Final Project
BS849 - Bayesian Modeling in Public Health
Data
The data in the file tics.data.2021.csv report up to 5 longitudinally collected measures of “Telephone
Interview for Cognitive Status” (TICS) in offspring of centenarians and controls enrolled in two studies of
longevity and healthy aging conducted at Boston University. TICS is a test that consists of various questions
that are designed to assess a person level of cognitive function and can be administered over the phone.
Variables included in the file are:
• ID: patient ID
• Fam.num: family ID
• Sex: 0-Females, 1-Males
• Ptype: 0 = Controls, 2= Centenarian offspring
• Age.at.Enrollment: Age when subjects enrolled in the study
• Age.Last.Contact: Age when last seen, or age at death for diseased subjects
• Deceased: yes if participants has died
• BMI
• SH.Ever.Smoked.: yes for current or past smokers at enrollment
• MC.Aspirin: yes for regular aspirin use at enrollment
• MC.Stroke: yes for history of stroke at enrollment
• MC.Diabetes.Mellitus: yes for history of diabetes at enrollment
• MC.HTN: yes for history of hyper-tension at enrollment
• MC.Coronary.Artery.Disease: yes for history of CAD at enrollment
• MC.Cancer: yes for history of cancer, excluding skin cancer, at enrollment
• MC.Heart.Attack: - yes for history of heart atack
• Years.of.Education: at enrollment
• TICS01—TICS05: TICS score in 5 longitudinal follow-ups
• Age01—Age05: Age at TICS tests.
Report writing
Use R Markdown to write your report. Scripts and software output should be included separate
from your report and can be uploaded to Blackboard. My suggestion is to keep the report brief
(around 3 pages, at least 10pts fonts and 2 additional page of figures and tables). You can include
details of the analysis as PDF or HTML file from R markdown.
1
BS849 - Bayesian Modeling in Public Health
In this project, you will use the data to try to address the following:
1) What are the variables that are different when comparing offspring of centenarians and controls at
baseline? This analysis will be useful to select a set of possible confounders to include in the next
steps. Hint: Analyze variables one at a time.
2) Are the distributions of TICS score at baseline different between offspring of centenarians and individ-
uals without parental longevity after you adjust for possible confounders?
3) Use hierarchical models to investigate whether there are differences in the rate of change of TICS
between offspring of centenarians and individuals without parental longevity
4) Are there groups of study participants with significantly different rates of change of TICS?
5) Report on the missing data. Are the results robust to missing data mechanism?
You should conduct the analyses that you think are most appropriate and then write a report that includes:
i) Introduction to the report and description of your research questions;
ii) Methods section that describes all statistical methods used in the analysis. Be complete in your
presentation, including the choice of priors and the justification for those choices.
iii) Results of your analysis. Your results should include a table with summary of patients’ characteristics
at baseline, and a summary table that shows the results of the analyses. Provide some graphical displays
of the results that you think are most useful to convey the results and convergence diagnostics.
iv) A brief paragraph with discussion and conclusions.
Spring 2022 2