What can we learn from 'big data'?
The red pill movement, whether you agree with its conclusions or not (which I wont go into here), has given us a really good example of how you can learn from “big data” and how you can use population segmentation to simplify messages. I wonder if it is something we should be doing more of in primary care.
Without going into too much detail – the movement appears to have analysed the results of masses of data from dating sites and apps and learnt some interesting insights.
By it’s very nature when a user registers on a site they give a lot of information about themselves – this is perhaps analogous to the information I hold on my patients. The site itself then records lots of interactions as users view other users and swipe left or right – there is even outcomes data in a way as if a user keeps using the platform or not gives some conclusion on whether the connection was successful, assuming the outcome was long term relationship which may not always be the case!
The movement now talks about groups of people, e.g. Chads and 403s, and some ruder terms I wont list but these are essentially stereotypes that act in a similar way that share behaviours or characteristics and in theory help when summarising behaviours or allow targeting of messages. This is a classic example of population segmentation similar to the “Karen” moniker that is sometimes used. It is easier to talk about a Chad or Karen acting in certain way than constantly talk about percentages or present tables of data.
My local hospital did some interesting segmentation work where they analysed groups of patients presenting to them and they sub-divided them into groups (with titles rather than names) such as “fix-me” or “keep me well” they looked at the size of these groups and their needs/wants. I openly wondered if we should have a primary care version which might include groups such as “presentation of a self limiting illness” “having terminal care” “well and for prevention and screening” “chaotic user” “daughter visiting elderly relative” “holidaymaker” “student” I guess the point of these grouping is almost the opposite of the 80:20 rule – its identify small groups that have particular needs/wants and come up with a way of dealing with them rather than just thinking about the whole.
The NAPC have a 3x3 matrix model of segmentation which is largely based on age and complexity which can be one useful way of thinking about dividing your population and their needs but isn’t as specific or stereotypical as the above grouping but is worth looking at. It tends to group people into well – have some chronic diseases and frail and or complex with the latter needing specialist care and continuity and the former access and good health promotion/screening/vaccination services which could be delivered in a different way. Indeed the rise of the fuller centre might break general practice into these two sides with urgent care in one and chronic care in another.