Predictive model of cardiovascular diseases implementing artificial neural networks
Currently, there is a growing need from health entities for the integration of the use of technology. Cardiovascular disease (CEI) identification systems allow a large extent to predict diseases associated with the heart, thus allowing early identification of cardiovascular diseases (CVD) to improve...
- Autores:
-
Henriquez, Carlos
Mardin, Johan
Salcedo, Dixon
PULGAR EMILIANI, MARIA ISABEL
Avendaño Villa, Inirida
Angulo, Luis
Pinedo, Joan
- Tipo de recurso:
- Part of book
- Fecha de publicación:
- 2022
- Institución:
- Corporación Universidad de la Costa
- Repositorio:
- REDICUC - Repositorio CUC
- Idioma:
- eng
- OAI Identifier:
- oai:repositorio.cuc.edu.co:11323/10232
- Acceso en línea:
- https://hdl.handle.net/11323/10232
https://repositorio.cuc.edu.co/
- Palabra clave:
- SOM neural networks
GHSOM neural networks
Feature selection
Cardiovascular disease
- Rights
- embargoedAccess
- License
- Atribución-NoComercial-CompartirIgual 4.0 Internacional (CC BY-NC-SA 4.0)
Summary: | Currently, there is a growing need from health entities for the integration of the use of technology. Cardiovascular disease (CEI) identification systems allow a large extent to predict diseases associated with the heart, thus allowing early identification of cardiovascular diseases (CVD) to improve the quality of life of patients. In this research, a comparative analysis of the results obtained after implementing a series of feature selection techniques (Info. Gain, Gain ratio), and classification techniques based on artificial neural networks (SOM and GHSOM) was carried out, using the same data set “Heart Cleveland Kaggle Disease Data Set” hosted in the Machine Learning UCI repository and under the same test environment. Thus, to establish which of the techniques mentioned achieve a higher percentage of accuracy and precision when identifying patients who suffer from the disease under study. For the performance of the tests, cross-validation was used to select a percentage of the data set to perform them and another for training. Through the implementation of load balancing, normalization, and attribute selection techniques, it was possible to reduce the number of characteristics used in the classification process of the predictive model of cardiovascular diseases, which generated a reduction in computational requirements. Based on the above, 81.45% of successes were obtained with the hybridization of the Gain ratio feature selection technique and the GHSOM training techniques with the use of 7 features. |
---|