The company will provide patients AI-powered conversations about their healthcare
The way we consume healthcare has changed exponentially in the last decade, mostly thanks to the smartphone. The ability to track our own health through apps and wearables has allowed people be in charge of their own care like never before, and there is an expectation of getting information quickly and easily.
At the same time, there is more health information than ever before thanks to advancements in genomics, which makes it possible to analyze our genes, or basically our DNA, which make up genes, to understand our physiology. All that data is almost more than doctors can handle.
So, we have patients who want to have their questions answered on-demand, and doctors who are already overburdened by data. There’s a solution that can fix both of those problems at once: advanced natural language processing technology, or an AI-based healthcare platform that can actually converse with patients, and answer their questions.
Enter doc.ai, a company that emerged on Thursday after a year in stealth mode on to provide just that, by building medical dialog systems and conversational AI for personalized healthcare.
"With precision medicine, and personalized medicine, coming along, with genomics and others, they have given us a very accurate value of our biology. We have a quantified biological profile, but it's getting to be too much for doctors without the use of machines. Doctors are overloaded with homework they have to do, and it's become an increased burden on the healthcare system," Walter De Brouwer, founder and Chief Executive Officer of doc.ai, told me.
"On the patient side, what we did in the past was bring digital into healthcare. So now we have mobilized health and telehealth, but that's no longer sufficient. We will be previewing one billion genomes by 2026, and we're running out of humanity to explain all of these things."
How doc.ai works
What doc.ai provides is to use an artificial intelligence natural language component to essentially become a robo-health monitor. It will be able to take that genetic data and provide patients with support, converse with them about on a number of different health topics including diseases, traits, pharmacogenomics, and family planning.
For example, if a patient has lab work done, they would be to go online and see the results and, instead of waiting for a doctor to call them, they can immediately ask questions about what the results mean and what they should do next. The platform will be able to see if their glucose or cholesterol is in range, and then check which medications they are, and make recommendations regarding dosage and follow-up treatment.
This, said De Brouwer, is essentially what a doctor would do, but rather than replacing the doctor, doc.ai would be able to free up the time spent on those phone calls
A platform like Doc.ai can help reduce the cost of healthcare simply by taking the human element out of the service.
"To take a full sequence genome, the cheapest you can do is for $1,000, along with genetic counseling, which costs $200, and which patients normally pay out of pocket. If you can do that in telehealth, it could probably get lower than $100, maybe even $60 or $70," said De Brouwer.
Eventually he sees it potentially costing a little as $1 for a year of unlimited visits since there will only be a silicon-based entity that doesn't need to be compensated.
The company currently offers three products:
Robo-genomics, a conversational agent designed to improve comprehension of genetic data and provide users with decision support.
Robo-hematology, which is designed to answer any question on over 400 blood biomarkers. It’s trained on hundreds of thousands of medical documents and common FAQs.
Robo-anatomics, its patent pending Selfie2BMI module uses state-of-the-art Deep Neural Network and optimization techniques to predict a variety of anatomic features including height, weight, and gender from face.
Doc.ai, which was founded in July of 2016, and has raised an undisclosed amount of seed fundng. It now has its first channel partner, Deloitte Life Sciences and Healthcare, which is working with doc.ai to test its Robo-Hematology solution.
A kind of healthcare chatbot
If the above description sounds sort of like a chatbot, De Brouwer wouldn't disagree. In fact, he classifies doc.ai as being the third level of advancement in conversational AI.
The first are the chatbots that most of us use, the ones on Facebook Messenger that allow us to reach customer support. Those involve a lot of pattern matching, and doesn't have much intelligence.
Second are virtual agents, which do agendas and scheduling. These, De Brouwer said, understand a bit more than the first category, but also have to use supervised methods.
The third category are platforms like doc.ai, which use neural nets, and which need a lot of data and they need a memory, so they remember who they are talking to. They also have to be able to look for information unsupervised so they can answer patient questions. Unlike chatbots for customer service, which are pre-programmed to answer the most common questions, doc.ai has to answer questions such as, "How is my health," which involves a much more complicated answer.
The fourth type of AI would be something akin to a personal assistant, something like J.A.R.V.I.S. from Iron Man or the AI in the movie Her.
The future of AI in healthcare
While there's currently a lot of AI in healthcare, there's not a lot of language AI, or conversational AI, in healthcare, De Brouwer said, due to regulatory problems and issues with compliance.
Eventually, though, he believes that we will enter into that fourth level of AI, where everybody will have their own personal trainer. And that stems from the ability that people now have to take greater control of their own care, with greater knowledge and education about how to take care of themselves.
"A few years from now the collision of the two new paradigms, personalized health and the comprehension of machine learning, will come to the point where you'll be able to train your own AI, your own machine. What I have in mind is a mathematical construct of neural nets. They will know everything from what vaccines you have, what medications you're taking, how active you are, what your conditions are, your latest blood result, and it will be able coach you a bit. Everything will happen in your smartphone. You'll be passing a Walgreens and it will tell you that you need a flu shot," he said.
The ultimate goal with doc.ai, it so create a world full of knowledge about healthcare, where people have the ability to learn and take care of each other.
"We want to provide 24/7 access to quality healthcare so people can learn and help others. It's a bit like Uber, where they found an oversupply of time and cars, or Airbnb, with rooms in your house. There's a healer aspect in each of us, we just don't know how to use it. Wherever we are, there's someone in that room who knows someone who is sick, and they cant do anything. Powerlessness is a very poignant emotion and one we want to get rid of," De Brouwer said.
"Younger kids are interested in health so they can take care of their grandparents. All that knowledge is out there, everywhere, and we are integrating it, packaging it, so we can use it and make it easy."