The Truth About AI In Healthcare

In heavily regulated industries such as healthcare, digital innovation can be slow to progress. However, once organizations push towards digital transformation and innovation, the benefits that can be achieved such as revenue growth, patient volume, and cost of care can provide tremendous value. Healthcare organizations are looking for an approach to cost-effective and technically efficient build-out to help on their digital transformation journeys. With investments shifting from core EMRs to infrastructure solutions that enable flexibility and adaptability, healthcare organizations are looking to digital innovation to solve these key issues. In an upcoming Enterprise Data &AI presentation on May 5, 2022, Vignesh Shetty, SVP & GM Edison AI And Platform, GE Healthcare Digital will discuss GE Healthcare’s digital health platform and how it’s helping companies in the healthcare sector on their AI and data journey.


Vignesh Shetty, SVP & GM Edison AI and Platform, GE Healthcare Digital


In this interview for Forbes, Vignesh shares how GE Healthcare is applying AI and ML, some of the challenges associated in adopting transformative technology in heathcare, as well as some of the things to consider when navigating privacy, trust, and security around data related use cases and needs.

How is GE Healthcare applying AI/ML in different application areas?

Vignesh Shetty: GE Healthcare uses AI to help healthcare providers achieve clinical and operational outcomes that create impacts for patients, providers, and health systems. For AI to be most effective, it should be seamless, invisible and within existing workflows while uncovering patterns (e.g., uncovering unknown unknowns) that are missed by humans.


Three areas where we see opportunities to apply AI are:

Platform as an AI engine: Healthcare systems experience fragmentation due to disjointed data sources, separate systems with incompatible vendors and other collection and collation issues. This “digital friction” makes it difficult for healthcare systems to adopt the applications and technology needed to access and manage enormous amounts of disparate clinical, diagnostic, and operational data.

We are developing Edison Digital Health Platform to accelerate app development and integration by connecting devices and other data sources into an aggregated clinical data layer. The goal of the platform is to enable hospitals and healthcare systems to effectively deploy the clinical, workflow, analytics and AI tools that support the improvement of care delivery, the promotion of high-efficiency operations, and supporting reduction in the IT burden that typically comes with installing and integrating apps across the enterprise.

On–device AI:

From big iron MRI scanners used by doctors to detect tumors on the prostate gland to mobile X-ray units in the ER or ICU that technicians use to image the lungs of COVID patients at their bedside, we are seeing a tangible impact with our AI embedded on the device.


Examples include:

Critical Care Suite which automatically analyzes X-Ray images for critical findings (such as pneumothorax) producing triage notifications. It also enables automated measurements and quality control that can help improve efficiency on the front lines.

Air Recon DL is our advanced deep learning Image Reconstruction Technology that works across anatomies – this technology can offer clinicians a significant reduction in exam times, which helps with the patient experience and address today’s backlog more quickly and with impressive image quality.

TrueFidelity™ CT uses deep-learning image reconstruction to generate razor-sharp with deep detail, true texture, and high fidelity for every CT scan.

Predictive insights at the department and enterprise level applications:

Early adopters have reported seeing significant reduction in no-show rates using the Smart Scheduling application which means more slots filled, greater efficiency for providers and payers, and a better experience for patient.

How do you identify which problem area(s) to start with for your data analytics and cognitive technology projects?

Vignesh Shetty: If you don't see AI’s incredible potential to help healthcare providers improve diagnostic confidence, efficiency, and productivity, look closer. Likewise, if you don't find some of the hype absurd, look even closer.

GEHC invests a lot of time to avoid potential pitfalls by:


  • Continuing to deeply understand the needs of clinicians and hospital systems
  • Spending tremendous energy developing that intuition
  • Studying and understanding nuances and workflows to complement the market research


We work closely to collaborate on data and expertise between the two worlds of practitioners and our developers. Both are passionately striving to solve the same problems but not necessarily talking to each other, early enough. The result is that some offerings do not address the right clinical or operational need, are not suitably integrated into existing workflow, or simply do not work.

As a global leading med tech and digital provider, we are committed to helping healthcare providers reduce pain points, improve diagnostic confidence, and focus on reducing digital friction.

What are some of the unique opportunities you have when it comes to data and AI?

Vignesh Shetty: Folks call data the 21st century oil – a better analogy would be crude oil. If harnessed well there is massive potential especially by focusing on these three areas:


  1. Creating a comprehensive 360-degree patient view (leveraging genomic, radiomic, imaging and other data)
  2. Deployment (ongoing validation of algorithms as it adapts to real world data) and regulation
  3. Building trustworthy, ethical, and explainable AI systems


AI, like other tools, is a new lever. Leverage by definitions amplifies an input to provide greater output. We are using data to understand the leverage points in a clinician’s workflow which helps identify where to apply various tools (AI being one of several) to yield nonlinear results.

Can you share some of the challenges when it comes to AI and ML adoption, especially for heavily regulated industries such as healthcare?

Vignesh Shetty: The head of radiology at a hospital in Europe, and one of our key customers, used this description as it relates to AI when he said, “The menu is spectacular, the spread is broad, the chefs are Michelin starred, the aroma is great, when do I get to eat?”

His sense of unfulfilled potential stems from the following learnings:


  • Massive friction with respect to implementation into existing workflows across disconnected IT systems
  • Hospital IT departments don’t have the bandwidth or the expertise to manage the implementation, integration, and maintenance of individual applications
  • Interoperability constraints
  • A hospital shouldn’t be a collection of disconnected IT systems that all speak a different language and break during upgrades of one or more components since there isn’t a standard


In heavily regulated industries like healthcare, clinicians rely on heuristics and habit formation by constructing workflows that are unique to them to minimize mistakes.

For many physicians, the main hurdle to AI adoption is familiarity and experience with the technology while minimizing risk to the patient and distraction to ensure the AI is going to help rather than hinder their clinical routine. It's a quandary that’s being resolved with thoughtful, targeted AI based on longitudinal patient data that builds trust and is quietly working behind the scenes so as not to disrupt or create another step in an already strained environment. Trust leads to utilization, which is a key to unleash AI's true potential.

How do you deal with varying levels of data quality for AI and ML systems?

Vignesh Shetty:


  1. We increasingly leverage synthetic data where appropriate for training and real-world data for validation.
  2. Modern data science owes a lot of its success to harvesting “data exhaust”: data of seemingly no use to an organization that would normally get discarded in an environment of high storage costs, but we believe has huge value in driving clinical/operational outcomes.
  3. We then use this to kickstart low-stakes experimentation, lowering the cost of failure.
  4. The following trends act as “data fuel” for the “AI fire” – data variety from wearables, sensors, and broad EMR adoption, proliferation of the internet, cheaper hardware, cloud computing and better algorithms.


How are you navigating privacy, trust, and security concerns around the use of your data?

Vignesh Shetty: When it comes to deployment, an important hurdle is how to ensure safety and efficacy over time as algorithms adapt and evolve, through the continual evaluation of performance and assessing the need for reapprovals of specific AI solutions.

Healthcare providers and AI companies like ours are coming together to put in place robust data governance, ensuring interoperability and standards for data formats, enhance data security and bring clarity to consent over data sharing. Collaborating on cybersecurity expertise is key because it will largely influence the trajectory of AI adoption. The necessity of HIPAA and HI Trust* compliance as well as evolving privacy regulations make the standard for service very high.

AI research needs to heavily emphasize explainable, causal, and ethical AI, which could be a key driver of adoption.

What are you doing to develop a data literate and AI ready workforce?

Vignesh Shetty: At GE Healthcare, we are focused on thoughtful integration of ML and AI throughout the fabric of the organization using a three-tiered approach


  1. Acquiring basic knowledge about how AI works in a medical setting to understand how such solutions might help them in their everyday job and what the limits are.
  2. Create the conditions for innovation ecosystems to flourish. Teams need to learn to start with new assumptions consistently and repeatedly.
  3. Continue to invest in training, engagement, and coaching resources for the end-users (rad techs, nurses) in the development of solutions (5% tech, 95% change mgmt.). Our philosophy is to treat every new idea as a challenge to your imagination, not a threat, so rather than listing the reasons why an idea won't work, try to think, and then find the ways in which it could.


We are optimistic about the future of AI, but we can’t leave it to chance. I’m convinced that the skills for responsible leadership in the AI era can be taught and that people can build safe and effective systems wisely.

What AI technologies are you most looking forward to in the coming years?

Vignesh Shetty: AI is central to building a future where healthcare is personalized, prevention-oriented, and affordable and we can make a difference to patients and providers in the moments that matter by offering both prescriptive and predictive AI driven insights to help healthcare providers improve both clinical & operational workflows.

It’s possible to envision a significant improvement in the patient/provider experience using multi-modal data that create a longitudinal patient record which helps healthcare providers to schedule a patient at the right time which would reduce no-shows, ensure that patients are scheduled on the right device and facility with the relevant logistics in place. Imaging a patient receiving proactive care (thanks to wearables and sensors interacting with AI models) and enjoying frictionless experiences (with robotic assistants for routine tasks), all while going about her daily life.

This will not occur by applying new technologies through the lens of old applications or existing ways of doing things. Building a better mousetrap is a great way to onramp users into the digital realm. But it also has limitations; you can only see what’s new in terms of what has always been.

The way forward will be native applications that are built with these new paradigms in mind. In retrospect, native applications can seem obvious, but in their early stages they can be difficult to imagine. The goal is to enable caregivers to get better, which means spending more time managing their patients rather than managing the patient record.

Lastly, bet right and early, when everyone (or most) others bet wrong, and try to build something people will look for, will talk about or would miss if it were gone.

In an upcoming Enterprise Data &AI presentation on May 5, 2022, Vignesh will dig deeper into some of the topics discussed above as well as share how GE Healthcare’s digital health platform is helping companies in the healthcare sector on their AI and data journey.

Previous post
Back to list
Next post