Diabetes Presentation updated 12.10.13

IntroductionIntroduction

Diabetes - group of diseases that deals with high blood glucose levels Type 1 - Usually found in Children & Young Adults

Body does not produce insulin Only 5% of diabetes cases

Type 2 – Hyperglycemia – body has blood glucose levels higher than normal Body does not use insulin properly Most common form

Gestational Diabetes - Diabetes that shows in pregnant women 2 to 10% of pregnancies Usually occurs during the 24th week of pregnancy Disease not necessarily present before or after (35 to 60%

chance of Type 2 in 10 to 20yrs

Diabetes StatisticsDiabetes Statistics

25.8 million Americans have diabetes (8.3% of US Population)

7 million unaware the have disease In 2010 1.9 million people ages 20 or older were

diagnosed with diabetes Estimated that 79 million have “prediabetes” Diabetes By Gender:

13.0 million Men (11.8% of Men 20 or older) 12.6 million Women (10.8% of Women 20 or older)

ObjectiveObjective

Observe how disease behaves over time

Determine key factors in progression of Diabetes

Compare how disease behaves between men and women

DataData

442 Diabetes Patients Ten Baseline Variables:

Age Sex BMI AVG Blood Pressure Six (6) Blood Serum Measurements

Y ~ Quantitative Measure of Disease Progression after One (1) Year

Statistical AnalysisStatistical Analysis

Linear Regression Assumptions Significance Refine Model

P-VALUE Approach Stepwise Model

Models by Gender Results

SAS Code for Full ModelSAS Code for Full Model

SAS OutputSAS Output(Full Model - ANOVA)(Full Model - ANOVA)

Parameter EstimatesParameter Estimates

Graphs for Full ModelGraphs for Full Model

Residual PlotsResidual Plots

Residual Plots Residual Plots (Continued)(Continued)

SAS Code for Reduced Model SAS Code for Reduced Model (P-Value @ alpha =0.05)(P-Value @ alpha =0.05)

SAS OutputSAS OutputReduced Model - ANOVAReduced Model - ANOVA

(P-Value Approach)(P-Value Approach)

P-value ApproachP-value Approach

MODEL: Y= -13.03601SEX +6.44833BMI +1.00866BP

+50.53982S5-328.84936 F-Value = 103.62 -> P-Value <0.0001 R^2 = 0.4868

Residuals do not violate assumptions

Parameter EstimatesParameter Estimates

Graphs for Reduced Graphs for Reduced ModelModel

Residual PlotsResidual Plots(Reduced Model)(Reduced Model)

Reduced Stepwise Reduced Stepwise

Stepwise FinalStepwise Final

Stepwise Final Stepwise Final (Parameters)(Parameters)

Stepwise ApproachStepwise Approach

MODEL: Y=-21.5910SEX +5.7111BMI + 1.1266BP -

1.0429S1 +0.8433S2 +73.3065S5-313.7666 F-Value = 76.095 -> P-Value <2.2e-16 R^2 =0.5149

Residuals do not violate assumptions

Graphs for Stepwise Graphs for Stepwise ModelModel

Comparison of Reduced Comparison of Reduced & Stepwise Models& Stepwise Models

P-value Approach MODEL:

Y= -13.03601SEX +6.44833BMI +1.00866BP +50.53982S5-328.84936

F-Value = 103.62 -> P-Value <0.0001 R^2 = 0.4868

Stepwise Approach MODEL:

Y=-21.5910SEX +5.7111BMI + 1.1266BP -1.0429S1 +0.8433S2 +73.3065S5-313.7666

F-Value = 76.095 -> P-Value <2.2e-16 R^2 =0.5149

Code for Gender ModelCode for Gender Model

Gender 1 Gender 2

Gender 1 Gender 2

Graphs for Gender 1 Graphs for Gender 2

Residual for FactorsResidual for FactorsGender 1 Gender

2

Comparison of Gender Comparison of Gender ModelsModels

Gender 1 (MALE) Model: F-Value = 23.66 -> P-value <0.0001 R^2 = 0.4862 Influential Parameter: AGE, BMI, BP, S5 Intercept: -367.97615

Gender 2 (FEMALE) Model: F-Value = 32.88 -> P-value <0.0001 R^2 = 0.6004 Influential Parameter: AGE, BMI, BP, S5 Intercept: -466.14236

Analysis ResultsAnalysis Results

Models Full Linear P-Value Reduced Stepwise Model Gender Models

Talk about strength Reduction Methods

ConclusionsConclusions

Findings based on Full VS Reduced Main Influences on Progression of Disease:

GENDER, BMI, BP and S5 Findings Base on Gender Overall Diabetes Results

Recommendations for Recommendations for Future Future

Compare Different Types of Diabetes (Type 1, Type 2, Gestational

Observe over longer period of time Include additional Factors: Ethnicity, Other Health

Factors, Other Behaviors (Smoking, Alcohol Consumption, Exercise Habits, etc.)

Treatments: Diet, Medications, Physical Activities, etc.

Include Comparison of NON-diabetics

Questions? Questions?

ReferencesReferences

Data: http://www4.stat.ncsu.edu/~boos/var.select/

diabetes.tab.txt Description

http://www4.stat.ncsu.edu/~boos/var.select/diabetes.html

LARS Paper http://arxiv.org/pdf/math/0406456.pdf

http://www.diabetes.org Others?