How to succeed with AI-powered healthcare solutions

Over the past 15 years, AI has become firmly embedded in healthtech, particularly in radiology. “Decision-support tools in radiology primarily rely on computer vision,” Jan Beger explains. “These tools often use so-called narrow AI models that have been trained to detect one specific condition and alert the radiologist when needed. But we’re now entering a new era, with technologies capable of much more holistic, far-reaching impact.”

At the forefront of this shift are foundation models – AI neural networks trained on vast volumes of raw data, enabling them to be applied across a wide array of use cases.

“Radiologists don’t look at images in isolation,” says Mr Beger. “They interpret them in context, drawing on lab results, patient history and other clinical data. Foundation models can do something similar – ingesting diverse data types and reasoning across all of them. It brings us closer to AI that can operate more like a clinician. That’s truly exciting.”

Making sense of multimodal data

Healthcare systems must handle a vast array of data types ranging from structured lab results and diagnostic images to free-text clinical notes and even handwritten observations. “This diversity presents a challenge, but also a major opportunity. With the latest generation of AI technologies, we can leverage and derive meaning from multimodal patient data. It’s a game-changer.”

We’re now entering a new era, with technologies capable of much more holistic, far-reaching impact.
Jan Beger, GE Healthcare

Mr Beger also highlights that artificial intelligence has a role to play in bridging the communication gap between clinicians and patients. “Medical reports can be overwhelming. AI can help translate jargon into clear, patient-friendly language, even responding in an empathetic way.”

A wave of opportunities in healthtech

These advances are creating a wave of opportunity, not just in terms of new AI-driven health products, but also for enhancing internal business operations. “Artificial intelligence will influence many areas of our work, and over time, it will support employees in various functions throughout the company. Every company should begin this journey, and if you haven’t already, start tomorrow. There’s no time to waste.”

Of course, the path is not without its challenges. “Using cutting-edge AI is relatively expensive and requires investment,” Mr Beger cautions. “And we are operating in a highly regulated industry, where getting approval for AI-enabled medical devices is no small feat.”

Every company should begin this journey, and if you haven’t already, start tomorrow. There’s no time to waste.
Jan Beger, GE Healthcare

Keeping pace is another concern, with major AI breakthroughs happening weekly. “Today’s latest research might be outdated in six months. This makes choosing the right technology tricky. Smaller companies may want to consider partnerships rather than trying to build their own models from scratch.”

Putting the human back into healthcare

While AI opens up exciting business opportunities and creates economic value, Mr Beger sees its most significant promise in restoring the human element of care. “Let AI handle the routine, so that healthcare professionals can focus on delivering the extraordinary,” he says. “We need to bring the human touch back to healthcare.”

Successful AI in healthcare: Three golden rules according to Jan Beger

Ensure your solution addresses a genuine clinical need.

Over the past 20 years, I have seen many startups building amazing technology – without any market fit. Start by identifying a genuine clinical challenge. And to do that, you need to be working closely with hospitals, clinicians and patients.

New AI solutions must be deeply embedded into existing workflows.

Healthcare professionals are already overwhelmed with admin. Introducing yet another standalone application or system with a different user interface requiring a separate login will not work. Your solution needs to integrate seamlessly into the software they already use – ideally with no extra clicks.

Focus on trust and adoption.

Many people are still rather suspicious of AI. Healthcare staff worry about losing their jobs, and patients fear a cold, impersonal care. We must build trust by demonstrating value and ensuring that users have the right level of AI literacy to use those technologies responsibly in day-to-day healthcare.

Photo credits: GE Healthcare