A view of the future from 2012: digital twins for healthcare

Here's a Facebook post from 2012 in which I describe what people now are calling a digital twin. The idea was fascinating to me then, and it turns out to have been quite influential in my career. Since then, I have worked on building systems to rapidly measure the molecular composition of human blood, and now I'm working on the computational framework for patient data that will mature into digital twins as we incorporate more and more data sources such as genomics and wearable monitors.

6 July 2012

Note to self: How to model an entire organism with molecular-level detail

Collect the most comprehensive data set regarding an individual that is possible. For a given state (at rest, for example), collect all vital signs, sequence transcriptomes from as many cell types as possible, sequence the genome, map brainwave activity, do a full-body MRI scan, measure the levels of everything we can in the blood, etc.

Repeat the process with tons of people in varying states and with varying health histories and genetic make-ups. Input data from people who are on various drugs and those whose physiology is otherwise altered.

Sick the data-crunching hounds of google translate on the data. Use the same empirical statistical approach to come up with thousands of tiny little rules that model the entire human body on a molecular level. By continually feeding in more data, the model becomes more accurate, precise and detailed.

Eventually, by feeding in a person's full data set, we will be able to make a model of that specific individual and test the effects of various treatments on the model before actually implementing them on the person. No more "Here, take this for six weeks and then come back and see me." Instead, we'll say "let me run that through your model - I'll be back in 5 minutes." Perhaps the model could even suggest treatments.

Would I follow exactly the same approach today? Not exactly. But the overall goal is still there.

What do you think it will take to build digital twins in healthcare? Let us know in the comments below!