The recent explosion of generative AI has pushed public discourse to a palpable tipping point. Nearly overnight, it seems, we’ve cycled from “must develop and/or adopt new AI applications ASAP” to existential fear, avoidance, and even acrimony.
For those of us who have been working directly in the artificial intelligence space for years, specifically in healthcare, these are both trying and exciting times.
On the “trying” hand, many people, including some AI experts, are raising legitimate red flags about various aspects of some AI technologies. These concerns, while valid, have sparked a widespread backlash against the entire spectrum of AI applications — even those that have little in common with the systems in question. But on the “exciting” hand, more people are also seeing the positive potential of AI. Wherever you might land, the most significant aspect of the recent conversation is just that — a discussion has begun. By and large, society is gaining a deeper understanding and growing curiosity about what AI can really do. And the massive gray area between these emerging, polarized perspectives is bursting with possibilities.
This is especially true in healthcare, where we can effect — and, in fact, are already effecting — real, positive change for health systems worldwide. We are on the precipice of massive breakthroughs on countless fronts, and halting progress now would simply and needlessly prolong problems already long overdue for solving.
This is why we must stay the course and push past the hype. Healthcare decision-makers must identify the solutions that address today’s most crucial needs — variability in care, budgets (and clinicians) stretched too thin, and slow, reactive, outdated quality improvement methodologies — while recognizing the difference between legitimate concern and unfounded panic.
Now is the time to determine which healthcare AI apps are in for the long haul. To focus on innovators who are taking the right steps for the right reasons — and lean in.
Why emerging healthcare AI is unique
AI has already made its way into many aspects of daily life, from GPS navigation to smart home technologies. It’s also already common in many areas of the hospital, such as radiology and pathology, which it has rapidly commoditized.
But the most transformative potential of AI is its ability to do things humans can’t do — like learn whale language or process massive amounts of data at lightning speed. It’s in this capacity that emerging healthcare AI technologies stand to provide the greatest value — and the greatest impact. Because the standard of proof for how we structure and implement healthcare is so necessarily high, progress has traditionally been incredibly slow. But AI promises to change all that.
Large Language Models (LLMs) are at the core of most of the latest AI debates. And while the current trajectory of these text generators signal their inevitable disruption of the software industry, emerging healthcare AI is in another league altogether. It’s capable of rapidly surfacing never-before-seen insights that can change not only individual lives but entire health systems in ways that are (and have been) desperately needed, such as:
- Reducing variability, inequality, and bias in care and outcomes
- Identifying, validating, and promoting adherence to data-driven best practices
- Improving the safety and quality of care for all patients, everywhere
- Providing visibility and transparency into previously inaccessible and/or “data-rich, insight-poor” (DRIP) areas of healthcare, such as surgery
Plus, as governing bodies are scrambling to regulate broad, public-use generative AI platforms, the healthcare space is already heavily regulated with regard to data, privacy, and patient rights. So the potential benefits of healthcare AI likely vastly outweigh the risks by comparison.
But how do you know which healthcare AI applications will actually deliver on all that promise? Which will emerge stronger and help your hospital system thrive when the dust from this latest hype cycle settles? I believe it comes down to three key things.
Focus, focus, focus
Perhaps the biggest difference between generative AI platforms and emerging healthcare AI technologies is the level of specificity required to derive value from them.
Generative AI is based on huge, varied datasets and used for extremely broad, wide-ranging tasks. For example, people use LLMs for everything from meal planning to cover letter writing to cheating on English homework.
But healthcare AI must maintain multiple layers of specificity to be of any use whatsoever. In fact, I’d go so far as to say that failure to embrace any of the following three essential focus areas spells certain death for healthcare AI hopefuls:
1. Vertical focus
Yes, healthcare itself is a vertical. But even within that vertical, the thinner we slice, the more impactful the resulting AI output will be. In surgery, for example, we’re drilling down deeper, training video-based AI to recognize individual steps and even different techniques used in specific procedures. By connecting that data to patient outcomes, we can analyze how those tools or techniques impact everything from safety and efficiency to patient recovery and other outcomes.
By narrowing the focus of the technology, we can better measure and control the results — and fine-tune value to individual needs.
2. Focus on differentiation & specialization
The most successful healthcare AI software provides specific, measurable value for specific stakeholders. Privacy, security, and improved care for patients. ROI for enterprise health system leaders. And unique functionality that meets real needs and fits into existing workflows — no “AI for AI’s sake.”
Take, for example, automating the identification of near-miss events in surgery and bringing those events to the attention of surgical teams. This empowers clinicians to adjust their techniques and proactively tackle quality improvement.
3. Focused, proprietary datasets
Generative AI output is notoriously vague and generic. To get relevant, valuable results, you must have exceptional control over the data flowing into the system — how it’s collected, stored, processed, analyzed, etc. — as well as out of it, meaning how it’s ultimately delivered in the form of insights. The data must be as tailored and “clean” as possible to ensure results are accurate, unbiased, and reliable.
Where do we go from here?
The possibilities of emerging healthcare AI are endless. I’ve seen firsthand the power of our rapidly growing surgical datasets, and I couldn’t be more excited about the future.
And regardless of public or personal opinions, AI is here to stay. What matters now is what we do with it. The technologies that emerge stronger from this inflection point will be those laser-focused on doing the right thing — delivering value for patients, providers, and health systems worldwide, every day.
About Dr. Tamir WolfDr. Tamir Wolf is CEO and co-founder of Theator, a health-tech company that leverages AI and computer vision technology to help top surgeons and leading hospitals around the globe improve the quality of patient care. With a mindset of where you live shouldn’t impact if you live, Tamir has grown Theator into an industry powerhouse, delivering meaningful, unbiased and actionable insights in the OR to help surgeons once and for all drive better outcomes and solve for disparity and variability in surgical care.