Dissertation > Excellent graduate degree dissertation topics show

Study on the Predictive Modeling of Non-small Cell Lung Tumor Progression

Author: LiShuang
Tutor: WangPu; FangLiYing
School: Beijing University of Technology
Course: Control Science and Engineering
Keywords: Tumor progression modeling longitudinal data hierarchical linearmodel Varying-coefficient model Logistic varying-coefficient model
CLC: R734.2
Type: Master's thesis
Year: 2013
Downloads: 34
Quote: 0
Read: Download Dissertation


Since the1980s, lung cancer has become the highest mortality disease in the areaof maligant. Researches on modeling of tumor progression can help doctoresunderstand the disease and choose appropriate treatment. So the modeling of tumorprogression issue has been considered as one of the most important area forresearching.This paper introduces tumor modeling method in cross-section data, longitudinaldata and explains the features of medical data firstly. The patient data is mainlylongitudinal data so the modeling research of longitudinal data is the key point. Weproposed different modeling approach aim at different research emphasis. Then, aframework of DTAMS (Diagnosis and Treatment Aided Management prototypeSystem) for Tumor progression is designed and some modules are developed. Finally,a summary and future work is presented. The main work of research is summarized asfollows:(1) Based on the investigation and analysis of data feature, an overall researchplan is proposed for tumor progression modeling, introduce the workflow and relatedkey problems. Describes the the characteristics of medical data, and carries on thepreliminary screening and disposeof the data.(2) We analyze the survival quality of NSCLS using hierarchical linear model.Taking the time of follow-up and individual characteristics as input, and taking theFACT score of different time for the output, studying the varying trend in differenttime. Through the established model, we can assist the doctor have a overall grasp ofthe patients’varying trend of FACT score and take take different treatment fordifferent groups to improve the quality of life.(3)Studying the progression of the tumor size using varying coefficient model.In the process of data collection, each patient have different follow-up measurementtime, therefore the tumor progression data is longitudinal data.We used the varyingcoefficient model to explore the relationship between the detection index and tumorsize used the proposed two-step iterative procedure based on basis expansion andadaptive-LASSO penalty least square method, estimates the coefficient and screensthe irrelevant variable, meanwhile analyzing which variable is varying effect, whichone is constant effect, then can conduct the doctor taking reasonable control measuresof the larger coefficient to control the development of tumor progress. At last, we canpredict the tumor progression to assist the doctor know well the variation trend.(4)Except use the tumor size to reflect the tumor progression, we can definewhether the tumor progression include recur, transfer and the emergence of newtumor, it’s a classification data,1means has progression,0means has no progression. We use logistic varying coefficient model to analyze if there has a progression.Varying model can’t analyse the classification data, so we use Logistic varyingcoefficient model to anlyse the tumor progression. The tumor progression data isdifferent in each decete so this is longitudinal data. We firstly use b-spline toapproximate the coefficients and use Maximum likelihood method to estimateparameters, and then can obtain the probability of tumor progress of each checkfinally. If is probability is larger than50%, we think the tumor will progress.Besides, finished the framework design and developed the tumor progressionmodule in DTAMS. This work is helpful to the future research of group in our lab.

Related Dissertations

  1. Evaluation of the Pairwise Approach for Fitting Joint Linear Mixed Models,O212.1
  2. Model Selection for Longitudinal Data Based on the Quadratic Inference Function under a Simple Order Restriction,O212.1
  3. The B-spline Two-stage Least Square Estimation of Varying-coefficient Model Function Coefficients,O212.1
  4. Study of the Heteroscedastic in Longitudinal Data,O212.1
  5. Missing half of longitudinal data analysis of parametric regression models,O212.1
  6. Multi-state Markov Model in Outcome of Mild Cognitive Impairment,R749.16
  7. Two Statistical Inference Methods within Marginal Models: GEE and QIF,C81
  8. Asymptotic Properties of Estimators in Semiparametric Regression Model for Longitudinal Data under Linear Restrictions,O212.1
  9. Longitudinal data model selection Lagrange Multiplier,O212.1
  10. Joint Semiparametric Mean-Covariance Model in Longitudinal Data,O212.1
  11. The Property of Cross-validated Kernel Estimates of Varying-Coefficient Regression Model with Longitudinal Data,O212
  12. Response variable is continuous and multisection type of multivariate longitudinal data analysis,O212
  13. EM Algorithm for Binary Markov Chains of Longitudinal Data with Missing Data,O212.1
  14. The linear model longitudinal data punitive Generalized Method of Moments,O212
  15. Study on Traditional Chinese Clinical Response Evaluation Based on HMM,TP391.41
  16. Mixed Effects Model for Longitudinal Data,O212.1
  17. Analysis of Mixed-effect Model Based on Longitudinal Data,O212.1
  18. The Estimation of Semiparametric Model for Longitudinal Data Basing on Quadratic Inference Functions,O212
  19. The Estimation and Testing in GEE Method without PA Condition,O213
  20. Parameter Estimations and Their Asymptotic Properties in Random Varying-Coefficient Model,O212.1
  21. Semiparametric Regression Model for Longitudinal Data,O212.1

CLC: > Medicine, health > Oncology > Respiratory system tumors > Lung tumors
© 2012 www.DissertationTopic.Net  Mobile