Descrição da proveniência de dados para extração de conhecimento em sistemas de informação de hemoterapia / Provenance Description to Extract Knowledge from Hemotherapy Information Systems

AUTOR(ES)
FONTE

IBICT - Instituto Brasileiro de Informação em Ciência e Tecnologia

DATA DE PUBLICAÇÃO

23/05/2012

RESUMO

The São Paulo Blood Center is responsible to maintain a database with information on each donation. However, this database does not have the quality required by techniques of analysis. For this reason, it is difficult to use it directly to establish systematic relationships between the variables. The main contribution of this paper is a provenance description of attributes selected using classification criteria defined by specialists. We show that it is possible to make detailed investigations using the data description without the need to change the structure of the database. During 1996 2006, 1,469,505 donors were responsible for more than 2.8 million of donation. After the provenance description, we obtained 252,301 male and 133,056 female that met our inclusion criteria. Of the 385,357 donors included in the analysis, 21,954(5.7%) were deferred due to low hematocrit, 3,850(1.5%) were males and 18,104(13.6%) were females. Our results show that, although the intervals between donations for female and male donors are wider, women presented anemia earlier than men. Approximately 12,84% of the females and 1,21% of the males would develop low hematocrit before the 7th donation. Our data suggest that individuals with low hematocrit level should wait longer before the next donation. Therefore, it is important to understand if there is a connection between blood donation and decrease in hematocrit level in order to prevent undesirable outcomes to blood donors. The provenance model presented here was not defined according to the generic provenance models already implemented. This thesis presents a provenance model that is able to add semantic information to acquire knowledge of an in silico experiment. One of the main purposes is to develop an approach based on declarations in order to answer biological questions. The provenance model described in this paper combines rich information for each process using the declarations, each having expert knowledge as a basis. To evaluate this provenance model we use descriptive statistics and Survival Analysis. Finally, with the validation of the model in a known domain, we intent to apply and validate this provenance model to other hemotherapy information systems.

ASSUNTO(S)

experimentos in silico in silico experiments proveniência de dados whole blood donations. anemia anemia blood donors data provenance doação de sangue total. doadores de sangue

ACESSO AO ARTIGO

Documentos Relacionados