I have said this before: More granular classification can lead to better care and, perhaps, better insurance coverage (in the US, at least). The current system clearly hasn't been effective enough - with people being misdiagnosed one way or the other - sometimes with terrible consequences. Type 1 clearly needs to be identified without unambiguously as soon as possible to result in less risk and better outcomes. Type 2. on the other hand is simply too diverse to be treated as one category: some need only lifestyle changes, some insulin, and some need one or more meds in order to achieve and maintain control. The many variations promote all the misinformation and misunderstanding we see today. In an age when other diseases and conditions are being defined with great specificity, it is ludicrous that Type 2 Diabetes is still a giant "catch-all" for non-Type-1 glucose dis-regulation.
I agree @Terry4 about the role of "big data." When I got my Masters in Healthcare Informatics, one of the goals we discussed was the aggregation of data among healthcare providers and across regions. I got excited about the information that could eventually be mined from all the health data that has been and is being collected. This appears to be a sample of what can be learned from this resource -- and it's not even close to a fully realized objective.