TY - GEN
T1 - Weka machine learning classification in identifying autonomic dysfunction parameters associated with ACE insertion/deletion genotypes
AU - Ng, Ethan
AU - Hambly, Brett
AU - Matthews, Slade
AU - McLachlan, Craig S.
AU - Jelinek, Herbert F.
PY - 2012/7/16
Y1 - 2012/7/16
N2 - This study was designed to investigate parameters of autonomic dysfunction that may be under the influence of ACE ID genotypes. 136 patients with (47) and without type II diabetes were genotyped. Biomarkers such as HbAlc and eGFR, blood pressure, blood cholesterol are in part regulated by the autonomic nervous system and heart rate variability is an indicator of autonomic balance between the sympathetic and parasympathetic division. Several statistical methods were used, including the J48 decision tree machine learning algorithm to associate parameters of autonomic dysfunction and other biomarkers with ACE genotype. Non-parametric and machine learning methods detected more variables, which were able to contribute to classification of patients into genotypes. We found that HbAlc and TC:HDL were important nodes for separation of ACE genotype classes when the J48 decision tree algorithm was used. These were also verified by the Mann-Whitney analysis. Parametric comparisons of normally distributed variables revealed that only HDL was significantly different between the genotypes. Our findings potentially demonstrate an association between parameters of autonomic dysfunction with ACE genotypes.
AB - This study was designed to investigate parameters of autonomic dysfunction that may be under the influence of ACE ID genotypes. 136 patients with (47) and without type II diabetes were genotyped. Biomarkers such as HbAlc and eGFR, blood pressure, blood cholesterol are in part regulated by the autonomic nervous system and heart rate variability is an indicator of autonomic balance between the sympathetic and parasympathetic division. Several statistical methods were used, including the J48 decision tree machine learning algorithm to associate parameters of autonomic dysfunction and other biomarkers with ACE genotype. Non-parametric and machine learning methods detected more variables, which were able to contribute to classification of patients into genotypes. We found that HbAlc and TC:HDL were important nodes for separation of ACE genotype classes when the J48 decision tree algorithm was used. These were also verified by the Mann-Whitney analysis. Parametric comparisons of normally distributed variables revealed that only HDL was significantly different between the genotypes. Our findings potentially demonstrate an association between parameters of autonomic dysfunction with ACE genotypes.
KW - ACE I/D polymorphism
KW - Autonomic dysfunction
KW - Classification
KW - Genotypes
KW - Heart rate variability
KW - Machine learning algorithms
KW - Renin angiotensin system
UR - http://www.scopus.com/inward/record.url?scp=84863684112&partnerID=8YFLogxK
U2 - 10.2316/P.2012.764-084
DO - 10.2316/P.2012.764-084
M3 - Conference contribution
AN - SCOPUS:84863684112
SN - 9780889869097
T3 - Proceedings of the 9th IASTED International Conference on Biomedical Engineering, BioMed 2012
SP - 61
EP - 66
BT - Proceedings of the 9th IASTED International Conference on Biomedical Engineering, BioMed 2012
T2 - 9th IASTED International Conference on Biomedical Engineering, BioMed 2012
Y2 - 15 February 2012 through 17 February 2012
ER -