Research interests

Speaker notes:

Here’s a brief description of what I want to talk about. I’m not going to read this, but I put it here so that I will remember what all this was for when I look at this sometime in the distant future.

Speaker notes:

I got a PhD in Statistics almost 40 years ago. My first full-time job was teaching Business Statistics to undergraduate and MBA students. It was a publish-or-perish position and I perished because I was too slow to get my publications out.

I switched to a supervisory position at the National Institute for Occupational Safety and Health, which is a part of the Centers for Disease Control and Prevention. I supervised a couple of statisticians and a couple of programmers. Although I was not involved directly with any of the large CDC databases like NHANES, I knew the people who did that work and learned a lot from them.

From there I moved to Kansas City and became the first statistician at Children’s Mercy Hospital. I did pretty much all of the statistical work until my last year when they let me hire an extra statistician.

In 2008, I decided to become an independent statistical consultant. That business was up and down, but I found a large number of clients who were working on their dissertations and their dissertation committee did not have sufficient quantitative expertise to guide the analysis and write-up of the results.

At the same time, I asked around a few places to see if I could get a part-time position at an academic institution so that I could keep a few of the perks that I had at Children’s Mercy, such as access to a medical library. I got a 25% position at a program that was just starting to recruit students into a Master’s degree in Bioinformatics. I spent much of my time helping develop the program and worked until 2016 in the Research and Statistical Consult Service. My percentage effort varied over the years, and I am currently working at but I became benefits eligible in 2018 with a 75% effort in the department.

For the past couple of years, I have been working informally with the Center for Economic Informtion. This group has a massive amount of finely detailed information about neighborhood and housing information that they have correlated with various health outcomes like asthma and lead poisoning. They know geocoding applications really well and would be a great group to work with if you wanted to do any spatial analysis involving the Kansas City metropolitan area.

Speaker notes:

UMKC pays me a huge salary, so I try to work hard to justify their investment in me.

I teach several classes. We’re always looking for new students, especially from outside our department for the Introduction to R, SAS, and SQL classes.

I’ve published more than 100 publications, most in peer-reviewed journals, though only a handful are first author publications. Four of these publications have won major awards. I want to highlight one in particular, because it highlights what my job is. There was a audiologist at Children’s Mercy who was working on her dissertation, and her committee was not capable of providing the statistical guidance that she needed. It was fairly simple, a Poisson regression model for predicting the number of hearing aids in children that were lost or damaged. She got through her disseration defense just fine, and published her results in the American Journal of Audiology. Her paper was selected as the best publication in that journal for the calendar year 2003.

I wrote a book in 2006, published by Oxford University Press, about the critical appraisal step in Evidence-Based Medicine.

Speaker notes:

I’m currently on three research grants, and have worked on ten other grants since 1996.

The Case Studies grant is a very small grant, but I’m hoping to turn it into a bigger project. I have been interested in failures in medical research, because we learn as much by what doesn’t work as we do by what does work. I want to present a series of case studies in a graphic novel format. I self-financed art work on one study of the TGN-1412 trial, and got funding to pay for an artist to complete that work and start a new case study on the Duke ovarian cancer study.

The CTSA grant is housed at Kansas University Medical Center, but it really is intended for support throughout the various research institutions in the Kansas City metropolitan area. There are five major components of the CTSA grant, and my work is in the portion dealing with Health Informatics. My job is to promote greater use of Electronic Health Records in research. We have a quarterly meeting to promote the work of various researchers in this area and to encourage new researchers to adopt this technology. The next meeting, via Zoom, will be Thursday, September 17 from 4 to 6pm. The topic this time will be COVID-19 research, both at KUMC and Cerner. If you need details about this meetup, please contact me.

I’m also helping out Jared Bruce on a clinical trial for patients with Multiple Sclerosis.

There are some grants in the queue. I helped write two grants on lead remediation that uses some very nice geospatial data collected by the Center for Economic Information.

I wrote my own grant on peer-reviewed datasets, datasets that are included with peer-reviewed publications. The quality of documentation for these datasets is very uneven, so I proposed a training module on data sharing that includes a final project where students attempt to replicate the data analyses in several published research articles that has made the raw data available.

I’m also on a grant being submitteed by Jannette Berkley-Patton where she is using a church-based network to encourage broader testing for COVID-19.

Speaker notes:

I can help with anyone interested in statistical methods, especially anything out of the typical norms of research. The weirder the problem, the more I like it. I am especially interested in Health Informatics, the research application of electronic health records.

I am a big propent of the R programming language and especially how it can be used to make research reproducible.

Speaker notes:

Another big aspect of my work is that I am very interested in research about the research process itself. I have several papers looking at a Bayesian model for forecasting patient accrual rates in clinical trials. Accrual is the accumulation of patients over the course of a study, and it often lags terribly from what was predicted at the outset. The Bayesian model incorporates knowledge about patient accrual into a prior distribution and then updates the prior as patients show up for the study.

I have the PowerPoint slides available, at least for the short term, at my github site. I include speaker notes in almost all my PowerPoint presentations, because I hate how limiting the bulleted structure of PowerPoint is.