The ‘Teslarization’ of Healthcare
Author: Dr Joe Taylor
Not only can you silently sweep past queues of panic buying hydrocarbon car owners but owning the latest and greatest in automobile innovations might just save your life.
When I bought my Tesla I was reprimanded by the charming chap on the phone for wishing to purchase a car. He told me in no uncertain terms that Tesla was a software company and not a car company. It was a disappointment at the time as I really did want a car, but now I get the point.
A slew of exciting healthcare dedicated technologies is now in the market to monitor our health and fitness, but there is a much more exciting revolution around the corner.
Your smart car will be your own personal state-of-the-art diagnostic laboratory.
Why ‘Dr Tesla’ could already have seen you now
The logic of turning an everyday consumer essential into a healthcare platform is a well-trodden path. You wear a watch to tell the time, it happens therefore to be conveniently placed to record continuous healthcare information too, so you integrate those functions and now telling the time becomes a secondary feature.
Cars are increasingly intelligent, none more so than Teslas. These cars are collecting a huge amount of data to support their semi-autonomous driving capabilities, but their healthcare potential remains untapped.
When getting into the driving seat of one of latest generation of smart cars you are potentially entering into the one of the most advanced diagnostic clinical environments imaginable.
It’s difficult to overstate how rich the dataset being collected about the Tesla driver is. As a neuroscience researcher at Oxford 15 years ago I was collecting a tiny fraction of the amount of data at huge cost and with enormous effort, and in clinical practice we had next to none of this data to support diagnoses.
The power of longitudinal data
A one-off snapshot is very weak compared to data that spans long-periods. Subtle changes in health gradually evolving over time often go unnoticed by people and those close to them. Objective trend data is therefore of enormous value in spotting drift in healthcare variables.
In the case of Tesla your personal account follows you from car-to-car and the loyalty of Tesla owners to the platform means that driver data could potentially span many years.
What could we currently achieve with existing car technology?
Tesla vehicles regularly receive over-the-air software updates that add new features and functionality. The car is constantly evolving through software changes in the context of static hardware. Consequently, there are many healthcare functions that could be added to vehicles already on the road rather than requiring redesign of the cars or new inventory to come onto the market.
1. How are your reaction times evolving?
As our cars can ‘see’ what’s going on around us they are stepping in to keep us safe when we’re not quick enough to do act ourselves.
Slowing reaction times to suddenly slowing cars, dogs running out into the road or changing traffic signals could indicate a cognitive problem and warrant formal assessment.
2. Are your hands shaking at the wheel?
As Tesla drivers already know that ‘nanny-nag’ is an often-annoying feature that means unless you have your hands applying pressure to the wheel every so often the car will refuse to drive itself. The feature relies on torsional force on the wheel not on grip pressure so that even when the steering wheel is straight the twisting pressure being applied is being measured.
If your previously steady hands start to shake at the wheels this could indicate that you have a developing movement disorder. One of the most important clinical diagnostic differences between tremor is whether you shake at rest or during intentional movements. Given that driving involves both periods of rest and active steering then there is the ability to differentiate between tremors with two very different groups of causes.
3. Do you have to play the radio at increasingly high volumes?
Maybe you bought your Tesla to belt out Bruce Springsteen as part of your midlife crisis. I’m not one to judge. However, if over time the average volume you have the tunes at is increasing this may indicate hearing loss.
Not to worry – your car will know where the nearest hearing test is available and maybe you could give them a call from now at its suggestion?
4. Why do you have the heating on in summer?
People with hypothyroidism are always too cold, their body doesn’t have enough of the thyroid hormone signal to burn energy creating internal heat.
If you used to be a free-wheeling wind in your hair sort of driver and gradually enjoy being cocooned in the toasty embrace of your car’s heated seats on max then maybe you should consider getting your thyroid levels measured.
5. Should you be making better use of your gym membership?
Weight sensors in the seats of cars are there to alert you to an unsecured passenger, but they could easy tell if you were piling on the pounds. I’m afraid your car also knows when your visits to the gym fall from those earnest weeks immediately after signing that 12-month contract.
I would quickly fall out with my car if it started to be overly judgemental, but if it suggested that my planned route would enable me to swing by the gym I would be less likely to take a sledge hammer to it.
1. ‘Who you looking at?’
In the latest Tesla models cameras are not just looking out at the road but also looking at the driver. This opens a world of new opportunities for continuous healthcare monitoring.
Vision-based healthcare monitoring has become increasingly capable. Oxehealth’s Oxevision system can derive medical grade cardio-respiratory vital signs. Whilst the company is currently using specialist hardware, it is possible to apply these same data extraction approaches to video feed from ‘normal’ cameras.
2. Keep your hands on the wheel
For the moment we still have our hands on the steering wheels of our cars offering the opportunity for integration of contact sensors into those wheels. Hyundai’s car healthcare system concept proposes that this is an ideal point to measure ECG data.
3. Integration with wearables
To provide a truly 24-hour monitoring solution, the healthcare features of our cars would need to be integrated with information about what we are up to when outside of our cars.
Either integration with existing ubiquitous wearables or development of entirely new technologies is the natural solution. When you buy a healthcare enabled car, it is likely to come with an accompanying wearable that seamlessly integrates the data from when you’re at the wheel to when you’re outside of that environment.
4. Co-ordination with static healthcare services
In an acute emergency, such as a heart attack, stroke or fit the car could automatically use its communication capabilities to alert emergency services and even the local hospital.
As autonomy improves not only is the car likely to bring you to a safe stop in the case of an acute incapacitating event, but in the future may drive you to the nearest appropriate healthcare service.
5. Healthcare interventions
The diagnostic potential of smart cars is clear, but could they intervene to improve our health? I often loose against my car at chess whilst waiting for it to charge – it turns out it’s smarter than I am. If the car suspects you are getting a bit sluggish in your cognition, it could suggest some brain training on screen whilst you’re next waiting.
Our cars will be able to detect rising levels of stress, and there could be mental health interventions such as a breathing exercises, that the car could suggest to calm you down when road rage looks to be your next destination.
The power of AI and mass recordings
The value of continuous driver monitoring will only be truly realised through the application of AI to the massive amounts of data from potentially many millions of drivers.
Developing useful AI is costly and time-consuming. Fortunately, this is not a problem Tesla is unfamiliar with and other manufacturers are quickly having to come to terms with.
The development of self-driving capabilities in Tesla vehicles has required collection of a mass of data. The solution to autonomous driving is training a large neural network. It is this same approach that we will see transform healthcare.
There are more than a million Teslas on the road sending information back to the ‘transformer’ – a single giant neural network. This model is data hungry and eight external cameras in addition to the new internal camera feed back into this beast. Tesla is vertically integrating its AI strategy by creating custom silicon to speed up the training process for its net, recently revealing details of its D1 chip to speed up the machine-learning algorithm behind Autopilot.
An AI semi-autonomous driving model might incidentally become a diagnostic healthcare platform without that being its primary goal. A neural network that has enough information to optimise its primary goals may find value in incorporating indicators of health and wellbeing.
Challenges to smart cars becoming healthcare environments
There are natural privacy concerns regarding how much information is being collected, how it is being used, stored, and distributed. These are the same questions we grapple with across the tech sphere, but given people tend to anthropomorphise their cars rather than their other tech perhaps this is platform people will be more willing to place their trust in.
It is also possible that as the car becomes a diagnostic clinical environment we will have to grapple with their regulatory status. The fitbit sense has FDA 510k approval for the detection of Atrial Fibrillation and it’s likely that similar considerations will have to be considered for the health functions of smart cars.
The capability and legality of truly autonomous driving on our roads is much closer than most imagine, within five years we are unlikely to be required to pay active attention to the cars that drive us. This limits the amount and type of healthcare information that can be recorded but there will be many cars that will require active driver participation for a couple of decades to come.
When we become used to cars as diagnostic healthcare environments, we are unlikely to let go of that value with the establishment of autonomous driving and will still consider getting into a car as a healthcare choice. I anticipate this will first be manifest in the development of healthcare diagnostic tools that are aimed at monitoring passengers – key indicators such as their eye movements in response to the outside world can provide some of the information that can be gleaned from drivers’ interaction with the environment.
Perhaps it’s ironic that the innovation that has supposed our sedentary lifestyles during the 20th century could be the most important vehicle for delivering healthcare improvement in the 21st.
Apple’s much awaited entry into the smart car space is likely to come with a big health and wellness aspect, and we should expect plenty of other manufacturers to realise the untapped healthcare potential of their smart cars.
As all automotive manufacturers fight for their existence in an industry rocked by technology this is likely to prove a key battlefield.
Author: Dr Joe Taylor