Modelo bio-inspirado para la aproximación de nacimiento por cesárea
Fecha
Autores
Autor corporativo
Título de la revista
ISSN de la revista
Título del volumen
Editor
Compartir
Altmetric
Resumen
Cesarean delivery is one of the causes of maternal mortality with the highest percentage, in Colombia for 2021 it was 46.4%. To mitigate this, health professionals monitor pregnant women; however, the density of information and the volume of patients make it difficult to consider their symptoms in detail. Taking into account a public database that contains demographic information of 45 pregnant women and the recording of the electrohystereogram (EHG) biosignal, which is a non-invasive technique that represents the electrical activity of the cells of the uterus. A bio-inspired model is developed implementing the classifiers: K Nearest Neighbors (KNN), Multilayer Perceptron (MLP), Support Vector Machines (SVM) and Deep Learning (DP), to process data sets with demographic information, characteristics of the EHG biosignal, and their combination. Finally, the bio-inspired model is evaluated by calculating the performance indices: sensitivity, specificity and precision. The best performance in the analysis of demographic data is obtained with: KNN where the sensitivity is 100% and the specificity is greater than 80% together with SVM with a sensitivity of 75% and a specificity of 83.3%. The classifiers with the highest performance in the analysis of the EHG biosignal are: MLP with a sensitivity of 82.3% and a specificity of 85.7%, followed by PD with a specificity of 72.8%. The percentages with the highest accuracy are found when considering only the demographic data or the EHG biosignal, in contrast to that obtained with the combined data. The bio-inspired model allows for 90% accuracy with demographic information and 90.9% with EHG information.