A 26‑year global review shows rapid growth, expanding uses in the operating room, and a call for safer, real‑world validation.  

By the numbers. Researchers mapped 821 scientific papers on artificial intelligence (AI) in surgery published between 1998 and 2024. Publications are rising fast—up from 1 paper in 1998 to 328 in 2023, a 14.5% average annual growth rate. More than 10,000 authors contributed, with Harvard University listed as the most prolific institution. The United States and China lead national output. Figure 2 on page 5 highlights Harvard’s position at the center of a global collaboration network, while the world map in Figure 3 (page 5) shows the U.S., China, Italy, the U.K., and Germany as key contributors.  

What’s hot. The field’s core topics reflect practical goals: classification, diagnosis, outcomes, risk, and management. Figure 5A (page 7) shows these high‑frequency keywords, and the thematic map in Figure 5D (page 7) points to rapidly developing areas such as neural networks, image segmentation, augmented reality, and intraoperative navigation. Together, these trends mark a shift from trial concepts to tools designed for day‑to‑day surgical care across specialties including orthopedics, oncology, and neurology.  

Where patients may notice AI. The pattern diagram on page 8 (Figure 6) lays out how AI fits into each step of care: before, during, and after surgery.

• Before surgery: AI can analyze scans and medical records to help tailor surgical plans and estimate risks.  
• During surgery: Computer‑vision tools can identify key anatomical structures and support precision moves in real time; early work with AI‑enabled robotics suggests fewer errors and more delicate operations.  
• After surgery: Wearables and sensors can funnel data to AI systems that flag potential complications sooner and support personalized recovery.  

Who’s publishing—and where. Figure 1 (page 4) shows Surgical Endoscopy as the most active journal by volume, while Annals of Surgery ranks highly by citations. The most‑cited papers include landmark work on deep‑learning image classification and early analyses of AI’s promises and pitfalls in surgery (Table 4, page 6).  

Why this matters. For patients, the near‑term impact of surgical AI is about safety, consistency, and access: smoother planning, extra “second eyes” in the operating room, and smarter follow‑up that can catch problems earlier. Importantly, AI is an assistant—not a replacement—for surgeons, who remain responsible for decisions.  

Mind the gaps. The authors stress that many tools still need real‑world, multi‑center validation, better data quality and security, and clear guidelines before they’re ready for broad deployment. The review also notes a limitation: it draws only from the Web of Science database, so some relevant studies may be missing. The conclusion (pages 9–10)calls for stronger collaboration across disciplines to turn prototypes into trustworthy bedside tools.  

Bottom line for readers: Surgical AI is moving from pilot projects to practical helpers that can enhance precision and recovery. Ask your care team how (and whether) AI tools are used in your procedure—and how your privacy and safety are protected as this technology evolves.  

Source: Li H., Han Z., Wu H., et al. “Artificial intelligence in surgery: evolution, trends, and future directions,”International Journal of Surgery, 2025.  

Editor’s note: This article is for information only and is not a substitute for professional medical advice.