Panel interview with senior members of some prominent healthcare AI startups #GiantHealthEvent

Data Science representatives from Eagle Genomics, Babylon Health, Touch Surgery, Myrecovery, Creation were interviewed by Cognition X.

All these companies use machine learning in different ways either to analyse the genome, diagnostic/surgical implications or helping patients directly.

All of the panel agree that AI is coming back after going through a ‘nuclear’ winter.

download-4

Eagle Genomics feel that by 2025 there will be over 250 million patients with their genomes sequenced. This constitutes several zetabytes of data. This cannot be analysed by humans but only machines. We have a burgeoning data management crisis that can be solved in no other way. We need to turn big data into actionable insights.

download-8

Babylon health say they want to make healthcare affordable to all – that is their vision. They believe in augmenting doctors and making them safer. Babylon feel the biggest need is in Africa. That is why they are working in Rwanda because the needs are so great.

They think that in the next 5 years there needs to be some sort of regulatory mechanism to govern the use of AI diagnostics. They want to improve the productivity of doctors not replace them.

download-5

Touch surgery feel that humans alone cannot deal with all the data alone. They feel that it is essential to use that data in order to learn how to make their product as good as they can. Currently they are trying to use machine learning to improve the user experience and training. They believe that the technology should also help the surgeon in the same way that power-assisted steering and GPS enhance the driving experience.

download-3

Myrecovery are analysing how their users are using the apps as well to try and predict their recovery. A lot of people complained about the US election predictions. Garbage in = garbage out. Longer term they see it as a core asset to the business. They feel that the present situation is unsustainable. We can’t compete with robots any more, we need to work with them in order to get the job done.

download-6

Creation feel that it will be essential in the future but at the present are focusing on building data sets and getting them verified by medical professionals. They talked about developing a system that could analyse a photo via a network – like instagram for medical diagnosis. They feel that doctors use social media quite a lot. Sometimes they have found doctors answering individual patients on social media. They feel the barriers are breaking down. They also sited the microbiome. He talked about machine sentiment analysis and how it is currently largely useless. Most ‘AI’ is still heavily dependant on human interaction to make it work but in the future this won’t be the case.

Then Charlie asked the panel what advice they would give to tech startups thinking about working in this field. Their advice was:

  • Get good at selling yourself.
  • You need to be in it for the long haul. (We are in the middle of a revolution).
  • Be prepared to change your business model several times.

GE: New Thinking in the Digital Health Ecosystem: Breaking down the silos #GiantHealthEvent

No company or individual is a silo. The internet has changed everything and it is all now connected

download-6

But what will happen when 50 billion machines become connected? Suddenly the data will show us the flow in the hospital, the results will be highlighted. The computers can detect problems and fix them.

seattle-childrens-hospital-turns-big-data-into-better-care-6-638

This is the collision of the physical and analytical, brilliant machines with industrial amounts of data and people interacting with them.

We are heading towards a ‘colossal clash’ between the consumer health technology and clinical healthcare. The two worlds are both merging but also on a collision course with one another.

accenture-colossal-clash-digital-health-disruptors

We are heading to an age where we want to rate everything. This is a consumerisation of healthcare. It will become the norm to rate your doctor online.

In the future nothing will be redundant. Everything will be rankable and in flux. We are heading towards an outcome driven world.

download-7

This is why GE are creating an open ecosystem cloud to host health applications because there is going to need to be a way to scale innovations quickly. This can provide a portal for developers to test their products quickly and get feedback.

There is no going back. Things are moving fast and GE want to be ready.

The Fourth Industrial Revolution

the-fourth-industrial-revolution

Shafi Ahmed then introduced the topic of the ‘fourth’ industrial revolution.

industry_4-0

The blurring of the lines between the physical and digital spheres.

It is about thinking differently in order to solve hard and complex problems. It will involve the interface of medicine, digital technology, societal changes, art and many other things.

It will involve:

Virtual reality

random42-oculus-gearvr-headset

Big Data / Informatics

images

Blockchain, IoT

blockchain

Machine Learning / AI

download

This is just what we know the fourth revolution will involve. There are many other things as yet undiscovered which will be involved.

The UK’s first ever exponential medicine conference

screenshot-2016-11-16-at-09-18-24

Welcome to GIANT

This marks the beginning of a shift in the mindset of health technologists in the UK. There have been other events like the NHS EXPO but their purpose was very different.

This conference as illustrated by the programme is going to be very different.

This is exponential medicine conference, UK.

discover-3

The speakers at this conference are not your usual health technology conference speakers. I am currently sitting in the middle of an audience full of outside thinking, border pushers within healthcare. A mixture of professionals from medical, business and technology spheres all in one place.

This is the infancy of a small silicon valley type community for health technology in London but it has none of the glitz and glamour of exponential medicine conference. This has much more the feeling of the first DEF CON in June 1993 (I was not there, I was only a child then but I imagine this is what it was like.

Barry Shrier, the Founder started by introducing the term moonshots. These are radical innovations which lack any immediate method of becoming financially successful but have enormous potential in the future.

screenshot2015-07-31at7-28-55am

The challenges that face us in healthcare now have never been greater. Therefore, the need for a ‘moonshot’ or more likely many smaller ‘moonshots’ to succeed has equally never been greater.

This is the interface at which they occur:

00000002326

Barry then talked about the ongoing impact of the NASA program. It is his conviction that we can achieve enormously big goals and we should set out to achieve them for the good of the world. The vision: “To improve the health and wellbeing of humanity, by supporting entrepreneurs and supping healthtech innovation.” He dedicated his presentation to the late Helen Keller. “Life is either a daring adventure, or it is nothing”. H.Keller

It is the spirit of ‘I can do it’ that will enable us to overcome the enormous challenges that humanity faces.

Liver Imaging for Gastroenterologists

Disclaimer: This article is not intended as medical advice for patients but is for informational purposes only, primarily aimed at clinicians.

357127cf4b242ed836e6ad39dcd6b979

The liver is supplied by multiple vessels and divided into 8 segments.

In the normal liver most of the blood supply comes from the portal vein.

Tumours however tend to be supplied by the arterial supply (more on this later).

Regarding imaging. Ultrasound is great but unfortunately results are not reproducible between different operators.

o0006sup3dt1C

Common benign findings include: Cysts

liver_cyst_ultrasound

but sometimes they are not so simple

hyadtidThis one above turned out to be a hyatid cyst (Echinococcosis).

haemangioma

Haemangiomas can be even more difficult to distinguish.

focal-nodular-hyperplasia-i15

Focal nodular hyperplasia (above) can be distinguished from hepatomas by using Primavist (came out 2004). This contrast is taken up avidly by healthy hepatocytes and not by cancers.

Diffusion weighed MRI can help to distinguish between small metastases and other lesions, however its effectiveness is limited.

fatty-liver

Above is a fatty liver which makes it very difficult to distinguish the liver parenchyma

hepatitis

This is what acute hepatitis looks like on USS. The liver looks ‘brighter’.

The radiologist can also see vessels such as varices, abdominal collaterals, reversed/blocked portal flow and associated ascites etc.

Cirrhosis can be seen but only at an advanced stage so these imaging modalities are not so useful for the early detection of liver disease. Of CT, USS and MRI – MRI is the best at picking up cirrhosis but it is also the most expensive and difficult to access test.

Hepatocellular Carcinoma

The above conditions are all benign, but what we really want to spot are the cancers. These primary cancers normally occur in those with pre-existing liver disease.

Diagnosis can be very tricky if the timing goes wrong which is why you need a skilled radiology team.

This slideshow requires JavaScript.

Above are some images of HCCs. They tend to be surrounded by areas of necrosis as the tumour develops a rim of dead tissue around it.

Treatment options include:

This slideshow requires JavaScript.

Sorafenib, TACE, transarterial chemoembolization, resection and transplant

This is where we are now in 2016 but there are many other investigation and treatment options under investigation. So we can now tell what’s probably going on without doing a biopsy, most of the time. There is still significant margin for improvement.

Next up: Liver disease in pregnancy.

More interesting tech at NHS Expo 2016

Lifelight

This is a really cool sensor system that is brilliant ad it is so simple yet massively effective. The founder Lawrence, is a legend and is doing great things via Southamptons catalyst programme. Watch this space! If I had money to invest I would invested in this.

XPERYE

Used by Shafi Ahmed in London this is a neat peice of kit using Google glass to transmit the wearers vision to a computer. The images can then be annotated or altered to be used in teaching etc. With the Google glass version 1 no longer available and a time lag for its successor and at £300/month this is potentially a very nice little training tool in the right context at the moment. It is certainly fun to play with.

PA consulting’s offerings 

PA consulting have introduced these products at the expo. This is a new probe for measuring Barretts oesophagus through the scope. It was next to the cytosponge but for some reason that had disappeared today. Both very promising innovations in my specialty (gastro) for Barrett’s surveillance.

TEDx NHS

TED has finally landed in the NHS. This proves people are gradually catching the vision. Bring it on…!

New tools to market at Expo 2016

Creavo
This beauty could revolutionise the admission of chest pain. It can more accurately diagnose ACS AND unstable angina in just 3 minutes in the ED. It’s currently going through studies to validate it but I can see it being very successful as it’s price point is only around £100k and it could save NHS trusts millions.

Creavomedtech only just formed. Based in Leeds with a strong support structure, good corporate backing and strong technical expertise linked to the university I think this is one to watch!

Early Sense

This platform by early sense could be useful for patients with dementia as a way to monitor them remotely. However, there are lots of cheaper bed alarms. At £3500 a bed I think this products USP is the monitoring rather than anything else.

Vanguard

This is not a new technology but it’s beginning to scale. Portable theatres and endoscopy units could help some hospitals who have capacity issues. This polentally also has a role in major incident management.

Centrihealth

This is a context based health platform. More advanced than a normal EPR (electronic patient record) system. This uses AI to filter the data and produce a useful record. This tool is FANTASTIC, but they have not yet broken into the NHS because their marketing strategy is wrong. It took me ages to get what they are actually doing and I think thats why they are struggling here. THIS IS A PATIENT SAFETY TOOL NOT A ‘CONTEXTUALISATION’ ONE. If you read this then get in contact because I think I can help you.

There are many more innovative solutions here but these ones caught my eye. To see more visit the EXPO website.

Novel Imagers / VR kit

In the last couple of weeks I have been contacted by several people who have designed various gadgets for Medical VR/interaction. They are iridescent imaging, VIZR, D-EYE and Gesturetek. 

First up Iridescent Imaging 

Mitch Downey founded this company to create fantastic small and life-size VR projections.m  see more images here: http://iridescentimaging.com  The technology looks fantastic but as you might expect it isn’t cheap. I can really see how these might be useful for virtual clinics, particularly in specialties where walking in front of the doctor is a key part of the exam: ie. Rheumatology / Neurology. However, a remarkable product needs a great website and this is something that needs a bit of work. 

Then there is D-EYE

This thing looks excellent. It can effectively enable an ophthalmologist to diagnose a child remotely in another country using just their smartphone but more likely it will be a great clinic application. To be honest I want one on my emergency unit as it looks so easy to use but again these things are new and therefore costly. https://www.d-eyecare.com

Next up VIZR

This is more like a clinician assistant (like Google glass) they feel their user interface and information is more up to date. It looks like a smart gadget but as the video on the site demonstrates it is not entirely unobtrusive to the patient looking at the doctor http://www.vizrtech.com

Time will tell whether there will be enough uptake. My feeling is that unfortunately for this device we are still several years away from IoT and the effects of Big Data having their full impact to help this device work. 

Finally we have http://www.gesturetekhealth.com

These guys are doing something a bit different. Trying to get physical controls to work for rehab /those with disabilities. Unfortunately the website is dire which is a shame as they have some really cool products like Irex-an upper and lower extremity training system. 

It looks like they are doing a great work in their niche. 

Note I have just started a medium blog as well and plan to integrate them later so watch this space. 

Telemedicine: Is it the future? 

​What surprises me most about the article below is the so far low uptake of Telemedicine. The applications for this are potentially vast but due to various restrictions particularly surrounding the reluctance of US health insurers to fund such consultations, the general uptake is still low. 

Several companies in the UK are trying to build these primary care alternatives now (Dr Now, Babylon etc) and it would appear that there is public desire for these things. However, they face lots of challenges. 

Firstly patients are offered more comprehensive services in the NHS even if there is a delay to access them; tests are very difficult to organise at home over the Internet; Patients can’t be fully examined and fundamentally there is a major shortage of GPs. This means that capacity is unlikely to ever be able to meet demand in a cost efficient way without adopting alternative models of triage. 
These challenges can all be overcome in time but the fundamental test will be what patients actually want. As physicians we think we know the answer to this question but have we really asked the right questions? 

Link: http://www.wsj.com/articles/how-telemedicine-is-transforming-health-care-1466993402 

Is this the future of patient centered healthcare records?

https://www.patientsknowbest.com/ is a new resource for patients to keep complete control of their records in a secure way. It seems like a fantastic idea but the key will be uptake. Do enough patients feel empowered enough to take control? or are we still living in too paternalistic an age?