Today’s morning keynote kicked off with Steve Mills talking about big data – “as if data weren’t big before”, he joked – and highlighted that the real challenge is not necessarily the volume of data, but what we need to do in order to make use of that data. A huge application for this is customer service and sentiment analysis: figuring out what your customers are saying to you (and about you), and using that to figure out how to deliver better service. Another significant application area is that of the smarter planet: sensing and responding to events triggered by instrumentation and physical devices. He discussed a number of customer examples, pointing out that no two situations are the same and that a variety of technologies are required, but there are reusable patterns across industries.
Doug Hunt was up next to talk about content analytics – another type of big data – and the impact on transforming business processes. He introduced Randy Sumrall, CIO of Education Service Center Region 10 (State of Texas), to talk about the impact of technology on education and the “no child left behind” policy. New technology can be overwhelming for teachers, who are often required to select what technologies are to be used without sufficient information or skills to do so; there needs to be better ways to empower the educator directly rather than just having information available at the administrative level. For example, they’ve developed an “early dropout warning” tool to be used by teachers, analyzing a variety of factors in order to alert the teachers about students who are at risk of dropping out of school. The idea is to create tools for completely customized learning for each student, covering assessment, design and delivery; this is more classical BI than big data. Some interesting solutions, but as some people pointed out on the Twitter stream, there’s a whole political and cultural element to education as well. Just as some doctors will resist diagnostic assistance from analytics, so too will some teachers resist student assessments based on analytics rather than their own judgment.
Next was Frank Kern to talk about organizations’ urgency to transform their businesses, for competitive differentiation but also for basic survival in today’s fast-moving, social, data-driven world. According to a recent MIT Sloan study, 60% of organizations are differentiating based on analytics, and outperform their competitors by 220%. It’s all about speed, risk and customers; much of the success is based on making decisions and taking actions in an automated fashion, based on the right analysis of the right data.
Some of IBM’s future of big data analytics is Watson, and Manoj Saxena presented on how Watson is being applied to healthcare – being demonstrated at IOD – as well as future applications in financial services and other industries. In healthcare, consider that medical information is doubling every five years, and about 20% of diagnoses in the US have some sort of preventable error. Using Watson as a diagnostic tool puts all healthcare information into the mix, not just what your doctor has learned (and remembers). Watson understands human speech, including puns, metaphors and other colloquial speech; it generates hypotheses based on the information that it absorbs; then it understands and learns from how the system is used. A medical diagnosis, then, can include information about symptoms and diseases, patient healthcare and treatment history, family healthcare history, and even patient lifestyle and travel choices to detect those nasty tropical bugs that your North American doctor is unlikely to know about. Watson’s not going to replace your doctor, but provide decision support during diagnosis and treatment.
Dr. Carolyn McGregor of UOIT was interviewed about big data for capturing health informatics, particularly the flood of information generated by the instrumentation hooked up to premature babies in NICU: some medical devices generating several thousand readings per second. Most of these devices may have a couple of days of memory to store the measurements; after that, the data is lost if not captured into some external system. Being able to analyze patterns over several days’ data can detect problems as they are forming, allowing for early preventative measures to be taken: saving lives and reducing costs by reducing the time that the baby spends in NICU. A pilot is being done at Toronto’s world-class Hospital for Sick Children, providing analysis of 90 million data points each day. This isn’t just for premature babies, but is easily applicable to any ICU instrumentation where the patients require careful monitoring for changing conditions. This can even be extended to any sort of medical monitoring, such as home monitoring of blood glucose levels. Once this level of monitoring is commonplace, the potential for detecting early warning signals for a wide variety of conditions becomes available.
Interesting themes for day 2 of IOD. However, as much as they are pushing that this is about big data and analytics, it’s also about the decision management and process management required to take action based on that analysis.