TY - GEN
T1 - Using HMM, Association Rule Mining and Ensemble Methods with the Application of Latent Factor Model to Detect Gestational Diabetes Mellitus
AU - Shetty, Jayashree S.
AU - Shetty, Nisha P.
AU - Das, Vedant Rishi
AU - Vaibhav,
AU - Olivia, Diana
N1 - Publisher Copyright:
© 2023, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.
PY - 2023
Y1 - 2023
N2 - Gestational diabetes mellitus (GDM) is a condition often seen during pregnancies in which a hormone made by the placenta prevents the body from using insulin effectively. Women with GDM are at an increased risk of complications during pregnancy and during delivery. The offspring and the mother are also at an increased risk of getting diabetes in the future. Therefore, careful screening is necessary to avoid further complications. The objective of this research is to facilitate proper prediction of the presence of GDM in women so that timely intervention can help prevent future adversities. Multiple machine learning algorithms with data analysis methods are employed to investigate the probability of GDM and reach an optimal solution. The methodology makes use of the latent factor model and stochastic gradient descent to account for the missing data. Information entropy is used to calculate the amount of information each variable presents. The final classification is done and compared using three methods. These include ensemble method, hidden Markov model, and association analysis. Experiments reveal that the ensemble method involving decision trees, k-nearest neighbors, and logistic regression with weighted averaging delivers promising performance. Test data accuracy of 80% was recorded on the ensemble method.
AB - Gestational diabetes mellitus (GDM) is a condition often seen during pregnancies in which a hormone made by the placenta prevents the body from using insulin effectively. Women with GDM are at an increased risk of complications during pregnancy and during delivery. The offspring and the mother are also at an increased risk of getting diabetes in the future. Therefore, careful screening is necessary to avoid further complications. The objective of this research is to facilitate proper prediction of the presence of GDM in women so that timely intervention can help prevent future adversities. Multiple machine learning algorithms with data analysis methods are employed to investigate the probability of GDM and reach an optimal solution. The methodology makes use of the latent factor model and stochastic gradient descent to account for the missing data. Information entropy is used to calculate the amount of information each variable presents. The final classification is done and compared using three methods. These include ensemble method, hidden Markov model, and association analysis. Experiments reveal that the ensemble method involving decision trees, k-nearest neighbors, and logistic regression with weighted averaging delivers promising performance. Test data accuracy of 80% was recorded on the ensemble method.
UR - http://www.scopus.com/inward/record.url?scp=85140446322&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85140446322&partnerID=8YFLogxK
U2 - 10.1007/978-981-19-4193-1_20
DO - 10.1007/978-981-19-4193-1_20
M3 - Conference contribution
AN - SCOPUS:85140446322
SN - 9789811941924
T3 - Lecture Notes in Networks and Systems
SP - 215
EP - 223
BT - Emerging Technologies in Data Mining and Information Security - Proceedings of IEMIS 2022
A2 - Dutta, Paramartha
A2 - Chakrabarti, Satyajit
A2 - Bhattacharya, Abhishek
A2 - Dutta, Soumi
A2 - Piuri, Vincenzo
PB - Springer Science and Business Media Deutschland GmbH
T2 - 3rd International Conference on Emerging Technologies in Data Mining and Information Security, IEMIS 2022
Y2 - 23 February 2022 through 25 February 2022
ER -