SUPERMATH: Better Living Through Analytics (2010 Conference)

Not everyone can be Indiana Jones, but thousands of people worldwide got a chance to help find the tomb of Genghis Khan as armchair explorers in National Geographic's huge "Valley of the Khans" project. Covered by national media, the project has a tie to San Diego and to analytics -  a local UCSD graduate student team played a central role in the project, and will be featured at SUPERMATH in the Media, Marketing and Social Networks expo neighborhood.

While we usually think of analytics as being purely machine driven, humans are often a valuable component. Ph.D. candidate Luke Barrington will showcase UCSD research on human computation and how to use collective online intelligence to solve analytics tasks that outwit automated approaches. Luke will also present the design, development and results of a “virtual exploration” tool for crowdsourcing the search for the tomb of Genghis Khan. Luke was part of the "Valley of the Khans" project, a multi-disciplinary team of researchers and explorers led by UCSD's Dr. Albert Lin. In partnership with National Geographic, they built an application to solicit help from online participants in analyzing satellite imagery to search for modern and ancient structures in a remote region of Mongolia. A field expedition to Mongolia followed up on the locations tagged by virtual explorers and resulted in the discovery of over 50 sites of archaeological interest.

Luke will also demo and discuss Herd It, a game his team invented that motivates human players to contribute reliable tags for music. Herd It's crowdsourced data teaches machines to understand music. Luke will share with attendees how to apply the combined perceptual expertise of thousands of human contributors to analyze massive amounts of data with and produce prioritized facts on which to take action.

Register now for the conference - early bird discount ends this week!

A bit more about Luke:

Luke Barrington received the BE (elec) degree from University College Dublin (UCD), Ireland in 2001 and MS degree from the University of California, San Diego (UCSD) in 2004.  He is about to complete his Ph.D thesis on "Machines that Understand Music", a dissertation on the intersection between music, signal processing, machine learning, data mining and distributed, online human computation.  He is a co-founder of the UCSD Computer Audition Laboratory and of Music Search, Inc.  As a member of the Valley of the Khans project he traveled to Mongolia on a National Geographic expedition to search for the tomb of Genghis Khan.  He is an avid musician and wails on the guitar.

Today, we have a deluge of information from a variety for sources ranging from the latest medical devices to laboratory and radiology reports and data from pharmacy systems. All this medical information is intricately linked to how effectively patients are cared for. However, when it comes to really understanding what’s going on in a systematic fashion, the state of the art does not do enough.  If it did, we would not have so much waste, so much redundancy, and so many avoidable yet fatal medical errors.

Part of the problem is integration of these sources. Most current systems are simply not capable of analyzing the “human narrative” portion of the physician-patient interaction, so they stick to what is easy and non-subjective: structured records (such as billing codes, patient forms, and machine data such as body temperature or blood analysis). Unfortunately, that means today’s medical systems are missing 80-90% of the information that is potentially more useful. Understanding human narrative is fundamentally a difficult problem. Intelligence agencies have been trying for decades to understand human communication. Academic and commercial research, plus changes in computer speeds and costs, have only recently allowed some degree of successes in a very constrained setting within the large domain of healthcare.

Now think about it for a second. Isn’t it interesting? We live in this paradoxical world in healthcare: on one hand, we have reams of information and on the other we don’t have enough information - at least, not in a usable form.

So, how do we get all this data together and transform it to actionable information? How do we prevent waste? How do we prevent fraud and protect privacy? How do we improve patient care while reducing costs and medical errors? That is the Holy Grail.

To help us understand the complexities of the healthcare analytics world, we are fortunate to have Dr. Michael Victoroff as a keynote speaker at SUPERMATH. Dr. Victoroff is a medical doctor by training and has worked as a family physician on the provider side, as a medical director for Aetna on the payor side, as a patient advocate (see CNN interview), and has numerous specializations including creation of a widely used medical error taxonomy. More recently, he has delved into Natural Language Processing and unstructured data management including dealing with human narrative in the form common in the medical setting.

That diverse background and his lively stage presence promises for a great keynote speech. I, for one, cannot wait to hear him speak.

I first heard about Charles Komanoff's fantastic work in Wired Magazine, "The Man Who Could Unsnarl Manhattan Traffic". I was fascinated at the idea that this one guy - ok, a completely obsessed math genius guy - could figure out using a simple spreadsheet in 3 years what politicians and traffic engineers have been unable to achieve for decades.

San Diego traffic is getting worse every year - the same can be said for any major city. No supersonic coastal train is going to solve our immediate issues county-wide, and once the unemployment in California returns to normal there will be a noticable increase in traffic jams and commuter delays. Can the same analytical approach that Komanoff used to calculate human costs and gains in New York City work in San Diego and Southern California? It seems unlikely that we'd ever approve highway fees or zone tolls in downtown areas, but in my opinion it's worth analyzing the underlying "real costs" of delays in terms of productivity and gross state happiness. Komanoff has graciously agreed to come to San Diego to keynote at SUPERMATH and tell us about the challenges of getting "business as usual" groups to adopt analytics-driven policy.

I can't wait to hear him share his vision. You can find out more about Komanoff's life and career at his website: http://www.komanoff.net/

Photo copywright Wired Magazine 2010

Wow, the vision of SUPERMATH 2010 is really coming together. Many thanks to the steering committee for the Forum on Analytics, the Board of the San Diego Software Industry Council, committee chair Tom Clancy and SDSIC chair Bob Slapin (and of course the tireless efforts and support of Jill Curson).

This is not the official site for SUPERMATH 2010 - that's in the works, but we thought it would be good to post some of our working documents  on Posterous so that people can get a feeling for how a conference like this comes together. Also, we're lining up sponsorships and speakers asap.

A little history: this is actually the fourth year SDSIC has hosted the Forum on Analytics in San Diego, but this year we really wanted to knock it out of the park since so many exciting things are happening in analytics and there's so much momentum around technologies and people driving amazing new discoveries and applications.

So we made some key decisions that ultimately drove the vision and the event planning:

• 2 tracks: one public and one invitation-only (approved attendees)
• Private conference will be a "TED for Analytics" featuring thought leaders who are really putting analytics to work in their lives and their companies
• Public expo will be like an adult science fair, with hands-on demo pavilions featuring research and business analytics innovations - NOT a standard, boring trade show

The emphasis of the conference is to make people think, "Wow, analytics impacts almost everything I eat, drive, wear, or watch!" - to inspire and to educate, and to socialize a broader level understanding of the opportunities and possibilities.

We hope you'll mark down November 10, 2010 to attend the conference and/or the expo. We'll post the url where you can apply for registration, or buy expo tickets. Check back for updates as we start to flesh out the floor plan and the various "Analytics Neighborhoods".