AI Is The Hype Real? What Does It Mean For My Business? A Healthcare Example
AI has been talked about a lot, but many of us don’t see it affect our business. Review the basics of AI and then how it is impacting the healthcare encounter. Learn the key strategies to help take advantage of the tech advances.
James Bates is an accomplished CEO, entrepreneur, and board member who has created and led high-growth plans for a number of technology companies. Presently, Bates is CEO and Founder of AdviNow Medical (ANM), the world’s first Artificial Intelligence and Augment- ed Reality driven automated medical visit platform. James conceptualized the idea, wrote the fundamental patents, raised $16 million in capital, recruited the team, and launched the company. Today, ANM leads the world in medical automation. Previously, James was an officer at Freescale/ NXP responsible for the $1 billion revenues Analog and Sensor Group the world lead- er in self driving vehicle technologies. Earlier in his career, James founded the Asia Pacific region for Silicon Labs where he was responsible for driving growth that eclipsed $500M in just 3 short years. James serves on multiple boards and advisory boards. Among them is view, the leader in automated tinting windows that incorporates a complete AI managed building automation system. Sophatar is also of note as a location-based screen management/ advertising company that pioneered personalizing the shopping and dining experience. James also serves for Arizona State Universities College of Nursing and Health Innovation. James earned his Master of Science degree in Electrical Engineering, from Brigham Young University. He is fluent in Japanese, lived in Tokyo for twelve years, holds nine patents, and has written in the Proceedings of the International Geoscience and Remote Sensing Symposium, and other technical leading conferences and publications. He resides in Paradise Valley, Arizona, and enjoys family, sports, and particularly hiking.
AI Is The Hype Real? What Does It Mean For My Business? A Healthcare Example
Hello, thank you for taking the time with me today. I’m excited to share AdviNOW Medical AI. As was mentioned, at my previous job I was at a company called Freescale. I did self-driving vehicle technologies. We made the technologies that gave the car the ability to sense the world around you and be able to take action with that world. When you start looking at the opportunities that that brings, of course artificial intelligence combined with self-driving vehicles enables a whole new world that freezes people from a certain location and allows goods to be moved at a very low cost. But in the summer of 2015, we sold that business. As an officer of the company, it was my time to retire. As I retired I started looking at other places to go in and I decided to invest in healthcare medical practices.
Introduction. In healthcare practices, there were a lot of issues. One of the greatest issues of our time is how do
you take care of people and give them health care that they need? However as I was looking at the business of healthcare, I couldn’t bring myself to invest in it. It has a lot of different issues associated with when actually making money, so if you’re a surgeon and you do surgeries all day you make a lot of money. If you’re a health insurance company – you make a lot of money. If you’re a doctor just seeing patients, you don’t really make very much money and that is the whole fundamental problem with healthcare today.
I started looking at this as an investor and I couldn’t bring myself to invest. But then I took a step back and I thought, what if we use the same concept of automating around a driver in a car and we automate around the doctor in a clinic and we completely automate that experience, will that actually deliver and an experience for patients and doctors inside the clinic that people would enjoy and more importantly, does the business of healthcare automatically become possible for everyone? And the answer is yes, it does. So that was the foundation of AdviNOW medical and
how we got started.
What I will be doing today is going through and talking about artificial intelligence – what
fundamental artificial intelligence is and then give the example of healthcare and how we ended
up where we are today.
Gartner Hype Cycle for Artificial Intelligence, 2019
This is what’s called a hype cycle from Gartner and this actually uses the hype cycle for artificial intelligence. There is a tremendous amount of hype around artificial intelligence and this is 2019 in what you can see here is that there are a lot of different technologies which we hear a lot about. It starts with animation that triggers a possibility and then there is a peak of inflated expectation – that’s where people hear so much about these technologies and they have all great
ideas. Then they realize it’s a lot harder to do than what we think and it turns into the trough of disillusionment. As you move farther up, technology advances. They become more mature and that’s when you have the slope of enlightenment and where you can actually start seeing these technologies move along. Next you have the plateau of productivity where the technologies are meeting their full possibilities to be used in society.
Looking at this graph you see some things like GPU accelerators sitting in the plateau of productivity. Why? Because graphics processors are everywhere. They’re in your phone, tv, car. They’re essentially all parts of our economy. Thus, they are in that plateau of productivity. We’re not expecting them to change much but they’re now a part of what everybody does. Speech recognition also kind of falls into that category. Natural language processing – not really, as you can see it is still sitting between peak of inflated expectations and trough of disillusionment. Ultimately, the computer being able to understand by dictate, not interpreted, but I’d be able to do a speech to text or text to speech is very much mature technology. So if we go back and we look at other technologies that are out there, what you see is edge AI which people talk about a lot that’s still sitting in the innovation trigger because gpus have become so cheap and so prevalent. But what you see chatbots reading the peak of inflated expectations where everyone is expecting
the chatbot to fundamentally change the world when really chatbots are just starting to be used. Natural language processing, self-driving vehicle, robotic process automation, computer vision – people are basically tired of hearing about this stuff and that’s why it’s sitting in the trough of disillusionment. This is what everybody here is about.
What is Artificial Intelligence?
But really, what is AI? When you start thinking about it, artificial intelligence is a number of things but it is not any one of these things alone. This is the very center. You have to have technologies. This is a physical enablement
Technologies Physical Enablement
This is like a computer, a GPU. There is a platform with some type of user interface, there are API’s associated with it. All of this is the physical piece of any artificial intelligence system but it’s not AI on its own although many companies claim that it is. Then you have the ability to reason. This is an if and that – some type of decision tree, maybe some simple AI logic that will predict an outcome, get data. But this in it of itself is not AI either. Now you can’t have the
method without the technologies.
Then you have the ability to learn. Many people think you have the ability to learn, you are AI. Well, yes and no. In order for a system to be able to learn, you need to have technologies and the methods. But if you can not communicate with the outside world where you can engage and observe, then you do not have the AI as isi defined. Artificial intelligence, what is it really? It’s the ability to sense, reason, engage, and learn. All of these items are what AI is all about. If you want to engage this is natural language processing, computer vision, voice recognition, robotics and motion, planning and optimization, and knowledge capture. All of this is engaging with the outside world. Learning has supervised and unsupervised learning. Why does it matter to you?
Supervised and Unsupervised Learning
Supervised learning is when you take a computer, feed it data, tell it that this data is similar and the computer then says, it does pattern matching, if it’s similar it must be the same. If you input a picture and define this picture as an apple and you write “these are apples,” put in a whole bunch of apples and then you feed it something that it doesn’t know and it says, “hey it looks like an apple,” it recognizes -hey we are an apple. That means an expert has essentially defined what the computers see and then has spit out an outcome.
Unsupervised learning is a little bit different. This is where we have input data. That input data is not defined by a human. It is fed into the model. The model actually will extract the things that are similar among these similar images. Whether it’s an apple, peach, or banana, it’s all put into its own little characteristic mapping within the neural network. This model will know this is this. We don’t know what it is but it looks the same as all of these so it goes here. Thus, a human is not an expert in what it is, it’s just pulling out things that are similar. Often you’ll hear a term called deep learning Big Data. This is often what they are referring to most of it is unsupervised learning where they’re looking for matches and characteristics that they didn’t know were really there and that they believe will help run businesses more efficiently.
Another concept which I think is kind of important is how does a system learn? In this example you have a nice little brain. This is going to take an action, it takes it out to the environment and let’s just call it a dog. You would go to the dog, tell the dog to sit, maybe you push on it’s bottom a little bit. The dog sits. The dog understands that then you define a reward. You get it a treat. The dog wants the treat. Thus over time as you do this multiple times and you give it food, tell it to sit, a dog sits then you see the dog sit, then you give it food – this is positive reinforcement. This is a very easy concept. We all use it with our children, pets and pretty much in everything we do. There’s positive and negative reinforcement that we use. A computer does this with mathematics. It either strengthens the correlation or weakens the correlation based off of an outcome of the environment on an action that was taken.
When we look at it within the framework of a neural network, we can get to see exactly how this works. In this example you have an image of a cat and an algorithm like tensor flow that will take this image and it will break it into different layers that have different similar traits. They don’t know which traits it’s grabbing on. It’s just looking for traits that are similar among a number of different images. It can be as big as the color of the cat, the color of the background, or as small as the cat knows. If we grab this little layer we can say wait a minute, this layer has the very tip of the cat’s nose, the next layer is a little discoloration after the cat’s nose, and next layer is that little area between the cat’s nose and the mouth, and the last area goes into the cat’s mouth in the bottom. This demonstrates the level of accuracy in the level of detail that these models are pulling out as it is comparing images. Now this is done without human intervention. Generally humans can influence what characteristics the classifiers are created under.
What’s happening as each one of these processes moves through, the more pictures these get stronger or weaker. This is reinforcement learning as it’s putting it in mapping in an unsupervised way. The amazing thing here is that intelligence is everywhere. It is obvious to us, maybe it’s robots, drones, or autonomous vehicles, something we can touch and feel in the physical realm. Maybe it’s obvious but it’s virtual like a chatbot, virtual assistant, or smart advisor. Things we all see today. Maybe it’s not so easy to see like an intelligent sensor. Maybe it’s your ring doorbell that recognizes when people come in you don’t even know about it but the ringing remembers it and puts it into their system. Maybe it’s in the very back end of your Smart Security operations that’s learning about attacks real time by tracking the IP addresses that are hitting your website.
This AI is being used in almost all aspects of our life today and it will only continue. Now us as entrepreneurs or advisers as board members of companies, our job is to make sure that we are early adopters so that we do not lose our advantages as other companies start automating pieces of their work flow. I’ll just go through a simple example of healthcare.
The interesting thing with healthcare is everybody knows what the problem is. Like the guy going to see the doctor with a big arrow shot in his head and the doctor goes up to him, sees that massive hole in his head and just says, “what seems to be the problem sir.” This guy’s looking back at the doctor and saying, “well everybody knows I got an arrow in my head.” In the United States per capita spending and Healthcare is just skyrocketed out of control. It’s got
to the point where the average family is spending $20,000 a year in healthcare maintenance when you’re paying $1,500 to $2,000 a month for insurance. That’s when you know something is fundamentally wrong. The problem is that ultimately clinics are going bankrupt. You see them close all the time and there’s a shortage out there at the same time clinics are closing. The reason is because they’re not making money and the system is not allowing them to make money ultimately the people who are paying the price are the patients if you live in a large Metro like Boston. You can be waiting up to two months to see your primary care doctor and depending on the type of specialty you’re waiting up to a month when you start looking at any of these cities throughout the United States.
We hear politicians talk about this all the time and on one side of the aisle you have the people that say the government should pay for it all – they’re the solution. On the other side of the aisle, you hear people say that the government was the problem just let anyone do medicine and it will be fine. The reality is neither of these are the solution and we know that which is why it hasn’t been solved. The only way we can actually really solve the problem is by getting to know the stakeholders, getting to know why the motivation is there and then using technology to deliver a true solution to the marketplace. These competing motivations weather payers Health Systems providers are not enemies but in our system today they compete against each other to grab that Healthcare dollar. How do we approach this?
Using artificial intelligence and augmented reality can automate the whole patient experience eliminating waste today. Health care clinics essentially have a tremendous amount of waste as they have become data collection centers rather than patient service centers. In fact, 2/3 of the time providers spend is wasted and that has created tremendous profitability struggles for the clinics that are out there in the marketplace. Whether it’s documentation that they have to collect, getting the right billing codes, or whether it’s treating patients over virtual consultations. All of these are leading to a tremendous amount of wasted time and that ends up creating the problem that we have.
The AdviNOW solution has resolved these issues by two main applications. One that faces a patient, and one that places the doctor. On the patient base in application we use artificial intelligence and augmented reality to collect all the information from the patient that a doctor needs to make a diagnosis and treatment. On the doctor facing application we present that information to the doctor in a simple one page format and allows them to complete a thorough objective diagnosis in a flash. This is why we say that these doctors have superpowers. These superpowers are two times throughput doctors that used to see through an outpatient an hour that now see seven patients. Doctors at the patients have higher than 90% patient satisfaction near-perfect accuracy and ultimately improved outcomes.
No Industry is Immune: All Industries are NOW “Tech” Industries
This whole process is what is driving the revolution in healthcare. The amazing thing here is that no industry is immune. All Industries are now tech industries. Whether this is in a hospital like I just talked about or whether this is in a car, driver’s, in a factory being replaced by robots. Artificial intelligence is here to stay and for us as leaders in the technology industry and in industries that maybe aren’t traditionally technology-savvy and we can go into construction
Industries, or farming industry. Things like this, they now are technology center industries. I would invite all of you to go through and create activities for you and your own industries to separate out two types of tasks. One which we call a repetitive task identification. These are not complex tasks but they are repetitive tasks that a human is doing everyday. Many times a good example of this would be paying invoices.
Groups of 2-4
● Repetitive Task Identification
○ RPA (Robotic Process Automation)
● Complex Decision Identification
○ Automated around liability owner
○ Machine learning
● Learning Data Identification
○ What can be used for expert or supervised trained data
○ What can be used for unsupervised data
AdviNOW received invoices in the mail, over email, and through other applications such as QuickBooks. Using robotic process automation, we’ve developed a system that reads all of these invoices and automatically populates them into our accounting system and a human just comes in and reviews the accuracy. That is the type of robotic process automation that we as business leaders must always do.
The second piece is complex decision identification. That is automation around a person so when you think about your business, you’re going to think about one thing, that one person that if you had 10 of those people, your business would just take off and be super profitable. Then you think about everything that that person does that essentially somebody else could do. Now a lot of times you get people to do that. What we’re talking about in these types of exercises is taking these complex decision identification and automating around that liability owner.
In AdviNOW’s case, we are automating around the doctor, machine learning, and artificial intelligence. This system, how this works is exactly how the adrenal system will make doctors more efficient. We eliminate the need for them to ask a patient a bunch of questions and the need for them to actually document that information into the EMR. AdviNOW system helps the doctor see more patients not by making them do more work but by eliminating the work it has less value for them to do.
I would ask all of you to go out into your own companies and find a complex decision maker that you can look at, see if you can increase the efficiency, and make them better at what they do. Once you’ve done this, then you have to identify the data that you could train using supervised and unsupervised learning to be able to help automate around that liability. This is how AI is changing the world. This is how startups are revolutionising industries. Whether we’re a startup founder looking for something to do or we are an actual business owner trying to make sure we’re competitive and take market share, this task of identifying repetitive tasks and complex decision-makers is required of all of us as we are entering the new age of artificial intelligence. To the answer of the very first question; Is artificial intelligence hype real? The answer is yes; it is very real. What does it mean for my business? Ultimately your business will be changed by it and it will either be changed by you or by your competitors who are eating your lunch. My name is James Bates . I am from AdviNOW Medical. I am open to any questions you may have. That is the end of my presentation.