AI also has the potential to integrate imaging with pathology, molecular diagnostics, and clinical data. In orthopedic oncology, there's no single source that usually provides the complete answer. Diagnoses depend upon the relationship between imaging phenotype, histologic appearance, genomic alterations, anatomic location, and, of course, clinical behavior. Multimodal AI could help synthesize these different streams of information, although the ultimate interpretation [of this information] probably will remain with the expert multidisciplinary team.
What specific steps in orthopedic oncology workflows can AI assist in? How can AI improve surgical planning?
A second important area is surgical planning. AI-enabled segmentation can help define the 3-dimensional boundaries of a tumor and its relationship to critical structures such as blood vessels, nerves, articular surfaces, and, specifically in children, the growth plate. These models can help support navigation, patient-specific instruments, and customized implants. In selected cases, [AI-enabled segmentation] may help us determine whether limb salvage is feasible and how to achieve an appropriate oncologic margin while preserving as much function as possible.
An especially promising frontier among the convergence of AI with navigation and robotic surgery. AI is going to help translate preoperative imaging into a 3-dimensional operative plan, while robotic systems can assist the surgeon in executing that plan with a high degree of precision and reliability. In orthopedic oncology, [3-dimensional operative plans] could enable more accurate bone resections, preservation of critical anatomy, and improved positioning of patient-specific implants or reconstructions. It may be particularly valuable for complex pelvic and periarticular tumors, where millimeters can influence whether we preserve a joint, achieve an appropriate margin, or compromise function.
Eventually, these AI systems may incorporate intraoperative imaging and other real-time information, allowing the operative plan to adapt as surgery proceeds, this is exciting. We always talk about plan A, and if plan A isn't going well, we have to shift to plan B, however, AI in real time may be able to help us with that navigation of thought from plan A to plan B.
[Importantly,] robotics cannot determine tumor biology or substitute for oncologic judgment. Technical precision resection is only valuable if the correct resection is executed, and AI will be an additive layer of discrimination in that regard. The surgeon must remain responsible for determining the necessary margin, and we still need to interpret unexpected findings and [know] when the preoperative plan must change. It's just like an airplane pilot, there's a lot of automation in airplane flying now, but having a homo sapiens in the pilot seat in case things don't go right is always going to be of value.
Where else can AI be applied in clinical practice? How should AI integration be approached in orthopedic oncology?
AI may also improve risk prediction and surveillance. Orthopedic oncology encompasses diseases with widely differing probabilities of local recurrence, metastatic progression, complications, and functional recovery. Predictive AI models may help tailor surveillance, identify patients who need closer follow-up, and support more individualized discussions about treatment options. AI could also compare serial imaging, identify subtle interval changes, and prioritize examinations, helping us to figure out the best surveillance pattern for these patients.
[There are] also less visible, but immediate, practical applications. Generative AI can assist with clinical documentation, summarize lengthy outside records, identify relevant findings across multiple reports, and support multidisciplinary tumor board preparation. Even as of today, we know that AI can make mistakes; it can hallucinate information, [therefore], a content expert does need to review and be accountable for this information. [These AI applications] can diminish administrative burden, further enhancing the physician-patient relationship. The physician can spend more time with that emotional and medical connectivity with the patients because of AI.
AI can also accelerate research by identifying candidates for clinical trials and extracting structured information from records, which is particularly valuable while studying rare tumors across institutions, because all of us, even the big centers, only have so many patients that we see a year, and we do need to collaborate to get large data sets.
We nevertheless need to approach these technologies with discipline and discretion. Musculoskeletal tumors are rare, and data sets are often small, heterogeneous, and derived from specialized centers. An algorithm may perform well when [it] is developed but fail when applied to another population. We must address bias, privacy, transparency, regulatory oversight, and responsibility for errors. Every tool requires rigorous external validation and continued assessment after implementation.
The goal is not simply greater technological sophistication. It's earlier diagnosis, more precision, better oncologic outcomes, [and] better function. Everyone stands to benefit from the discretionary application of this exciting technology.
Editor's Note: This transcript has been edited for grammar and clarity using artificial intelligence tools.
References
- Ryu S, Imaizumi Y, Goto K, et al. Artificial intelligence-enhanced navigation for nerve recognition and surgical education in laparoscopic colorectal surgery. Surg Endosc. 2025;39(2):1388-1396. doi:10.1007/s00464-024-11489-0
- Parker CTA, Huang HC, Grist E, et al; STAMPEDE Collaborators. Multimodal artificial intelligence prediction of abiraterone efficacy in two STAMPEDE phase 3 trials of non-metastatic very high-risk prostate cancer. Ann Oncol. Published online June 5, 2026. doi:10.1016/j.annonc.2026.05.708
