Artificial Intelligence continues to be one of the most talked-about topics in healthcare. But while the conversation often focuses on futuristic visions, few experts bring together medical knowledge and deep technical expertise to realistically evaluate where AI can deliver value today — and where it still falls short.
In a recent episode of Verizon’s Healthcare on Air podcast, Dr. Nemanja Kovacev, Head of the Healthcare & Life Sciences Center of Excellence at HTEC, shared his perspective on AI in medicine — where it’s evolving, how it can support clinicians, and what we should be cautious about.
From software to surgery: Nemanja’s unique career path
Nemanja’s story is a rare one. His passion for programming started at a young age and continues throughout his life. Originally trained as a trauma and orthopedic surgeon, over time, he found himself uniquely positioned at the intersection of medicine and technology. As AI technology development picked up, he got interested in this new technology.
He pursued a specialization in AI in Healthcare at Stanford and has since been activelly involved in healthcare and life sciences projects that leverage cutting edge medical technology. His mission is to build tools that genuinely improve care.
Today, he leads HTEC’s Healthcare & Life Sciences Center of Excellence, where his team provides guidance both to HTEC’s engineering teams working on MedTech and health projects and to clients across the healthcare ecosystem. His dual perspective allows him to translate clinical reality into meaningful AI applications — and vice versa.
A realistic view on AI and the promise of Agentic AI
Nemanja sees AI not as one monolithic technology, but as a diverse toolbox of algorithms with applications across diagnostics, data analysis, and operational efficiency.
He also discussed the growing field of Agentic AI, which refers to AI systems capable of reasoning, planning, and acting more autonomously. He described its potential in healthcare as promising — but only in certain domains. For instance, a hospital system might use an agentic AI to proactively identify and suggest diagnostic tests for patients based on medical records, without explicitly being told to do so.
“AI is a software entity able to learn without being explicitly programmed and Agentic AI goes a step further in terms of autonomy.”, Nemanja suggests.
Still, even with these advancements, he was quick to caution that medicine requires far higher standards of reliability than most industries.
The biggest benefits of AI in healthcare: clarity and order
While much of the AI conversation is clouded by inflated promises, Nemanja sees two areas where AI is genuinely moving the needle in medicine at the moment: image recognition and data structuring.
As a surgeon, he knows firsthand how essential speed and precision are in diagnostics — and AI is proving to be an invaluable second set of eyes.
“Image recognition has become one of the most powerful tools in diagnostics,” he explained, particularly when it comes to complex imaging like MRI and CT scans. These are high-stakes scenarios where a small detail can make a big difference, and AI can help catch patterns the human eye might miss.
But the bigger revolution may be happening behind the scenes.
Healthcare is infamous for its “data swamps” — massive amounts of unstructured, disconnected, and often inaccessible information. From patient records to device readouts, the system is drowning in data that’s too vast for any human to process meaningfully in due time.
“AI is incredibly effective at structuring this chaos — spotting correlations, and even pointing toward potential causations, hidden in the noise,” Nemanja shared.
These patterns, invisible to the human mind, become actionable insights with the help of AI. It’s not about disruption — it’s about enhancement. “There’s no disruption in medicine, only evolution through small steps,” he emphasized. “It’s a high-risk area, and no one is going to go fast and break things.”
The real concerns with AI in medicine today
Nemanja didn’t shy away from pointing out the challenges of integrating AI into healthcare workflows. Among his top concerns:
Data privacy
Medical data is incredibly valuable — and increasingly vulnerable. It is known to be one of the most expensive black-market data.
“As we increase the amount of structured data through AI, cyberattacks will become more prevalent because the loot is very valuable.”
Accountability
Nemanja expressed concern about the push from some experts to incorporate AI into doctors’ decision-making process. If doctors are required to use AI as the primary decision-maker, then who holds responsibility for the outcome?
“As long as AI is in a consulting capacity, it’s OK. But for a doctor to be obliged to use it as the main expert is a big problem.”
Bias and ethical use
Training data is currently a major concern in medical AI. The models can only be as good as the data they’re fed. Poorly curated or biased training data can lead to real harm in patient outcomes.
Regulatory balance
Too much regulation could stifle innovation — too little invites chaos. He referenced the EU AI Act as an example of how well-meaning policies might unintentionally hamper progress and require revisions.
AI should be used as a consultant
Nemanja’s position is clear: AI should act as a consultant, not a replacement for a doctor.
While large language models are making headlines, Nemanja points out that they still hallucinate and lack consistent accuracy. A recent study by Stanford University he cited found LLMs have an output accuracy of just 70%, which is simply not acceptable for clinical use:
“In medicine, it’s just not enough. It needs to be close to 100%.”
He also pointed out that AI notetaking doesn’t free doctors entirely as it is supposed to. For now, many doctors spend more time checking AI-generated notes than if they’d written them themselves. Notetaking also helps doctors collect their thoughts and remember the specifics of a patient’s condition. AI is useful in documenting textbook injuries and illnesses, but nothing specific and more complex should be automated at this moment.
Where AI will drive the biggest impact in the near future
Nemanja believes AI will enter healthcare gradually, improving specific areas, where different AI agents can be inserted gradually, long before any “total takeover” scenario. Areas of immediate promise include:
- Improvement in imaging methods: AI is already enhancing the accuracy and speed of all standard imaging methods.
- Operation room AI utilization: In operating rooms, AI will be used for tracking non-critical parameters like temperature and humidity to improve patient experience and outcomes.
- Remote monitoring and structured data: AI can extract meaning from unstructured clinical records, revealing new correlations and improving continuity of care.
- Back-office tasks: From logging patient data to maintaining hospital records, AI can reduce the administrative load on clinicians.
- Cybersecurity: AI systems will soon play a key role in detecting and preventing exploits, proactively protecting sensitive medical information.
- Robotic AI advancement in limited use cases: In the near future, robotic AI will likely be applied in low-risk, supportive roles.
Together, these advancements signal a future where AI quietly transforms healthcare from the inside out—not by replacing clinicians, but by reinforcing every layer of care with speed, structure, and security.
Meet HTEC’s team at the MedTech Forum 2025 in Lisbon
As this conversation with Dr. Nemanja Kovacev makes clear, creating purposeful AI in healthcare requires a clear-eyed view — one that cuts through the hype and focuses on solutions that deliver clinical and practical value.
From May 13–15, HTEC’s Healthcare & Life Sciences team of experts will be at the MedTech Forum in Lisbon, presenting AI-powered solutions developed in collaboration with partners — including Humeds, Aloe Care, and Marani— and meeting with the MedTech companies looking for strong partnerships in the areas of MedTech AI innovation.
Meet our team to explore:
- What are the unique challenges related to your medical AI development
- Architect AI solutions that clinicians will trust and adopt
- Bridge the gap between technical innovation and clinical reality
If you’re looking to implement AI and data science in MedTech or healthcare solutions, HTEC’s versatile team can help you build smart safe and regulatory-ready systems with the focus of improving patient experiences and outcomes.
