Blood typing for families: a novel hybrid human–computer application

Abstract

Crowdsourcing is a new approach which obtains information or input for a particular task by enlisting the services of the crowd. In recent crowdsourcing applications, the hybrid human–computer approach has been widely studied, to take advantage of both human beings and computers. In this paper, we propose a novel such application: blood typing for people in a family. We propose the BloodTyping method. It selects some members to take medical blood type tests, and to determine other family members' blood types, based on the inheritance rules. The aim is to reduce the number of medical tests, and thus, lower the cost. We extract rules for both +induction and − induction. The former is to predict children's blood types from parents', and the latter is to backward-induce a parent's blood type, given those of children and the other parent. Different combinations of blood types can induce different results: some may be an exact blood type, while others are composed of several blood types. Our method is optimised by conducting the cases which generate exact blood types first. The order is guided by extra-information via crowdsourcing, including the distribution of blood types with respect to the birthplace, and the personality, which may indicate some specific blood types. Taking a family with two parents and all children as a basic unit, the algorithm can be conducted simultaneously among different families in a decentralised way. The simulation results show that BloodTyping can significantly reduce the required number of blood tests.