No, I am not talking about when couples dress the same, I am talking about your twin. Not, not the one born 5 minutes before you that will forever hold that over you, your digital one. And, finally, no this is not some 2024 version of the Sims. Digital twins are what you’ve always wanted: someone to test something so that you know how it will work for you or similar. The only difference is that someone is you, or at least the highly likely scenario of what will happen to you based on all your previous data.
That’s right, the ‘holy grail’ so to speak of predictive technology is not far away, so prepare yourself. If you still aren’t sure what this may look like, the early days will be things like, scanning a barcode or taking a photo of some food, and being able to use that to understand what ingesting it will do to your unique physiology. Through ongoing iteration there will also probably be the ability to go the other way, for example, what should I eat for dinner given my deserved physiological state and in my current context.
Digital Twins in Medicine
I have previously written about the challenges of interpreting group data for the individual, or indeed how to apply research to yourself as an individual. This challenge extends beyond you and your own attempts to improve your health, it exists for doctors too. They make treatment decisions for the same individual (you) based on the same group outcomes of research, they do have the advantage of some clinical experience to help them but this is still a difficult task when ot comes to knowing what may work for you in certain cases.
Enter digital twins. These will allow modelling of response to various treatments for any issue, to understand the best one for you in the given situation. This becomes particularly pertinent when considering issues such as polypharmacy and the fact that if you are taking enough medications you will see an adverse drug interaction. In this context a digital twin may allow a doctor to foresee an issue and/or know which medications will have the best outcome as a group, allowing for better decision making and minimisation of harm.
Digital Twins and Health
When it comes to daily decisions that impact our health I think the power of a digital twin is significant. Generally, unhealthy behaviours are akin to a credit card: fun now pay later, rather than their healthy counterparts which are more like cash: you have to do the work up front before you see the benefits. The problem with this, is that the cumulative ‘costs’ of poor choices aren’t immediately apparent; making the habits hard to change/break. A digital twin foreseeably helps in this realm: understanding what any one choice will make immediately may be enough to change behaviour.
So, using your digital twin you will understand how to better deal with the previous night’s poor sleep, or what you should do today for your workout given your current status. As mentioned, early versions of digital twins already exist in the nutrition space, allowing you to better understand what any one food will do to you metabolically.
Digital Twins in Performance
I was recently discussing elite sports performance with someone who asked about the individualisation of training to people in team sports, as it is in endurance sports. Specifically, he asked my opinion about whether this would ever come to team sports, to which my answer was yes and it would be via means of things like digital twins. The challenge in team sports is resources required to truly individualise training, but with the use of big data and digital twins coaches would better understand what an appropriate dosing of different training modalities would be and could then free up resources of the individual to address other areas of need. Similarly, on an individual level in the future for people training in a non-professional environment, a digital twin will allow better decision making when it comes to how to train for optimal outcomes based on your goals.
One thing to note, which can be a challenge for some, is that digital twins or similar big data type of solutions output the ‘what’, the solution. They do not output the ‘why’ or the reasoning behind it. As mentioned this can be a real challenge to those with domain specific knowledge especially and reminds me of my time in primary school mathematics where the teaching was always ‘show your working’. These solutions are very much ‘black boxes’ though this does give us the opportunity to reverse engineer the reasoning and mechanisms from there, perhaps also giving us a quantum leap in learning that way and unlocking some knowledge that was otherwise not on the horizon.
In reality, the data required for digital twins is significant. The challenge being, appropriate volumes, quality and the temporal nature to truly create a wholistic digital twin. As mentioned to this point the’ve been very much focussed on things like nutrition, which is a little more of a ‘known quantity’ so to speak. Once we start adding factors that are more lifestyle based, things will get much more difficult, for example, predicting when to go to bed for optimal sleep given all the necessary inputs is foreseeably doable but also extremely difficult given the number of inputs that impact sleep (previous sleep, mental and physical activity, nutrition and the list goes on). All this is to say, we are moving towards digital twins but we aren’t as close to something wholistic as many would like.