Banca de DEFESA: SIMONE KARINE DA COSTA MESQUITA

Uma banca de DEFESA de DOUTORADO foi cadastrada pelo programa.
STUDENT : SIMONE KARINE DA COSTA MESQUITA
DATE: 26/08/2022
TIME: 13:00
LOCAL: Departamento de Enfermagem
TITLE:

Artificial Neural Network to assist nurses in decision making about venous ulcer dressings

KEY WORDS:

Nursing. Artificial neural networks. Validation Study. Varicose Ulcer. Technology. 

PAGES: 150
BIG AREA: Ciências da Saúde
AREA: Enfermagem
SUMMARY:

Venous ulcer is considered a major public health problem due to the socioeconomic impact it causes on society due to the increase in its incidence, as well as the gaps in health services, in particular, the difficulty of professionals in clinical judgment during treatment. choice of injury coverage, which slows down the healing process. In this way, a computer system can help nurses to minimize the difficulties facing the treatment of venous ulcers, through artificial intelligence. Among the intelligent technologies, the Deep Artificial Neural Networks have been standing out for their high power to solve complex problems, which makes it possible to maximize the quality of care offered to patients and contribute to the decision-making of professionals during clinical practice. Given the context, the study aims to provide computational assistance to nurses in decision-making about dressings for venous ulcers, through two neural networks. This is a methodological, descriptive research with a quantitative approach. The study was divided into four phases, according to the Unified Process (UP) methodological model. Phase 1 refers to the design and requirements gathering. In this phase, an integrative review was carried out on dressings used in the treatment of venous ulcers, construction of an image bank and the classification of tissues and coverage suggestions, by expert nurse judges in the treatment of chronic ulcers. Phase 2 comprises the elaboration of the structure of neural networks. Phase 3 refers to building, testing and evaluating the networks. To make it possible to evaluate usability, it was necessary to create a VenoTEC application containing 9 screens, aimed at classifying coverage options from photos taken of venous lesions of patients. Thus, the usability assessment process took place with 13 nurses, in the period of June 2022, using a scale, guided by the System Usability Scale (SUS). Data were analyzed according to a calculation guided by the SUS, as well as exported to the Statistical Package for Social Science 20.0 program, where they were analyzed using relative and absolute frequency statistics. The construction of the neural network aimed at tissue classification obtained an accuracy of 70% of correct answers, while the neural network aimed at the classification of coverage obtained an accuracy of 100% of correct answers. In this way, the networks proved to be effective for the application. Phase 4 refers to the finalization of the product. The result of the usability evaluation obtained an average of 88.07; which characterizes a very good usability. The participating nurses, when using the system, felt satisfied and stated that they would use the technology in their work activity. The development of the two deep ANNs aimed at offering computational support to nurses during VU treatment may bring contributions as a work tool based on scientific evidence and protocols, in order to standardize conduct in health services, which may favor the improvement of care , by minimizing errors, in addition to providing quick and professional decision making.  

COMMITTEE MEMBERS:
Externo à Instituição - JULIANO TEIXEIRA MORAES
Interna - 4665456 - DANIELE VIEIRA DANTAS
Presidente - 2882013 - ISABELLE KATHERINNE FERNANDES COSTA
Externa à Instituição - JESSICA DANTAS DE SA TINOCO - UERN
Interna - 1220986 - RHAYSSA DE OLIVEIRA E ARAUJO