One in four people have mental health issues and in men under 50, suicide is the main cause of death. It is a huge, often misunderstood problem that pervades every society and most families.
This was the opening message delivered by Valentin Tablan, an artificial intelligence expert and scientist at Ieso Digital Health, at the Healthcare Summit in London last month.
His message was clear: the mental health problem is endemic, but we know the solution in cognitive behavioural therapy. The issue is not everyone has access to this service. There are not enough therapists to deliver the necessary therapy.
The UK is better than most countries, however, with 15% of those suffering receiving treatment and a five year plan aiming to treat 25% (a further 10%). But clearly there is still a “scandal”, as Tablan states during his keynote speech.
>See also: Transforming healthcare with tech
Ieso’s mission is to try and help the mental health problem, by putting therapists and patients together via an online channel. It is a sizeable provider, with nine million people using the service and it has seen positive results.
Of those that use the service, there has been a 48% recovery rate and a completion rate of 55%. People using this service are getting better faster.
Working with the NHS, Ieso has treated 10,000 people suffering from mental health. The reason this platform has had success in tackling the issue surrounds it’s use of data.
Using it’s online channel, the health provider accumulates patient data based on their interaction with the therapists.
The digital triage
Every new patient creates a hypothesis for a certain condition. This data is then transferred to neural network, where AI and analytics begin to build a picture of different illnesses and how to treat them. Or, what treatment works best for different individuals.
With something like this, the more data the better. As the patient numbers grow on the service, this in turn will provide therapists with tools that go beyond a notepad and paper in the form of an AI assistant. Not, an AI therapist, because human interaction is still crucial to the healing process.
Tablan discussed with Information Age how AI can help address the stigma surrounding mental health and effect real change within this field of healthcare. He also discussed the benefits of technology in healthcare in general and the challenges it faces being implemented nationwide.
How can AI help mental health care?
What we’re doing is essentially trying to make therapists as effective as they can be, by making them the best therapists they can be, while delivering the treatment protocol, which equates to something like a physical treatment.
If they do deliver that and the patients get better, faster then the same number of therapists can see more patients, which will help dent the huge number of people suffering from mental health disorders. Not only in the UK, but worldwide.
Our therapists, through our processes and training, are managing to get patients better, faster than the average therapy based in the UK. By adding this kind of technology on top of it we are hoping to further improve on that.
Essentially, when a patient goes to see a therapist in the traditional face-to-face setting, they go in the room and close the door, and nobody knows what happens there. This means you can’t take any actions based on these meetings, what worked and what didn’t.
>See also: The next necessary step in healthcare: remote mobile solutions
In our system, all the exchanges between patients and therapists are done via a platform so we can collect all that data, and you can start making associations within patients in this age range, or a particular gender or with a particular condition. This allows us to provide a more reliable reaction for an individual’s treatment with different interventions based on that data set.
As you have more and more data you can start making these models more and more precise. And as I mentioned during my talk, our number of patients has been growing exponentially: doubling every year.
The more this continues, the more data that we’ll have and the better the models we will get. As such, the therapists will see better results in line with the tools we are providing them with. So it is kind of a virtuous circle, of more data and more patients that will lead to better outcomes.
In general, how do you think technology will enhance the capabilities of the health services?
There’s several elements. One of them is the tedious nature of paperwork. No doctor likes to fill in forms and it is also a terrible waste of their time. We have highly trained professionals and we have them filling out forms. A lot of that can be taken care of with automation.
Another thing is providing tools, like diagnostic models that are built from data. Medical science is really good at using data-driven approaches. So, all therapeutic interventions need to be validated for analyst control trials, so we know that they work properly.
But, humans have a limited capacity in understanding data. There’s a certain amount of data that they can keep in their brain at one moment. Computer’s don’t have this limitation, so they can actually apply these data-driven models interactively.
>See also: Software in the healthcare industry: what do you actually need?
There are all kinds of tools you can use to understand the entire literature in a particular field. Imagine having a technology platform that understands medical literature and provides focused searches that are dedicated to particular diseases and treatments within the medical profession.
It will empower the doctor. A tool like this would have access to a patient’s medical history, which would help when prescribing medicinal treatments, for example. It would reduce error, because computer’s don’t make human mistakes.
Computers don’t have the qualities humans have, but they possess other qualities which is they don’t forget things and they don’t overlook things. If you can get the humans and the machines to cooperate, with each using each other’s strengths you would end up with a better outcome for patient care and treatment.
How prevalent are technologies like AI and automation in the NHS?
I don’t know directly. But I know there is a lot of research done in university hospitals that use medical data as a testbed for this deep learning work.
They have been very successful at looking at radiograms. So, looking at cancer identification on X-rays. There was a paper released in Nature, which showed that computers are better than humans at identifying malignant cancer in skin images. Not much better and probably not as good as the best radiologist, but better than the average radiologist. So we already have progress being made in the medical area of deep learning.
What is holding back technology-driven, or data-driven healthcare?
The biggest problem is access to data, and this is a very difficult problem to solve. Machine learning and deep learning in general work better with the more data you have. The problem is, that in the medical area data, patient data is restricted (and with good reason for various personal and ethical concerns).
>See also: How digital health could revolutionise the mental healthcare system
Ideally, from the point of view of a machine learning scientist you want more data. But as a patient it is important that your data is secure. So, there is kind of a conflict of interest here. But, I think we will have to work out what patients are comfortable with sharing and maybe, inform them more about the potential benefits of sharing their data and how we can help.
There is this work at the Moorfields Eye Hospital that uses the Google DeepMind. If this shows positive results then it may cause a more positive view of tech in healthcare.
So, there is a public awareness issue?
Well, a few years ago there was a project run by the NHS where they were trying to get blood patient data shared so they could research it more easily. And there was a public outcry, because either their case wasn’t very well presented or the potential the benefits were not explained clearly enough.
Yes, there are risks and there are benefits as well. It’s about finding the balance point where society is comfortable. The more they share, the more it benefits our work, but it also increases the risk surrounding data privacy.
I would think more people would be concerned about their health rather than their privacy?
It depends what happens to this data. If this data gets used by insurance companies and everyone’s premiums get higher, or people can’t get insurance anymore then obviously this is a bad scenario of data being mismanaged.
It’s a matter of trying to push the envelope and see where people stop being comfortable with it and stop there.
Do you see wearable technology improving patient healthcare from a preventative standpoint?
Wearables have an important role to play and this has become more pervasive. A lot of information can be obtained from your mobile phone. In my field, for example, using GPS to see if someone has left the house or not (for a long period of time) would be a red flag for someone with a mental health problem.
>See also: Improving patient care through technology
There are lots of medical benefits that can be obtained from the data that wearables produce. Again, you have the same problem in that it is siloed. The data belongs to the company that sells to the device. And even if your doctor could benefit from it, they probably won’t be able to get to it.
It’s an area that is moving very fast and I think there are initiatives that will push towards this kind of data being used more. Hopefully we’ll end up with ways of the person actually owning that data and being able to share it with their doctor if they want. Maybe via an app that both them and the doctor, which enables the passing of the data from one to another.
There’s wonderful things you can do with data, but in the real world there are all kinds of administrative barriers that mean you can’t get to that data. It is frustrating. But the real world will hopefully catch up with the research being done at the university hospitals in the UK.
