PVRM Independent Study Course on Biomedical Informatics Driven Clinical Research Jasmin Nwachokor‎; Stacy Keller‎; Parker Bohm‎; Shaima Alothman

postscript: MitraProject

January 28, 2015: Introduction to HERON and Biomedical Informatics

Potential Topics:

PCORI Focused Aspirin Trial replication and support


  • Bariatric Surgery
  • Pediatric Antibiotics

Breast Cancer


Cross over with Patient Powered Networks

  • Vasculitis
  • Duchennes
  • MS
  • ABOUT/Breast Cancer


  • Scott Weir and IAMI: bladder cancer for high risk but not yet metastatic (premarket analysis, under treatment)
  • Joeseph LeMaster?, family medicine, diabetic foot ulcers and amputation (implementing best practices to avoid; predict who's at highest risk)

Discussion of topics. Come to class next with topic more fully formed and variables of interest.

February 4, 2015 Potential Fishing Trips


Shaima: Pain (out this session at a conference) ShaimaAlothmanProject

Stacy: PCORI ADAPTABLE aspirin study population, StacyKeller

  • cohort characterization regarding medication usage,
  • benefit for subsequent event
  • versus bleeding risk
  • read protocol
  • optimizing therapy during hospitalization
  • blood pressure meds if a diabetic, not on an ACEI or ARB, are they on the most recommended
  • education: do we have tracking in Epic or in the billing systems regarding pharmacy education and reconciliation
  • Current standard for cardioprotective use of aspirin. how many people who are aspirin naive are placed on it.
  • use of statins for lipids.
  • how well are aspirin and anticoagulants documented
  • do they have timely clinic visits after discharge? How often are their changes to the medication list?
  • bonus points wrt to pharmacy dispensing data

Jasmin: Breast Cancer JasminNwachokorProject

  • African American Females with Lymphodema
  • Amanda Amin faculty contact
  • Obesity
  • Age, smoking, alcohol, history of uterine/ovarian surgeries
  • immediate reconstruction
  • stage of cancer
  • type of surgery
  • menopausal status
  • family history
  • comorbidities
  • treatment (radiation, chemo, neoadjuvent)

Parker: Alzheimer's Disease with Jeff Burns ParkerBohmProject

  • Looking at medication use (namenda aricept)
  • supplements like reservatrol


  • Comorbidities and discharge placement
  • complications/variability in LOS and other characteristics where AD is a comorbidity
  • general exploration of what is richly documented on this population,frequency, intensity of care
  • mild versus moderate/severe
  • comparison with what is collected in the ADC and NACC forms REDCap database.
  • with diabetes, we have a concrete lab like hemoglobin A1c. Do we have cognitive tests with alzheimers?
  • look at reports note writer and flowsheets?
  • family history of alzheimers. Look at mother versus father.
  • how is dementia writ large characterized?

February 10, 2015 Practicing Your Cast

This class, discussion of cohorts and topic summaries

Introduce Data, databases, and SQL: Tamara

For next class, refine your one pager. Define your initial data elements and your query or queries in i2b2.

Next class will introduce SQL and data further.

  • Russ and Tamara's homework is finding a website that does a good SQL introduction for you.

February 18, 2015 Managing Data

SQL Tutorial and Exercises W3Schools has a good interactive website with the SQL basics and an online tool that lets you query a tiny relational database

We will go through some of the exercises in class together but become familiar as the data you will receive will be in a SQLite database

February 25, 2015 Data Request Oversight and Protocol Development

Need everyone's projects defined/scoped down.

  • what's your title?
  • what's the fundamental hypothesis(es) or descriptive aim?
  • what's your project proposal?
  • who's on your study team?
  • Queries written to define the cohort
  • data elements chosen for analysis

Keep in mind, some of you may have broad or ambitious initial thoughts as we've talked with interested parties. For example:

  • Stacy: you're fortunate with a protocol already established but for this class we want to be razor sharp on what we'll pull and then what specifically you might evaluate by May.
  • Parker: had a good discussion with Jeff Burns on several ideas. Let's funnel down on one in particular and a couple elements initially.

I want to see your projects elements above by Friday at 5 so next week we can submit data use requests.

March 4, 2015 Protocol Refinement

Everyone turned in their requests and they were all rapidly approved by the HERON DROC. We will spend this meeting today on refining the queries and the data elements required.

March 11, 2015 Casting your Net

We were successful getting Jasmin and Stacy to have access to a shared folder on the campus server so we can securely store the files.

Everyone could download and install the sqllitebrowser program

We then needed to download the "Extractor" program for the Mac Users from the Apple Store so they could unzip the .gz compressed files.

We could then import the data into a SQLLite database by downloading from the REDCap database for their project. Note: you will need to bribe Tamara to know which lines need to be deleted from the SQL file.

We then looked at the data views.

Russ then ran some example queries on Stacy's sample to show how to look at the modifiers and the patient dimension

select count(*)
from patient_view pv
where pv.race = 'black' 

select count(distinct dv.patient_num), dv.code_label, dv.modifier_label
from data_view dv join patient_view pv on pv.patient_num=dv.patient_num
where pv.race ='black'
group by dv.code_label, dv.modifier_label
order by dv.code_label, dv.modifier_label

select (count(distinct dv.encounter_num)/count(distinct dv.patient_num)) as frequentflier,
(count(distinct dv.encounter_num) % count(distinct dv.patient_num)) as frequentfliermod,
 count(distinct dv.encounter_num), count(distinct dv.patient_num),  pv.religion
from data_view dv join patient_view pv on pv.patient_num=dv.patient_num
group by religion
order by count(distinct dv.encounter_num)/count(distinct dv.patient_num) desc

Now that everyone has data access, focus on defining the first questions you want to ask so we can either resolve them in the class or define the data file format you will want to feed into your analysis program.

For example, you can probably analyze the basic demographics from just the patient view alone.

But, defining clopidogrel dosing at the encounter where the cath is done will take more manipulation.

March 23, 2015 Landing your Fish and Data Transformation

SQLite - Useful Functions

Sample queries

select min(start_date) as firstfact, (strftime('%s', end_date) - strftime('%s', start_date))/3600 as durationdays, dv.patient_num, dv.variable
from data_view dv
group by dv.patient_num, dv.variable
order by dv.patient_num, firstfact
select *
from data_view dv
where dv.variable like '%aspir%'
order by dv.patient_num, dv.encounter_num

select distinct  dv.modifier
from data_view dv
where dv.variable like '%aspir%'

select count(distinct instance)
from data_view dv
where dv.variable in ('Aspirin Oral Tablet') and dv.modifier in ('MedObs:Outpatient')
order by dv.patient_num, dv.start_date, dv.encounter_num, dv.instance

For next class, we will review methods for pivoting and also using Stacy's example of the frequency of aspirin at discharge during the encounter where the stent was placed. Initial approximation using Outpatient Modifier and the Aspirin Oral Tablet variable.

April 1, 2015: Gutting Fish and Introduction to Programming

You have now explored the EMR, registry, and billing data captured in HERON. You've been sponsored for data use and received your requested data in REDCap and the attached SQL database file formats covered in the last class.

As discussed last class, some questions can potentially be answered directly through clever use of i2b2 queries while others can be summed or managed via the REDCap case report forms or SQL statements. But, many of your topics involve more complicated data manipulation to enforce temporal sequences and criteria for the resulting data; not just to define the cohort.

April 8, 2015: Data preparation; scoping and REDCap analysis options

We spent time this class seeing how people were coming on their data manipulations in SQL and/or R to format their data in a manner amenable to the needs for their analysis. For simple analyses at the patient level (Jasmin's project) REDCap's built in export methods could be sufficient for some basic needs.

The student's preliminary data file formats, in Excel or as CSV files, need to be attached under their project specific pages on this wiki. Also enter a couple manual records of exemplar data for your analysis.

For next class, need to review the output data files created by the student's programs. If needed, the students could meet prior to next class with Tamara/Li?/DanC.

April 15, 2015: SQL Manipulation and Temp Tables

I turns out that if you use create tables to merge data, you can almost always merge things together to create an analytics ready data set. See the folders for each person's project

April 22, 2015: Projects' Status and Project Finals: Submission Ready Abstracts

Everyone please put your descriptions of your data transformation steps on line like Parker has done: and upload the word document of the steps and transformations including small tables that illustrate the data steps

Need you to pick your conference, coauthors.

Example of AMIA-CRI conference abstracts is here

Shaima: American Diabetes Association Scientific Sessions:

Combined Section meeting for the American Physical Therapy Association Note a June 23, 2015 deadline. That's what Shaima will target

Parker: American Academy of Neurology Ask Jeff Burns

Jasmin: Breast Cancer Symposium San Antonio Breast Cancer Symposium

Stacy: 2015 Midyear Clinical Meeting

Final project is the abstract and summary descriptive statistics on your population.

For next class, have your dataset completed, your abstract written without the final results. We will then review the datasets to determine what the final results might be for the abstracts and the descriptive stats for the final deliverable.

April 29, 2015: Reviewing your projects and abstracts

AMIA-CRI sample abstract specifications

See attached example of Sravani Chandaka's AMIA-CRI Abstract this year.

Next class, have your results to determine which resonate the most for the abstract.

May 6, 2015: reviewing analytic files and abstracts

Shaima has her data and abstract! Stacy has her data! Parker has his data and prelim analysis in SAS Jasmin has her data and prelim analysis in SAS

Go ahead and put up a brief paragraph on your wiki pages explaining how you did the data integration. Also attach the SQL you used to create the data file.

Go back to your REDCap projects, upload the SQL files and the resulting data files for archiving. Also load your analysis code as a file so we have a complete record of the project for future students.

May 12, 2015: present your abstracts, data, and preliminary analyses. Course evaluation

Last modified 2 years ago Last modified on Jan 21, 2016 6:41:47 PM

Attachments (4)

Download all attachments as: .zip