best practice

The business and science behind AI in healthcare – a Q&A with Peter Morel and Dr. Sergey Morozov

Osimis CEO Peter Morel and Chief Innovation Officer Dr. Sergey Morozov discuss five key questions you may be asking yourself regarding the practical application of AI in healthcare and why Osimis is at the forefront of this intelligent sea change.

The use of artificial intelligence is on the rise across countless industries, and healthcare is certainly no exception. And with more and more hospitals and practices beginning to see the benefits of leveraging smart technology in a clinical setting, we’re seeing more opportunities to relieve the pressure on radiology teams and other departments that are suffering from staff shortages.

Osimis CEO Peter Morel and Chief Innovation Officer Dr. Sergey Morozov join us to discuss five key questions you may be asking yourself regarding the practical application and rapid evolution of AI in healthcare, the digitisation of the industry, and why Osimis is at the forefront of this intelligent sea change.

 

Q - How do you define healthcare data-as-a-service? How do you hope to contribute with your previous expertise and your current role at Osimis?

 

Peter Morel: The application of AI to extract intelligence and capturing knowledge from the documents was initiated about five years ago. The adoption of this new technology into existing business processes providing new operational value is a learning curve and a common trip for both suppliers and customers in which sometimes undiscovered effects allow new dynamics in the organisation. I have a background in electronic document management, archiving, data recognition and classification so there are a lot of similarities with medical imaging data processing.

Sergey Morozov: To answer this question, we need to take a broader view of the transformation in healthcare. Firstly, there’s diagnostics and the data that is essential for the treatment selection.Diagnostics have been empowered by multiple instruments starting with observations made with the naked eye all the way to the use of digital stethoscopes and MRI machines that enable us to look deep into organs and tissues.

These instruments have evolved to support the transfer and storage of digital data, which has led to the creation of new workspaces such as remote reporting of radiology images or pathology slides. Digital diagnostic tools have continued to develop with more expertise shifting away from the hospital premises and into the cloud or internal/external reference centres.

That’s why we’re seeing healthcare dataflows, empowered by these digital infrastructures, facilitate new workflows and new levels of automation. For all of us here at Osimis, the ultimate goal is to increase the availability of excellent diagnostics for patients, enable new digitally liberated workplaces for physicians and allow cost control and scaling for healthcare systems.

I’ve been lucky enough to have been working with these goals in mind for almost 20 years now, moving from a physician role to the executive role while applying digital methods and IT systems for the healthcare transformation. Among my previous results are large-scale projectsof deploying a city-wide medical IT system connecting 1,500 scanners to 100 hospitals, organising a teleradiology centre with 200 full-time employees and 1.5M reports annually and integrating almost 50 AI services to analyse 6 million studies on the fly. With Osimis, I am blessed to have a new opportunity to build another bridge between healthcare and IT industry for the common good. 

Q - When it comes to implementing reliable and transparent medical diagnostics with AI, what is the current state of play?How fast has AI in healthcare evolved and how has Osimis evolved alongside this evolution?

 

Peter Morel: For some years now AI applications for medical imaging have been in various stages of development. The successful AI solutions have focussed on a specific domain as it requires a lot of combined effort to obtain a confirmed result.

Due to varying dynamics in the healthcare market, adoption of AI solutions varies in different regions. For instance, there are the early adopters, especially in university hospitals, and the broader application in some teleradiology centres. There is a growing awareness in these hospitals to introduce AI into their processes to benefit fully from it in a few years. It requires a learning and transition process, and that’s something Osimis has the expertise to support with as a trusted partner. We have the knowledge of the AI market and hospital processes, while helping to pre-qualify solutions for specific healthcare domains.

 Sergey Morozov: AI has quickly evolved from laboratories to independent teams thanks to the availability of publicly available tools. This democratisation of AI application development has rapidly increased public interest and ushered in a deluge of investments. However, the lack of real-life applications has naturally led to a certain level of frustration once the challenges of AI services became clear.

Currently, multiple AI applications are being employed in the hospitals without any real tangible improvements to healthcare quality. That’s why we are approaching the stage when AI tools transition from early adopter interest to the early majority application.

Interestingly, a somewhat chaotic market of AI services representing SaMD (software as medical device) is also starting to emerge. Osimis has gained invaluable expertise in medical imaging informatics and reached a high level of trust from partnering hospitals, making us both a reliable and a trustworthy provider of the best new medical informatics tools.

 

Q - AI in healthcare, in radiology in particular, is on the rise. Where do you see the biggest needs and how do you see them evolving?

Peter Morel: Something that will be very important in the adoption of the ‘automation’ aspect will be the possibility to participate in the training of the AI, building a trust relationship between the radiologist and their new intelligent tool. In the same way you’d look to enhance quality procedures, the result should be periodically verified, and the feedback should be looped into the training process. That’s where radiologists and medical professionals will really build their confidence in these tools and the benefits they can bring to their clinical settings. 

Sergey Morozov: The biggest demand for AI augmentation is in the routine process where there are lots of repetitive tasks. There is an immense number of functions within a healthcare system which represent standardised low-risk procedures with a huge cost-saving potential. These include data collection (scanning in vivo and in vitro, recording physiological signals, collecting subjective data from patients) and analysis tasks (such as increasing signal-to-noise ratio, looking for patterns, identifying predictors of deterioration or an emergency).

All of these can be augmented for a healthcare professional, thus enabling that specialist to focus on more complex tasks instead of performing those that feel like chores. Building on these smart applications of AI, the reliability of healthcare services naturally increases, the communication between clinical departments improves, and patient compliance evolves along with it.

 

Q - In your experience, which job in radiology is heavily underrated, underappreciated? And why?

 

Peter Morel: Over to you, Sergey.

 

Sergey Morozov: Thanks, Peter! I think the game of soccer works as a good analogy here. The person with the ball is the one dictating the flow of play. In a similar way, a nurse or radiology technician is the one taking care of the patient at the time of scanning. It’s these professionals that are both technically savvy while possessing strong emotional intelligence skills. That’s why it’s vitally important to acknowledge this and help develop the specialty of imaging technicians and radiographers to ensure patient safety and satisfaction is always kept at the heart of healthcare.

 

Q - To end on a positive note, what do you consider the biggest achievement of AI in radiology. And why?

 

Peter Morel: We are still in the early days of exploring the capabilities of AI in radiology. AI has and will continue to contribute to several domains including providing more quality time to the radiologist and extending the capabilities of the radiologist.

A big contribution AI is providing is the normalisation of parts of the process, where machines will provide the same result throughout the organisation. Being able to loop the feedback provided by the radiologist into the AI learning process, the knowledge will be shared with the internal community using the AI solution.

 

Sergey Morozov: AI is already here to help! With proper care and a professional teaching, it will evolve to become a must-have physician-assistant allowing hospitals and healthcare worldwide to grow and provide better services. We can make excellent healthcare available to everyone with the power of AI.

Newsletter
Get great insights from our expert team.
Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.
By signing up you agree to our Terms & Conditions