Surgery might seem like the last place for AI—it’s hands-on, physically demanding, and requires split-second human judgment. But AI is finding applications across the surgical journey, from planning through to postoperative care.

I’ve been exploring this area with surgical colleagues, and while it’s not my primary expertise, the developments are worth understanding for anyone involved in healthcare AI strategy.

Preoperative Applications

Surgical planning. AI analysis of imaging to support surgical planning is becoming common in some specialties. Orthopaedic surgery, for example, uses AI to analyse X-rays and CT scans for joint replacement planning—identifying optimal implant sizing and positioning.

Risk prediction. AI models that predict surgical complications based on patient factors, procedure type, and other variables can inform consent discussions and identify patients who might benefit from preoperative optimisation.

Case scheduling. AI optimisation of theatre scheduling to improve throughput, reduce cancellations, and match case complexity to available resources.

Intraoperative Applications

This is where AI gets interesting—and challenging.

Surgical navigation. AI-enhanced surgical navigation systems that provide real-time guidance during procedures. This is most developed in neurosurgery and orthopaedics, where precision matters most.

Computer vision. AI analysis of video feeds from surgical cameras. Applications include:

• Identifying anatomical structures
• Detecting surgical phase (where in the procedure are we?)
• Flagging potential complications
• Providing trainees with augmented feedback

Robotic surgery. AI enhancement of surgical robots. The Da Vinci system and competitors are incorporating more AI for motion smoothing, instrument tracking, and (eventually) autonomous sub-tasks.

Real-time decision support. AI that analyses physiological monitoring data during surgery to provide alerts and recommendations. This is early-stage but developing.

Postoperative Applications

Complication prediction. AI analysis of postoperative data to identify patients at risk of complications—sepsis, bleeding, anastomotic leak—before clinical deterioration.

Recovery monitoring. AI-powered monitoring of recovery milestones, flagging patients who are off-track.

Outcome prediction. AI models predicting functional outcomes, length of stay, and rehabilitation needs.

What’s Actually Deployed vs. What’s Emerging

To be clear about maturity:

Deployed and working:

• Preoperative planning tools in orthopaedics and some other specialties
• Basic surgical navigation
• Postoperative risk prediction (similar to broader clinical prediction tools)
• Theatre scheduling optimisation

Emerging but not mainstream:

• Real-time surgical video analysis
• AI-enhanced surgical robotics with meaningful autonomy
• Integrated intraoperative decision support

Research but not clinical:

• Fully autonomous surgical procedures (and probably shouldn’t be)
• General-purpose surgical AI that works across procedures

Surgical Specialty Variations

AI maturity varies significantly by surgical specialty:

Orthopaedics. Most advanced in AI adoption, particularly for joint replacement planning and navigation. Clear anatomical targets and standardised procedures make AI application more straightforward.

Ophthalmology. AI-assisted cataract surgery and retinal procedures are developing. The precision requirements of eye surgery make AI assistance potentially valuable.

Neurosurgery. Navigation and planning AI for brain tumour resection and other procedures. High stakes make accuracy critical.

General surgery. Lagging other specialties. More variable procedures and anatomy make AI application harder.

Cardiac surgery. Some applications in planning and risk prediction. Intraoperative AI is limited.

Implementation Considerations for Health Services

If your organisation is considering surgical AI:

Start with planning applications. Preoperative planning AI has clearer regulatory pathways, lower risk, and proven value in some specialties. Begin there rather than intraoperative applications.

Engage surgeons as partners. Surgical culture values autonomy. AI that feels imposed will face resistance. Surgeon champions are essential.

Understand theatre workflow. Operating theatres have complex workflows, strict sterility requirements, and time pressures. AI that disrupts these won’t succeed regardless of technical merit.

Consider training implications. If AI assists surgeons, how do trainees develop independent skills? This is a genuine concern that surgical education is grappling with.

Regulatory clarity matters. Intraoperative AI that influences surgical decisions is higher-risk from a TGA perspective. Understand the regulatory classification before investing.

The Surgeon-AI Relationship

I find the discussion about surgeons and AI particularly interesting. Surgeons have traditionally been seen as highly autonomous practitioners—skilled individuals making independent decisions.

AI challenges this in several ways:

Decision transparency. AI recommendations create records of what was suggested. If a surgeon deviates, there’s documentation. This changes the medicolegal landscape.

Skill development. If AI assists with aspects of surgery, does the surgeon develop the same manual skills? What happens when AI isn’t available?

Professional identity. For some surgeons, the idea of AI assistance feels like a challenge to their expertise and identity.

These aren’t reasons to avoid surgical AI, but they need acknowledgment and management.

Ethical Considerations

Surgical AI raises specific ethical questions:

Informed consent. If AI assists in surgery, should patients know? What if they prefer human-only surgery?

Liability. When AI contributes to a surgical complication, how is responsibility allocated between surgeon, institution, and AI vendor?

Equity. If AI-enhanced surgery improves outcomes, but only some institutions can afford it, does this create new inequities?

Data use. Surgical video is rich data. Using it for AI training raises privacy questions that extend beyond standard clinical data.

My Assessment

Surgical AI is real but not transformative yet. The most practical applications today are in preoperative planning, where AI can genuinely improve surgical precision and outcomes in specific procedures.

Intraoperative AI is developing but faces technical and adoption challenges. Real-time surgical assistance requires extreme reliability—if AI fails during surgery, consequences could be severe.

For healthcare organisations, I’d recommend monitoring this space and implementing proven planning applications where relevant, but maintaining realistic expectations about transformative intraoperative AI in the near term.

The operating theatre will remain a fundamentally human environment for the foreseeable future. AI will assist; it won’t replace the surgeon’s hands, judgment, and responsibility.

---

Dr. Rebecca Liu is a health informatics specialist and former Chief Clinical Information Officer. She advises healthcare organisations on clinical AI strategy and implementation.