Abstract
OBJECTIVE: To build a linguistic model using the properties of fuzzy logic to estimate the risk of death of neonates admitted to a Neonatal Intensive Care Unit. METHODS: Computational model using fuzzy logic. The input variables of the model were birth weight, gestational age, 5th-minute Apgar score and inspired fraction of oxygen in newborn infants admitted to a Neonatal Intensive Care Unit of Taubaté, Southeast Brazil. The output variable was the risk of death, estimated as a percentage. Three membership functions related to birth weight, gestational age and 5th-minute Apgar score were built, as well as two functions related to the inspired fraction of oxygen; the risk presented five membership functions. The model was developed using the Mandani inference by means of Matlab(r) software. The model values were compared with those provided by experts and their performance was estimated by ROC curve. RESULTS: 100 newborns were included, and eight of them died. The model estimated an average possibility of death of 49.7±29.3%, and the possibility of hospital discharge was 24±17.5%. These values are different when compared by Student's t-test (p<0.001). The correlation test revealed r=0.80 and the performance of the model was 81.9%. CONCLUSIONS: This predictive, non-invasive and low cost model showed a good accuracy and can be applied in neonatal care, given the easiness of its use.
Highlights
The fuzzy logic theory was introduced in 1964 by Zadeh, when he worked with problems of classification of sets that did not have well-defined boundaries
Os valores do modelo foram comparados com os fornecidos por especialistas e seu desempenho foi estimado pela curva ROC
A significant difference can be noticed between the risks predicted by the model according to the outcome
Summary
The fuzzy logic theory was introduced in 1964 by Zadeh, when he worked with problems of classification of sets that did not have well-defined boundaries. The theory of fuzzy sets became an important mathematical approach in diagnostic systems, medical imaging treatments, and more recently, in epidemiology and in Public Health[2]. The application of this theory in the medical field has shown great ability to enhance and develop equipment and models in various hospitals and research activities[2]. The first newborn would have an hypothetical membership of 0.85 in the Low Weight set, and of 0.15 in the Normal Weight set, while the second newborn would have a membership degree of 0.15 in the Low Weight set and of 0.85 in the Normal weight set[2]
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
