A new medical review charts where artificial intelligence is already helping in orthopedics—and what still needs fixing.
Artificial intelligence (AI) is beginning to change how doctors diagnose, operate on, and rehab bone and joint problems—conditions that are rising as populations age. A 2025 review in the International Journal of Surgery highlights four big wins in orthopedics and sports medicine: sharper diagnostics, surgical decision‑support, rehab monitoring, and next‑generation training and research. A wheel‑shaped diagram on page 3 maps these domains at a glance.
What’s new—and why it matters
- Faster, more accurate diagnoses. AI systems now spot fractures, arthritis changes, and disc problems on X‑rays and MRIs with specialist‑level accuracy. In one limb X‑ray experiment, adding AI raised fracture‑detection sensitivity from about 81% to 94% and nearly halved misdiagnoses. AI models have also graded lumbar disc damage on MRI with ~96% accuracy.
- Surgical backup in the OR. From planning joint replacements to guiding screw placement in the spine, AI‑driven tools and robots can shorten procedures and reduce complications. One robotic system has been linked to an ~87% drop in intraoperative bleeding, and hip‑planning software can auto‑measure anatomy to optimize implant sizing and positioning. The graphic on pages 10–11 shows how multiple deep‑learning networks monitor key steps in hip surgery.
- Catching “hidden” injuries. Knee MRI programs modeled on MRNet can flag ligament tears and other occult injuries that busy clinicians might miss; the examples on pages 8–9 illustrate how AI highlights suspicious regions and boosts readers’ accuracy.
- Smarter rehab and follow‑up. Wearables and algorithms can track muscle activity, judge recovery, and even predict who’s on track after knee or hip replacement—with reported prediction accuracy up to 92% in one series—so care teams can intervene sooner.
What to watch
The authors stress that these tools aren’t push‑button cures. Many models still need large, diverse datasets to avoid errors and bias; some underperform when applied to different fracture types or patient groups. Important equity flags include differences in bone density and maturation across ethnicities that current algorithms may miss, potentially skewing diagnoses. And today, most advanced systems are limited to major centers. The review’s page 15–16 section details these bottlenecks and calls for broader data‑sharing and better validation.
The road ahead
Expect tighter integration with AR/VR training, 5G‑enabled remote support, and cross‑talk with fields like biomechanics and genomics—illustrated in the future‑directions panel on page 16. Done right, the payoff is practical: fewer missed fractures, safer surgeries, faster recoveries, and clearer guidance for patients and clinicians alike.
Source: Guan J, Li Z, Sheng S, et al. “An artificial intelligence‑driven revolution in orthopedic surgery and sports medicine.” International Journal of Surgery. 2025;111:2162–2181.