Mostrar registro simples

dc.contributor.advisorCampos, Antônio Carlos Ligocki, 1958-pt_BR
dc.contributor.otherPinto, José Simão de Paula, 1963-pt_BR
dc.contributor.otherUniversidade Federal do Paraná. Setor de Ciências da Saúde. Programa de Pós-Graduação em Clínica Cirúrgicapt_BR
dc.creatorCruz, Magda Rosa Ramos dapt_BR
dc.date.accessioned2022-11-30T18:45:58Z
dc.date.available2022-11-30T18:45:58Z
dc.date.issued2018pt_BR
dc.identifier.urihttps://hdl.handle.net/1884/55329
dc.descriptionOrientador: Prof. Dr. Antônio Carlos Ligocki Campospt_BR
dc.descriptionCoorientador: Prof. Dr. José Simão de Paula Pintopt_BR
dc.descriptionTese (doutorado) - Universidade Federal do Paraná, Setor de Ciências da Saúde, Programa de Pós-Graduação em Clínica Cirúrgica. Defesa : Curitiba, 09/03/2018pt_BR
dc.descriptionInclui referênciaspt_BR
dc.description.abstractResumo: Introdução: A cirurgia bariátrica é considerada a forma de tratamento mais eficaz para a obesidade mórbida. A perda ponderal, controle ou remissão das comorbidades e melhora na qualidade de vida são resultados esperados no pós-operatório. A utilização de data mining é indicada na definição de variáveis preditivas e os modelos de regressão permitem a correlação das variáveis definindo o quanto uma influencia na outra. Objetivo: identificar os preditores para a baixa perda de peso no primeiro ano de pósoperatório de cirurgia bariátrica e de risco de reganho ponderal, com 5 anos de cirurgia. Objetiva-se também verificar a correlação destes fatores com a presença de comorbidades e satisfação do paciente com os resultados da cirurgia e estabelecer uma equação para estimativa do IMC médio com 5 anos de pós-operatório. Metodologia: Trata-se de um estudo longitudinal e retrospectivo, realizado com dados de prontuários de 388 pacientes submetidos ao Bypass Gástrico em Y de Roux, por videolaparoscopia. Foram coletados dados do pré-operatório (n=388), com 1 ano de pós-operatório (n= 388) e com 5 anos de pós-operatório (n=204). A aplicação de data mining foi realizada através da técnica de elaboração de árvores de classificação e regressão para a definição dos preditores do percentual de perda de excesso de peso (%PEP) com um ano de cirurgia, de IMC com um ano e 5 anos de cirurgia e de reganho de peso (RP) com 5 anos de cirurgia. O modelo ajustado para verificar as variáveis associadas ao %PEP e IMC com um ano de cirurgia e o RP com 5 anos, foi o de regressão linear múltipla. Para a elaboração da equação para estimativa do IMC médio com 5 anos de cirurgia bariátrica, foi utilizado um modelo linear generalizado com distribuição tweedie. Para finalizar, o teste de Mann Whitney (p<0,05) foi aplicado para verificar a significância estatística entre as variáveis com 5 anos de pósoperatório e o RP. Resultados: A utilização de data mining gerou duas árvores de regressão, identificando os fatores preditivos para baixa perda de peso com um ano de cirurgia e do IMC com 5 anos de cirurgia. Não foram identificados preditores para o RP com 5 anos de pós-operatório. Os preditores do menor %PEP com um ano de cirurgia bariátrica foram: presença de hipertensão arterial, maior IMC (IMC>39,84 kg/m²) e idade superior a 32 anos no pré-operatório. A baixa PEP, IMC no pré-operatório e IMC com um ano de cirurgia foram preditores para IMC mais alto com 5 anos de cirurgia. Fatores dietéticos do préoperatório não influenciaram na perda de peso no pós-operatório. Os modelos de regressão possibilitaram identificar a correlação entre as variáveis preditoras de perda de peso e a elaboração da equação para estimativa do IMC médio com 5 anos de cirurgia. A partir da interpretação dos modelos de regressão, observou-se que pacientes com baixos níveis sanguíneos de ácido fólico e de HDL, maior IMC e diagnóstico médico de ansiedade e/ou depressão no pré-operatório tinham maior IMC com 5 anos de cirurgia bariátrica. Observouse também que o maior IMC aos 5 anos de cirurgia esteve associado à maior prevalência de comorbidades, reganho de peso e insatisfação com os resultados da cirurgia. Conclusões: Os preditores identificados para PEP com um ano de cirurgia foram: idade, maior IMC e presença de hipertensão arterial. A PEP e IMC com um ano de cirurgia foram preditores do IMC com 5 anos de cirurgia. Níveis sanguíneos de ácido fólico e HDL, IMC e quadro de depressão e/ou ansiedade no pré-operatório foram fatores de risco para falha nos resultados do pós-operatório tardio e devem ser monitorados durante a avaliação clínica desde o préoperatório. Palavras-chave: Data mining. Cirurgia bariátrica. IMC. Perda de excesso de peso. Reganho de peso. Comorbidades.pt_BR
dc.description.abstractAbstract: Introduction: Bariatric surgery is considered the most effective form of treatment for morbid obesity. Weight loss, control or remission of comorbidities and the improvement in life quality are the expected postoperative results. The use of data mining is indicated in defining predictive variables and regression models enable the correlation of such variables, defining how much of an influence they have on each other. Objective: to identify the predictors for low weight loss in the first year of postoperative bariatric surgery and the risk of weight-increase, at 5 years following surgery. It is also aimed at checking the correlation of these factors with the presence of comorbidities and the patient's satisfaction with the results of the surgery, in addition to establishing an equation for estimating the average BMI following 5 years postoperative. Methodology: This is a longitudinal and retrospective study, carried out using data taken from the medical records of 388 patients that underwent Roux-en- Y Gastric Bypass, through video laparoscopy. Preoperative data were gathered (n=388), at 1 year postoperative (n=388) and at 5 years postoperative (n=204). The application of data mining was carried out through the elaboration of classification and regression trees for defining the predictors of the excess weight loss percentile (%EWL) at one year of surgery, of the BMI at one and 5 years following surgery and weight regain (WR) following 5 years of surgery. The adjusted model for checking which variables are linked to %EWL and BMI at one year following surgery and the WR following 5 years, was that of multiple linear regression. For elaborating the equation for estimating the average BMI at 5 years of bariatric surgery, a generalized linear model with Tweedie distribution, was used. Finally, the Mann Whitney test (p <0.05) was applied, to check the statistical significance between the 5-year postoperative variables and the WR. Results: The use of data mining generated two regression trees, identifying the predictive factors for low weight loss following one year of surgery and the BMI following 5 years of surgery. No predictors were identified for WR at 5 years postoperative. The predictors for the lowest %EWL at one year of bariatric surgery were: the presence of hypertension, a higher BMI (BMI>39.84 kg / m²), and age above 32 in the preoperative period. Low EWL, preoperative BMI and BMI following one year of surgery were identified as predictors for a higher BMI at 5 years following surgery. Dietary factors of the preoperative period did not influence weight loss in the postoperative. The regression models enable the identification of a correlation between the predicting weight loss variables and the elaboration of the equation for estimating the average BMI at 5 years post-surgery. From the interpretation of the regression models, it was possible to observe that patients with blood results showing low levels of folic acid and higher levels high HDL cholesterol, a higher BMI and having been diagnosed with preoperative anxiety and/or depression, had a higher BMI at 5 years following bariatric surgery. It was also observed that the highest BMI at 5 years following surgery was linked to a higher prevalence of comorbidities, weight regain and patient dissatisfaction with the results of the surgery. Conclusions: The predictors identified for EWL at one year following surgery were: age, higher BMI and presence of arterial hypertension. The EWL and BMI at one year of surgery were predictors for BMI at 5 years following surgery. Folic acid and HDL cholesterol blood levels, BMI, and a preoperative depression and/or anxiety may be considered risk factors for late postoperative outcomes and should be monitored during the pre-operative clinical evaluation and at one year of surgery. Keywords: Data mining. Bariatric surgery. BMI. Excess weight loss. Weight regain. Comorbiditiespt_BR
dc.format.extent1 recurso online : PDF.pt_BR
dc.format.mimetypeapplication/pdfpt_BR
dc.languageMultilinguapt_BR
dc.languageTexto em português e inglêspt_BR
dc.languageporengpt_BR
dc.subjectÍndice de massa corporalpt_BR
dc.subjectMineração de dados (Computação)pt_BR
dc.subjectPerda de pesopt_BR
dc.subjectGanho de pesopt_BR
dc.subjectComorbidadept_BR
dc.titleData mining e modelos de regressão na determinação dos preditores de perda de peso com um e cinco anos de Bypass gástrico em Y de Rouxpt_BR
dc.typeTese Digitalpt_BR


Arquivos deste item

Thumbnail

Este item aparece na(s) seguinte(s) coleção(s)

Mostrar registro simples