We are seeing extraordinary development based on new infrastructure to electronically document, reference, and most importantly analyze health data. The amount of data generated from electronic health records, medical claims, laboratories, health sensors, genetics, social media, and more is beyond big, its 'gargantuan' data.

Considering the volume and the velocity of new health data creation, big data analytics has the potential to revolutionize both the science of medicine and the delivery of care.

Between the two opportunities, I believe the greatest value of big data and analytics is to improve the 'science of care delivery'. Big data provides a new channel for personalizing care, efficiently identifying improvements, and diffusing new research in care delivery science at scale. That's why the potential for health analytics and big data excites me the most ’ the ability to continuously learn from the experience of others and have it improves not only the strategy, but also the day-to-day care.

What many may not have realized is that the insights generated by the data will not realize their full potential until they are aligned with the science of medicine.

Put another way, until the practice of medicine changes--until physicians are trained in a different way of processing information--the clinical insights generated by many analytics platforms today will be difficult to integrate into care delivery.

Insights from many of today's big data analytical packages are entirely based on statistical correlates and complex algorithms. While these insights are immensely valuable, they are not grounded in medical science.
The Amazon recommendation engine, Google search algorithms, and IBM Watson are examples of amazing technologies that have changed the way we shop, the way we access information, and now helping us find better cancer treatments. But before these "insights" are used, they need to undergo rigorous scientific validation.

After all, it would be unreasonable to expect a medical professional to blindly hold correlate-generated insights on par with the clinical causal deductions based on validated medical science. Would you?
I believe that the incongruence in pedigree between correlate-generated insight and the scientific reasoning of medicine will prove to be a fatal downfall for a large number of healthcare analytics innovations.