At the Heart of AI in Healthcare

When we listen carefully, what do our customers want to tell us about how they feel about AI in Healthcare. And, how do they want us to drive this change, and not just be a witness to it.

As I delve into some observations and reflections, wanted to share what a #customercentric #strategicoutlook could look like. Please join the conversation as I try to unravel the layers of what lies at the heart of AI in Healthcare.

We had a great European Congress of Radiology Conference #ECR2024 at Vienna last week. And, like with many industry conferences nowadays, the invariable overarching theme that came about was - #Artificial Intelligence, and its 'potential' impact on the #FutureofHealthcare.

Am deeply thankful for the varied and insightful discussions with our customers, partners, and industry colleagues, on what the future holds for all of us, and what would be our part in shaping this future.

Among many interesting ideas, highlighting the most prominent ones; some spicy, some a little provocative, but compelling, nonetheless.

Discovery mode - ON

It’s evident that AI in Healthcare is in a space where we as innovators and engineers are discovering this space and stretching the canvas of our imagination. We are learning as much about AI, as we are making AI learn about our world.

As I see it, AI in Healthcare is its studentship phase, fast growing up to be employable and start delivering results. We are training AI to solve real world problems, while getting trained ourselves to train AI with relevant data.

We both, AI and us, are on the path to this Discovery, together.

?AI in Healthcare is its studentship phase, fast growing up to be employable and start delivering results.

Observation - Data is the new oil. Are we treating it like Oil?

Reflection - Yes, data is indeed the new oil. #BigData surely is. But, in the case of our industry, being at times equally guilty, we are doing #DataMining for this oil very differently.

Imagine, a Saudi Aramco – the top oil producer, would have said that the oil it extracted would only be processed by a Reliance - the biggest oil refinery; to be used by only a Volkswagen car – the largest automobile manufacturer. And, they would have technically restricted the usage in any other vehicle with some chemical markers. What would have been the ‘value’ of that oil! The currency of such oil would have been drastically restricted. The true value of oil is unlocked in its enabling transportation, running industry, and at the core of it – being universal.

Likewise with #BigData, where we seem to be creating #DataLakes that are not interconnected. Narrow #supervisedlearning seems to be the norm, which is creating further silos of fragmented AI solutions.

Collaborate on Learning and Compete on Value.

Strategic outlook –?Collaborate on Learning and Compete on Value - All of us are in the business of healthcare. We need to build profitable and sustainable businesses to continue to innovate and serve our customers and patients around the world. But, we all are in the business of healthcare because ‘the patient’ is in the center of center of our universe. We serve our customers – hospitals and clinicians, who serve the patient, to keep the global population healthy.

AI is still in a learning mode, and industry wide collaboration through #SmartAPIs, to fuel this learning through interconnected data lakes, is critical for everyone’s success and above all, for a healthier planet.

#Standardization of this learning would generate the true value for our customers, and then, there is enough business to be unlocked by creating value with products and solutions, that emerge from this standardized learning.

Regulatory framework is a definite #guardrail, in parallel to standardization and often gets into a chicken and egg situation. We all know that #continuouslearningsystems are still not getting approvals, probably rightfully so.

Someone said, I could 'love' a product, but I’d rather 'use' a product that I trust. Today’s AI solutions are yet to gain trust of our customers as mainstream solutions. Collaborating for standardizations and regulations could help move the needle of trust, into a direction where we would be able to serve our customers faster and at scale.

Observation - AI is a tool, it’s an instrument, and a very intelligent tool at that.

Reflection - Like any other tool, this one too, is going to be as good as we make of it. The overuse of marketing jargon is taking away from the fact that AI is an instrument that has the most significant potential to influence the rate of change, for this generation and next.

What if, our good old carpenter claimed – Made with best Hammer! No, he didn’t. He was always focused on delivering the best chair and never claimed that the chair was great because of the best hammer!

We, on the other hand, have been talking so much about AI and so little about the #value and #outcomes we create for our customers - #clinical, #financial, and #operational. A humble submission, especially to friends and partners in the #startup world – Marketing compulsions aside, as long as our customers understand our WHAT, they’d rarely be interested in the HOW. What is the Value. How is the Tool.

Focus on the Problem Space is vital.

Strategic Outlook –?Focus on the Problem Space is vital - The most important challenges continue to be around #staffshortages, #access to #qualitycare and #affordablehealthcare. We as innovators in AI in Healthcare need a deep and detailed understanding of the clinical and operational challenges that our customers are trying to solve every day. Just because there has been an advent of AI in the solution space, doesn’t mean the problem space has changed.

From a business perspective, executives promising boards and startups promising investors, that AI would be a silver bullet for quick profits, would do a great deal of disservice to the potential impact of AI. Again, for friends and colleagues in the start-up world with institutional funding, the value of AI would be created, not in the usage of AI, but in the solving of problems, with or without AI or by employing other tools as necessary.

#Neuralnetworks and #cognitivecomputing have been around for decades and have served a great purpose with #imagerecognition and #imagereconstruction. What our customers are now looking for are systems that utilize #deeplearning, not merely on #relational datasets but also #contextual data, to assist them with #predictiveanalytics and #prescriptiveinsights.

?

Observation – Our customers would love to spell AI as Intelligent Assistant and not as Instant Automation.

Reflection - Automation is one of the use cases for AI, but a very narrow one, and surely not where rules-based databases have been working for ages. Sorting data, prioritizing results, pixel-based image analysis have all been solved years back. AI algorithms need to bring in the learning component to gain relevance.

Moreover, in most cases the automation being aimed for, is like having a secretary who manages only one day of the calendar. Now think of a situation where one needs to manage five different secretaries for managing a calendar for the workweek! And, each of them refusing to look at the other day. For sure, we won’t feel comfortable with such automation, indeed find it frustrating!

A Trusted Generic Intelligent Assistant

Strategic outlook – a trusted generic intelligent assistant - For AI in Healthcare, we need a different kind of #GenAI, where Gen stands for Generic - a #GenericExpertSystem of sorts.

In umpteen discussions with our customers, they seem to be overwhelmed with piecemeal solutions that lack integration, lack cross-communication, lack standardization, and ultimately lack being trusted by our customers.

The answer lies in how we approach training AI algorithms for the future. Continue with ultra specific use cases or prepare AI to replicate how #learningcurriculums have evolved and how our customers have been working for ages.

A combination of #supervised and #unsupervisedlearning. An assistant that elevates the silos of radiology, pathology, lab etc. to the Office of the #ChiefDiagnostician, who is ready to deliver #PrecisionMedicine in its true sense and scale. Building?a trusted generic intelligent assistant?for our customers, is what they want us to do. Something that they truly value, they use as mainstream solution and ultimately, they trust.

?

To summarize, cornerstones of this discovery that inspire us for building the future of AI in Healthcare, without forgetting the oil, the hammer, and the secretary ?? :

• Collaborating on Learning and Competing on Value.
• Focus on the evolving problem space.
• Building a trusted generic intelligent assistant.

?

Special thanks to friends who joined the conversations at #Healthcare5.0 #Roundtable on the sidelines of #ECR2024, let’s continue to fuel our discussion with the ‘oil’ that we discovered during our meeting, it's not 1859 ?? - Dr Amine Korchi from 3R - Réseau Radiologique Romand , Dr. Anjum M. Ahmed (MBBS, MBA, MIS) from 爱克发 , Ankur Sharma from 拜耳 , Antoine Disset, PhD from Median Technologies , Brandon B. Suh from Lunit Cancer Screening , David Hilderbrand from Blackford , Dr. Erik R. Ranschaert, MD, PhD from St. Nikolaus-Hospital Eupen , Jan Beger from GE医疗 , Julie Sufana from contextflow , Maria Proud , MHA from TeraRecon, Inc. , Michael Coulter from 上海联影医疗科技股份有限公司 , Murat Akturk from GE医疗 , Robert Wagner from 飞利浦 , Sinan Batman from TeraRecon, Inc. , Thomas Juhl Olesen from CEREBRIU , Vittorio Sportelli, PharmD RPh MNutr & Yael Misrahi from 爱思唯尔 , and of course from MarketsandMarkets? - Siddharth Shah , Siddharth Saha and Farhan Hussain with myself from 西门子医疗

Dr. John Sheehan Prof. Dr. Johannes Haubold Dr. Peter Mildenberger Marcel Wassink Herman Oosterwijk Ben Panter Dr.Arunkumar Govindarajan Dr Thanga Prabhu Dr. Gunnar Trommer Dr. Rajendra Pratap Gupta, PhD Ashkan Afkhami Andre Heeg, MD, PhD Stefan Larsson Brian Anderson, MD Bakul Patel Rajeev Raghuvanshi Alison Wesley