Page 19 - Contributed Paper Session (CPS) - Volume 4
P. 19

CPS2106 Julio M. Singer et al.



                         A linear mixed model for segmented regression
                                      with smooth transition
                                                   2
                                                                                     1
                            1
              Julio M. Singer , Francisco M.M. Rocha , Antonio Carlos Pedroso-de-Lima ,
                         Giovani, L. Silva , Giuliana C. Coatti , e Mayana Zatz
                                        3
                                                                          4
                                                          4
                              1 Department of Statistics, University of Sao Paulo
                 2 Paulista School of Politics, Economics and Business, Federal University of Sao Paulo
                              3 Department of Mathematics, University of Lisbon

                               4 Institute of Biosciences, University of Sao Paulo

            Abstract
            We consider random changepoint mixed segmented regression models
            to  analyse  data  obtained  from  a  study  conducted  to  verify  whether
            treament  with  stem  cells  may  delay  the  onset  of  a  symptom  of
            amyotrophic lateral sclerosis in genetically modified mice. The proposed
            models  capture  the  biological  aspects  of  the  data,  accommodating  a
            smooth transition between the periods with and without symptoms. An
            additional  changepoint  is  considered  to  avoid  negative  predicted
            responses.  Given  the  non-linear  nature  of  the  model,  we  adapt  an
            algorithm  proposed  by  Muggeo  et  al.  (2014,  Statistical  Modelling)  to
            estimate the fixed parameters and to predict the random effects by fitting
            linear mixed models at each step.

            Key words
            amyotrophic lateral sclerosis; fitting algorithm; mixed models; random effects.

            1.   Introduction
                Amyotrophic Lateral Sclerosis (ALS) is one of  the most common adult-
            onset  motor  neuron  disease  causing  a  progressive,  rapid  and  irreversible
            degeneration of motor neurons in the cortex, brain stem and spinal cord. In
            the majority of cases ALS occurs sporadically; in about 10% of the cases it is
            caused by familial reasons. No effective treatment is available and cell therapy
            clinical trials are currently being tested in ALS affected patients. The SOD1
            gene  encodes  an  important  antioxidant  human  enzyme  and  mutations  in
            SOD1 represent one of the most frequent causes of ALS.
                Among the different animal models for ALS, SOD1 mice are the most used
            in pre-clinical studies. After the initial tremor in the limbs they develop muscle
            weakness in early adulthood, become fully paralyzed and die. These mice over-
            express the human SOD1 gene bearing the G93A mutation, a point mutation
            found  in  familial  ALS.  Interestingly,  in  this  animal  model  the  disease
            progression is different between the genders. Males have a shorter lifespan
            and a clinical condition apparently more severe than females and differences

                                                                 8 | I S I   W S C   2 0 1 9
   14   15   16   17   18   19   20   21   22   23   24