Monday, April 15, 2013

A Holistic View of SAS: Health User Group


Last Friday, the SAS Health User Group (HUG) took place in Toronto. I am consistently in awe of this amazing group of professionals… perhaps even over-awed to the point of intimidation, if I’m to be honest. It’s amazing for me to listen to talks which bely such a fundamentally strong grasp of statistics, health indicators and SAS programming… and a little daunting when I’m on the agenda as well. It feels a little bit like a young Luke Skywalker charging into battle with the wily Darth Vader before he was ready. I think we all know what the outcome of trying to play with the big boys before ones’ time can be. I was just hoping I would exit the meeting with both of my hands – and my pride – still attached.

Fortunately, this group is as magnanimous and forgiving as it is intelligent. I’ve always been received well despite my technical shortcomings. Perhaps that’s because the agendas are consistently filled with incredibly strong presenters who’s SAS skills easily compensate for mine. This past Friday was to be no exception to this rule.

I led off the meeting with an overview of the new SAS VisualAnalytics product and I did my best to contextualize it for the health practitioners in the room. The talk centered around fictional data of approximately 3 million individuals with multiple data points and around 150 variables. The purpose of the talk was to visually explore the data without excluding or sampling right away with the goal of finding some trends – so matter how obscure – which could be impactful upon the onset of renal disease in an older population of Type II diabetes sufferers. I mentioned to the executive committee over lunch that the topic was actually quite personal for me, as I recently lost a family member quite suddenly due to complications from renal failure. It’s fair to say I wanted to do the topic justice. I began my talk working with billions of correlations in a basic matrix and worked down to some geo-mapping functionality to show the ease and power of the product. Now I’m no expert – not even close! – but I’d like to think that given the amount of questions which followed the talk, there was enough interest generated to have make the topic worthwhile to bring forward. I’m certainly grateful for the patience of the group as I wrestled through some of the terminology and methodology behind the talk.

Our next presenters took the group in a totally new – and welcome – direction. Yuriy Chechulin and Amir Nazerian of the Ministry of Health detailed how they used advanced modeling techniques to identify highcost users of the healthcare system. The objective of this research was to identify ways to ease the burden on the health community in terms of financial investment and time, as well as to help identify triggers and potential solutions to preventing negative health outcomes in the early stages. I really loved this talk. Even as a non-statistician, I could understand the principles behind the modeling techniques used… and I’m always fascinated by the data which helps inform these decisions. Once again the audience proved to be more than engaged with the topic as they peppered both of the presenters with questions about other potential applications for the model and inquiries around some of the exclusions they had made. It was a shame to have to cut off the questions for our break!

Upon our return, we had three tremendous guest speakers. SAS Canada’s own Judy Orr Lawrence walked us through some data cleansing and validation techniques in SAS Enterprise Guide. While more of an ‘ad hoc’ approach to data cleansing, it remained a very intriguing talk… especially for this long-time EG user. I’ll be giving a few EG talks myself in the coming months and I certainly plan on stealing some of her material (with permission, of course) for the talks. I love the way Judy presents: accessible, thorough and with the ability to keep the topic very high-level or dive very deep. As a frequent student in her classes I can tell you with unbiased certainty that anyone would benefit from her tutelage.

Next up, Ellen Maki gave us a deep dive into how she used Bland-Altman bounds to work towards a new understanding of sleep apnea treatment. Utterly fascinating. As I mentioned at the meeting itself, I’ve had sleep consultations several times and I’ve received wildly different diagnoses on each occasion. Ellen’s talk about the potential of taking future tests at one’s home nearly made me weep with joy: those of you who may have had to experience an over-night visit at a sleep clinic can attest that it is not necessarily the most comfortable of experiences to say the least! Even more importantly, Ellen implied that her data indicated quite strong and clear results in her study which is certainly good news for the future diagnosis and treatment of this very harmful sleep disorder. Her presenting style was extremely relaxed and easy, and I do believe that everyone in the room benefitted from her understanding of how she worked through her tests… and hopefully were inspired for their own work as well.

Finally, the incomparable Ruth Croxford concluded the meeting – in a mournfully abbreviated fashion – with a talk on using Perl expressions within her SAS code. I don’t have enough positive things to say about Ruth. As the President of the Executive Committee, she consistently finds great guest speakers (including herself) and is just such a fantastic representative of the healthcare community. I felt horrible that her talk was truncated due to earlier extended question periods, in no small part because I’ve been fascinated by Perl expressions for a long time and wanted to learn as much as I could about them. In her brief time, Ruth did manage to convey a lot of information. She easily moved through an introduction to Perl and also made it very simple to understand just how they could be leveraged in SAS. She certainly had a few people excited to try the procedure which allows you to do so!

All in all, the meeting was extremely successful. We had a record number of registrations – nearly 120 people – and I’m willing to bet we achieved a record for attendance as well. I give all credit to the fantastic guest speakers and the compelling agenda. Thanks once again to all of our guest speakers and to the Executive Committee who made it all happen! If you weren’t able to attend, not to worry: you can always pick up the presentations on the HUG website or connect with us in the SAS Canada Community I do hope we’ll be able to deliver just as strong a meeting when the Health User Group next meets in person on November 15th, 2013. In the meantime, I’ll be headed to Quebec City for their user group meeting and I’ll be enjoying a trip to San Francisco for SAS Global Forum at month’s end. I’m sure I’ll have lots to say about both meetings, so stay tuned.

Until then…

 

1 comment:

  1. Matt
    Agree about this group. Very interesting to see data mining in healthcare.

    ReplyDelete