Contributions to flexible bivariate regression models. Applications in Medicine and Environment

• Autores: Óscar Lado Baleato
• Directores de la Tesis: Francisco Gude Sampedro (dir. tes.), Javier Roca Pardiñas (dir. tes.)
• Lectura: En la Universidade de Santiago de Compostela ( España ) en 2022
• Idioma: inglés
• Tribunal Calificador de la Tesis: Guadalupe Gómez Melis (presid.), César Andrés Sánchez Sellero (secret.), Bruno de Sousa (voc.)
• Programa de doctorado: Programa de Doctorado en Estadística e Investigación Operativa por la Universidad de A Coruña; la Universidad de Santiago de Compostela y la Universidad de Vigo
• Materias:
  • Matemáticas
    • Estadística
      • Calculo en estadística
      • Métodos de distribución libre y no paramétrica
      • Análisis multivariante
      • Técnicas de inferencia estadística
• Enlaces
  • Tesis en acceso abierto en: MINERVA
• Resumen

  • This thesis proposes bivariate regression models useful in clinical setting, in medical research, and in other fields. To make these models as widely as possible, no parametric restrictions for the responses variables were contemplated. A probabilistic region covering a specific percentage of the data points can be estimated for these models. This region characterises the bivariate distribution shape depending on the effect of covariates. In practical terms, this identifies which values are most likely to be observed in the general healthy population after adjusting for patient characteristics; a region containing 95% of healthy patients results as conditioned by covariates can thus be obtained for diagnostic purposes. This reference region is a natural extension of the use of reference intervals.