AiThority Interview Series With Ron Wince, Founder and CEO at Myndshft

Tell us about yourself and the journey that led you to start Myndshft.

Our goal in starting Myndshft was to make artificial intelligence more accessible. As an example, despite the high level of optimism about what AI can do, less than 5% of healthcare organizations have active AI projects underway or in operations. With new technologies like AI and blockchain we have an opportunity right now to transform industries like healthcare. Just like Microsoft was started with the mission of putting a computer on every desk Myndshft is driven to demystify AI and make it as ubiquitous as PCs are today.

How open has the healthcare industry been to adapt to using AI to better function? What’s their biggest concern here?

Many organizations in healthcare don’t have experience in using machine learning and AI. They still see AI as a black box. To take the mystery out of it, clients can and should start on the business side of healthcare and away from the clinical and patient-facing processes. By focusing on the business and administrative processes, organizations can see pragmatic, measurable impact and AI becomes more tangible. Once they see this kind of impact, they become comfortable with expanding the application of AI in other areas of their organizations and progressing closer to the patient.

What kind of tasks can Myndshft’s M:IATM help manage?

M:IA is Myndshft’s intelligent process automation solution which was built on top of CognitiveBus, our cognitive blockchain platform. M:IA was developed as a jumping-off point for organizations to being able to immediately leverage basic artificial intelligence while also capturing a near immediate return on their investment.

Currently, we are using M:IATM for automating various routine and repetitive administrative tasks that also have higher levels of complexity and needs for data and machine learning. As an example – we are working in specialty pharmacy where simply getting an authorization for a needed medication is challenging and complex. M:IATM is able to capture all of the needed clinical and health benefits information and submit it for a prior authorization. Once submitted, M:IATM can then monitor for updated statuses and predict the impact to patient care or reimbursement. In this example, the process automation captures efficiency while the application of machine learning allows the organization to anticipate an outcome of the authorization request and react accordingly.

How does M:IA’s technology emulate the human thought process? How is this used?

M:IATM typically relies on the same business rules that people follow for basic, repetitive business processes and combines that with a machine learning model or cognitive service (natural language processing, image recognition).  Throughout the execution of a process M:IA is continuously monitoring the outcomes of the process and augmenting decisions of people or altering the process itself to make to achieve better efficiency and effectiveness than a person doing the process alone.

Machine learning in its simplest form emulates how a human learns; it completes a task, it monitors the outcome of the completed task (was the outcome successful as expected or was it a failure) and then the technology can make a refinement to the process. Currently, you are seeing this technology applied to a fairly narrow task or set of tasks.

What is the scope of the data that one collects at healthcare facilities like hospitals, and how can it be leveraged to provide better aid?

The data that can be collected at healthcare facilities is fairly broad but in most cases, only the data that is needed for the specific task is collected and prepared for use. Unfortunately, this step can take up to 80% of the actual work of leveraging machine learning. A high percentage of data in healthcare is in free text and is highly dependent on the originator of the data – such as a doctor who takes notes in a manner different than another doctor.

At its core, AI is using the data to make predictions and then checking that prediction against the actual outcome. As it completes this cycle, it becomes more and more accurate.

AI is one of those technologies that, because it continuous to improve over time, it becomes more and more valuable. This is why it’s critical for companies to start using the technology now versus waiting for it to be forced on them.

How does using Myndshft’s technology have an effect on ROI?

Our current clients’ ROI on initial business process projects is very high. Most projects are self-funding – meaning they have recovered the original investment cost and delivering a return to the client – within 6 months or less. This is focused on the business of healthcare where quantifying impact is much more straightforward and measurable.

What kind of technology training does the staff at healthcare facilities need to use MyndShft’s offering; how does this measure against freeing them from repetitive tasks that can now be automated?

In actuality, almost no training is required for most of the organization beyond the processes changes that happen as a result of deploying M:IA or a machine learning model. We’ve built our technology to allow organizations to continue to use their current applications, systems, databases, and workflows so the machine learning disappears into the fabric of the process itself.

Typically, most training is around moving on to other tasks because the technology is relieving a person of steps in the process – from simple data entry to augmenting decisions they make with the output of the machine learning.

Can you help us understand the OCR and Image recognition capabilities of M:IA?

Like most machine learning applications, it is really a combination of 3-4 different technologies bundled together. An example is a fax – still a common way to submit patient information in healthcare. When a fax arrives at the receiving party it is typically an image – the first step is to use a visual tool and break the image into small parts. Then, each part is examined by another technology that recognizes the letter “patterns”.  From there, another technology can recognize the words that came through on the fax and write them to a field in a software application. At the start of the deployment of a solution like this the technology is “trained” by human “supervisors”. In our experience, the technology usually starts with a lower level of accuracy – just like any new employee – and then improves quickly through more processing cycles.

How does Cognitive Bus leverage Blockchain technology for the healthcare industry?

In a couple of different ways; the first is providing what’s called ‘metadata’ in an auditable, unchangeable, encrypted format. Essentially the metadata is “broadcast” via a block to the participating nodes – or participating entities… such as a single computer or an entire organization – on the blockchain. The block is approved or validated based on rules that were established when the blockchain was launched. The block then becomes a permanent encrypted record on the chain.

This technology allows us to do everything from executing a transaction – such as submitting a bill between authorized parties on the blockchain – to facilitate the exchange of data that can streamline clinical care or care coordination.

In our use cases, we are using private, permissioned blockchain technology versus the open blockchains that are covered in the media. Public blockchains support cryptocurrencies like Bitcoin while private blockchains are much more relevant for enterprise applications in healthcare and other industries.

What other enterprise software does Myndshft’s technology integrate with?

The Myndshft platform integrates with major electronic medical records systems, customer relationship management solutions, enterprise resource programs and we are constantly adding more integration. We are also able to integrate with legacy systems and applications that allow our clients to extend the value capital invested in their IT systems while also modernizing those systems with machine learning.

The ability to modernize current systems without having to make capital investment or rip out and replace systems is a critical element to increasing the adoption of AI for cash strapped organizations like most hospitals.

What are the privacy concerns that exist with regard to patient data and what are the precautions that Myndshft takes to combat this?

Personal data privacy is a big topic; and in healthcare, it is even more significant. Currently, we use the highest levels of data privacy and security and have proprietary encryption methods. We’re also testing a methodology called “homomorphic encryption” where we are able to perform machine learning on encrypted data, which brings an additional layer of security.

Finally, we are leveraging blockchain for making de-identified data accessible and we are also testing decentralized machine learning – meaning data remains stored on its owners’ systems and the machine learning model is distributed versus having to transmit data over the internet where it is most vulnerable.

How do you see the healthcare industry evolving with the adoption of AI technologies?

AI is the future of healthcare.  Automating basic processes is a great starting point, and we’re getting more and more requests from clients to jumpstart their journey with AI in those areas. That said – I think it may be another 5 or more years before it will be widely accepted on the clinical side. For now, AI has an immediate opportunity to free up time for clinicians and care managers by taking on the more mundane administrative tasks. Our goal at Myndshft is to allow clinicians and caregivers to focus more time on patients versus spending a high percentage of their time navigating the heavy administrative burden now required in healthcare.

Thank you Ron! That was fun and hope to see you back on AIthority soon.