Data collection is not the challenge. We have cheap sensors, ubiquitous connectivity, mature database tools, and inexpensive computing power. We also have decades of legacy data – both digital and analogue – in government and industry archives waiting to be used.
Developing models and frameworks to encapsulate, integrate, interpret, and utilise the data is the next big challenge.
Raw data is not that useful; other than to induce headaches or sleepiness. Models and frameworks are needed to convert the raw data to insights. Deriving insights requires the identification of patterns and trends; followed by the analysis and interpretation of these patterns and trends. Ideally, we can use these patterns and trends to make predictions about the future.
In a personal healthcare scenario, we may identify cycles of low/high mood that are correlated with sleeping and eating cycles. The insight in this case would be: our eating and sleeping patterns are affecting our mood.
Making insights useful requires understanding context and the change that needs to happen as a result of using these insights.
Following on from the example above, we can say the change is: gaining more conscious control of our mood by actively regulating our eating and sleeping cycles.
From these insights, we devise strategies for actions to affect change. In this example, the actions could be: a mobile phone app with reminders and alerts. Or a virtual dashboard displayed on a large screen.
One key impediment in the model making process would be: How can we answer these questions without being caught up in technology implementation issues too soon?
Building a model, a framework, or a methodology is not the same as building an app, an expert system, or training an AI matrix. Coming from a tech and design background, it can be incredibly difficult to separate these two foci.