How AI Is Revolutionising Surgical Planning in 2026

The pre-operative planning workflow in orthopaedic surgery has undergone a quiet but seismic transformation. What once involved acetate overlays on a light box, then graduated to digital templating on a PACS workstation, has now entered a third era: AI-driven autonomous planning. In 2026, commercial platforms are delivering fully automated surgical plans from a standard CT scan in under five minutes, with accuracy that rivals or exceeds fellowship-trained surgeons.

For orthopaedic trainees and consultants alike, understanding these tools is no longer a "nice to have." It is a core part of modern practice.

The Speed Problem That AI Solves

Manual 3D surgical planning for a complex revision hip replacement can take 45 minutes to two hours. The surgeon must segment bones from soft tissue, reconstruct anatomy, template implant sizes, plan osteotomy levels, and simulate range of motion. In a busy public hospital with 15 arthroplasty cases per week, that workload is unsustainable.

AI-powered planning platforms like Formus Labs, Stryker's Blueprint, and Smith+Nephew's CORI system now automate the most time-intensive steps. Deep learning models trained on hundreds of thousands of annotated scans can auto-segment the femur and acetabulum from a standard pelvic CT in under 90 seconds. The surgeon's role shifts from manual contouring to reviewing, adjusting, and approving the AI-generated plan.

What the Latest Research Shows

A landmark multi-centre study published in The Bone & Joint Journal in early 2026 compared AI-generated total hip arthroplasty plans against those produced by fellowship-trained arthroplasty surgeons across 500 consecutive cases. The results were striking:

• Cup size accuracy: AI matched the surgeon's chosen implant within one size in 94% of cases (vs. 89% for manual digital templating)
• Stem size accuracy: 91% agreement within one size
• Combined leg length and offset prediction: AI plans achieved a mean discrepancy of 1.8mm, compared to 3.2mm for manual plans
• Time savings: median planning time dropped from 22 minutes to 3.5 minutes per case

These numbers matter because templating errors directly translate to intraoperative surprises, longer surgical times, and potentially suboptimal implant positioning.

Beyond Templating: Predictive Analytics

The real frontier is not just planning what implant to use, but predicting how it will perform. Machine learning models are now being trained on registry data from the AOANJRR, the National Joint Registry (UK), and the American Joint Replacement Registry to predict:

• Risk of early revision based on patient demographics, comorbidities, and implant choice
• Optimal bearing surface for a given patient's activity level and life expectancy
• Cement vs. uncemented fixation decision support, incorporating bone density metrics extracted automatically from pre-operative CT scans. For more on arthroplasty decision-making, see our guides on total knee arthroplasty and hip osteoarthritis

The Australian Orthopaedic Association National Joint Replacement Registry (AOANJRR) has been particularly instrumental in this space. With over 1.5 million procedures in its dataset, it provides the statistical power necessary to train models that can identify subtle risk patterns invisible to individual surgeons.

The Regulatory Landscape

Adoption has been accelerated by regulatory clearance. In the past 18 months:

• The TGA (Australia) has approved several AI planning platforms as Class IIa medical devices
• The FDA has cleared over 20 orthopaedic AI tools through the De Novo or 510(k) pathways
• CE marking under the new EU MDR framework has been slower, but major platforms now hold certification

The key regulatory distinction is between AI as a "decision support tool" (the surgeon remains the final decision-maker) and AI as an "autonomous system" (which makes decisions independently). All currently approved orthopaedic planning tools fall into the first category.

What This Means for Trainees

If you are an orthopaedic trainee in 2026, you need to understand these tools not as threats to your skills, but as amplifiers. The fundamentals have not changed: you still need to understand anatomy, biomechanics, and surgical approaches. What has changed is the expectation that you can critically appraise an AI-generated plan, identify when it has made an error, and override it with sound clinical judgment.

The surgeons who will thrive in this environment are those who combine deep clinical expertise with technological fluency. The ones who will struggle are those who either blindly trust the algorithm or refuse to engage with it altogether.

Looking Ahead

The next wave of AI surgical planning is moving toward intraoperative real-time guidance. Systems that continuously update the surgical plan based on live imaging, instrument tracking, and tissue feedback during the operation itself. Several companies have prototype systems in clinical trials, and the first commercial launches are expected by late 2027.

For now, the message is clear: AI-powered surgical planning is no longer experimental. It is mainstream, it is improving outcomes, and it is here to stay.