Using classification techniques for statistical analysis of Anemia

Highlights：

• Anemia in children is becoming a worldwide problem owing to the unawareness among people regarding the disease, its causes and preventive measures.

• This study develops a decision support system using data mining techniques that are applied to a database containing data about nutritional factors for children. The data set was taken from NFHS-4, a survey conducted by the Government of India in 2015-16.

• The work attempts to predict anemia among children and establish a relation between mother’s health and diet during pregnancy and its effects on anemic status of her child. It aims to help parents and clinicians to understand the influence of an infant’s feeding practices and diet on his/her health and provide guidelines regarding diet in order to prevent anemia.

• The two techniques, decision tree and association rule mining has been applied and compared to select more appropriate technique for this task and a model is proposed in the healthcare domain with the aim to reduce the risk of the blood-related disease anemia. A statistical analysis has been provided about Anemia in children with respect to various factors like type of residence, state-wise, gender-wise etc.

• The results are presented in the form of rules with count and overall accuracy for rules from decision trees and Support, Confidence, Lift and Count for rules from association rules. Some corollaries are suggested to Clinicians, parents and government based on the rules that should be undertaken so as to reduce the risk of anemia in children.

摘要

•Anemia in children is becoming a worldwide problem owing to the unawareness among people regarding the disease, its causes and preventive measures.•This study develops a decision support system using data mining techniques that are applied to a database containing data about nutritional factors for children. The data set was taken from NFHS-4, a survey conducted by the Government of India in 2015-16.•The work attempts to predict anemia among children and establish a relation between mother’s health and diet during pregnancy and its effects on anemic status of her child. It aims to help parents and clinicians to understand the influence of an infant’s feeding practices and diet on his/her health and provide guidelines regarding diet in order to prevent anemia.•The two techniques, decision tree and association rule mining has been applied and compared to select more appropriate technique for this task and a model is proposed in the healthcare domain with the aim to reduce the risk of the blood-related disease anemia. A statistical analysis has been provided about Anemia in children with respect to various factors like type of residence, state-wise, gender-wise etc.•The results are presented in the form of rules with count and overall accuracy for rules from decision trees and Support, Confidence, Lift and Count for rules from association rules. Some corollaries are suggested to Clinicians, parents and government based on the rules that should be undertaken so as to reduce the risk of anemia in children.