdsourcing to generate novel hypotheses for research. In this study, we apply crowdsourcing to explore previously unknown allergy-associated factors. Objective In this study, we aimed to collect and test hypotheses of unknown allergy-associated factors using a crowdsourcing service. Methods Using a series of questionnaires, we asked crowdsourcing participants to provide hypotheses on associated factors for seven different allergies, and validated the candidate hypotheses with odds ratios calculated for each associated factor. We repeated this abductive validation process to identify a set of reliable hypotheses. Results We obtained two primary findings: (1) crowdsourcing showed that 8 of the 13 known hypothesized allergy risks were statically significant; and (2) among the total of 157 hypotheses generated by the crowdsourcing service, 75 hypotheses were statistically significant allergy-associated factors, comprising the 8 known risks and 53 previously unknown allergy-associated factors. These findings suggest that there are still many topics to be examined in future allergy studies. Conclusions Crowdsourcing generated new hypotheses on allergy-associated factors. In the near future, clinical trials should be conducted to validate the hypotheses generated in this study.