Artificial intelligence and orthopaedics

INTRODUCTION

A clinician evolves over a lifetime in the fields of disease diagnosis, clinical decision-making, and outcome prediction.¹ The quest is permanent as research uncovers newer facts every day. Even as it becomes impossible for an individual human brain to keep pace with the massive amounts of developing research data, artificial intelligence is making it possible to sort and assess data at unimaginable speeds. Artificial intelligence (AI), machine learning (ML), and deep learning (DL) are being rapidly adopted in industry and science.¹ This adoption is underpinned by the fact that 21st-century science has ensured that data is gold.² This is closely tied to improved computing power, cloud-based computing and development, and the development of medical task-specific software algorithms.² Conventional thinking would suggest that AI should increase healthcare delivery, improve indications and interventions, and minimize errors. However, clinical medicine does present some barriers unique to the field of human health. These include ethical concerns, research biases, and a lack of uniformity and consensus among health professionals. Even as AI heralds a new revolution, there is a mix of apprehension, concern, cynicism, and hope among medical professionals. The purpose of this review is to have a look at the current literature and assess the current and future role of AI in orthopaedic surgery.

We used a Boolean query to search Pubmed, Central, EMBASE, and Ovid for the terms orthopaedic trauma and artificial intelligence. There were 991 results. As artificial intelligence is a rapidly evolving branch, we considered papers published in the last 3 years only (2022-2024). This reduced the search to 400 papers. All authors participated in reading the abstracts of the papers. Papers discussing the same type of surgery with respect to artificial intelligence were excluded. Papers that did not discuss orthopedic trauma were also excluded. Papers comparing various AI models were also excluded. This narrowed the search to 31 papers. A review of these manuscripts is presented below.

DISCUSSION

John McCarthy first used the term Artificial Intelligence. The term, therefore, is not entirely new. AI is data-based and entirely reliant on inputs. One of its subsets, machine learning (ML), a field where machines can make decisions, represents a more intriguing future that could impact the field of medicine significantly.³

Artificial intelligence's (AI) accelerating progress demands rigorous evaluation standards to ensure safe, effective integration into healthcare's high-stakes decisions. As AI increasingly enables prediction, analysis, and judgment capabilities relevant to medicine, proper evaluation and interpretation are indispensable. Erroneous AI could endanger patients thus, developing, validating and deploying medical AI demands adhering to strict, transparent standards centre on safety, ethics and responsible oversight.⁴

AI can be supervised, where data is fed into the system, and predictable decision-making is witnessed. It can also be unsupervised, which is relatively unpredictable. However, this type of AI does not have inherent human biases. The unsupervised type is also known as deep learning.⁵ We will examine the present status of AI in orthopedic surgery under several subheadings, which will help us understand its status in disease diagnosis, clinical decision-making, and outcome prediction.

Automated diagnosis

This is a field that is expected to benefit a lot as AI evolves in the near future. Liu et al in their study in 2024 found that the YOLOX-SwinT network algorithm enhances the accuracy of AO/OTA subgroups classification of ITF by orthopedic trauma surgeons.⁶ Bhatnagar et al mentioned that among the most prevalent causes of inaccurate diagnosis and medical lawsuits is the overlooked fracture on radiographs taken in the emergency room, which can range from 2% to 9%. The workforce will soon be under a great deal of strain due to the growing demand for fracture detection on multiple imaging modalities. AI, according to them, can overcome this issue.⁷

DellAria et al, in an interesting study, found that AI performed better than the radiologists in detecting common fractures, but not subtle fractures.⁸ In a study in 2024, Yuh et al, concluded that deep learning-assisted measurement is expected to expedite the diagnostic process and enhance reliability, particularly benefiting less experienced clinicians.⁹ Deep learning models can identify implants before hardware removal. This information is lacking in 10 percent of cases where implant removal is planned, causing problems during the procedure. Obviously, prior knowledge can be of help in such situations.¹⁰

Radiological imaging is central to the establishment of orthopedic diagnosis. Current studies with AI show that the diagnosis rates tend to approach expert levels with AI. However, AI will be used more widely if it can reduce radiology time and interpretation time, lower misinterpretation, improve patient outcomes, and reduce complication rates.¹¹

Fracture fixation

Zhao et al used TiRobot assisted F screw technique and inverted triangle parallel nail internal fixation in the treatment of unstable femoral neck fractures. They found that the percutaneous cannulated F screw technique using Tirobot navigation positioning system is a safe and effective treatment for patients with unstable femoral neck fractures.¹² Jiang et al compared femoral neck fixation with and without AI assistance. AI assisted surgery showed less fluoroscopy times, guide needle drilling times, hollow screw replacement times, and intraoperative bleeding volume. All other parameters remained comparable.¹³

Rakhshankah et al, concluded their manuscript on avascular necrosis of the femoral head. The ability of AI to achieve high accuracy with remarkable efficiency makes it promising for early detection and intervention, ultimately contributing to improved patient outcomes.¹⁴ Paik et al, studied vertebral compression fractures. Their main focus was detection. They concluded that their model could be a potential tool for detecting VCFs from a simple radiograph and assisting doctors in making appropriate decisions in initial diagnosis.¹⁵ Preoperative planning of implants aided by AI can streamline surgeon workflow.¹⁶

Outcome prediction

Several models have been tested for outcome prediction including Support Vector Machine [SVM], Least Absolute Shrinkage and Selection Operator [LASSO] and random forest.¹⁷ Nijhuis et al, used AI models to detect and classify distal end radius fractures. They rated the ability of AI as ‘promising’. However, they were unable to detect any algorithm that could predict loss of alignment.¹⁸

Knee et al. conducted a similar study and concluded that Chat GPT has significant limitations and is of limited use in its current format.¹⁹ Silva et al, made a similar observation regarding the use of AI for outcome prediction in femoral neck fractures. They felt that AI was in its infancy.²⁰ Yao et al, used a Python-based extraction tool to make an automatic diagnosis of supracondylar fractures. They concluded that the method had potential.²¹ Perhaps Kaspovic et al, reached a reasonable conclusion regarding the current status of AI in orthopedic trauma. They mentioned that while AI-based language models like ChatGPT hold significant promise for medical education and patient care, the current quality of information provided in orthopedics and trauma surgery is suboptimal. Further development and refinement of AI sources and algorithms are necessary to improve the accuracy and reliability of medical information.²²

However, Kaarre et al, praised the ability of Chat GPT for its correctness, completeness, and adaptiveness. The adaptiveness is perhaps something that could change the current consensus over a period of time.²³

Complication prediction

Dai et al, studied the prediction capability of AI in elderly patients for the development of post-operative comorbidity. This is an area that could see early implementation in the future.²⁴ Whiteside et al used AI as a predictor. They felt that an artificial intelligence algorithm that analyzes the pulse oximeter waveform in the fingertip can be used to determine the compensatory reserve index (CRI) in trauma patients. Thus, one can identify patients at risk for transfusion and the need for urgent medical care and may aid in the management of blood loss and transfusion in major orthopedic surgery.²⁵ Naranjo et al, used AI similarly to predict surgical site infections. They felt that more data is required to make the predictive power more reliable.²⁶

Length of hospital stay

Lai et al, showed that AI models could predict the length of hospital stay with fair accuracy.²⁷ Youssef et al, recommended AI for data-driven, personalized patient care, strengthening patients' self-responsibility and supporting interdisciplinary healthcare in orthopaedic trauma.²⁸

Preventive medicine

Shariatzadeh et al, developed models which they felt could serve as reliable screening tools to assess the risk of distal radial fracture in the general population before bone mineral density testing. In addition, they can be integrated into decision support systems to help healthcare providers identify high-risk patients for additional evaluation and education.²⁹ Chalhoub et al, concluded that ChatGPT falls short in providing reliable management recommendations, with a 30% misdiagnosis rate and 53% mismanagement rate in our study. Its limitations, including reliance on outdated data and the inability to interactively gather patient information, must be acknowledged.³⁰

Is it time to store and update personal data?

AI has the capacity to create and maintain individual data sets. As an example, healthy patients can regularly update their gait analysis, bone density, body mass index, baseline radiographs and hematology and serology. All these could factor into fracture management if there is a subsequent injury. Farhadi et al, put it beautifully when they divided future areas into two groups. Areas of interpretive uncertainty, and areas of substantial time and resource consumption. Both areas need separate AI application modules, with the former needing a strong ethical debate.¹ Lex JR et al, wrote that complicated, uninterpretable models may not provide benefit over traditional, interpretable models for patient specific outcome prediction.³¹

CONCLUSION

While AI is set to change radiology, outcome prediction, and rehabilitation, its use on a broader scale in the management of orthopedic injuries is still not possible. One obstacle to this end is the problem of data acquisition. Data security, legal issues, data heterogeneity, and ethics have to be debated and discussed before further progress can be made in this regard.

Funding: No funding sources

Conflict of interest: None declared

Ethical approval: Not required

REFERENCES

1. 1. Farhadi F, Barnes MR, Sugito HR, Sin JM, Henderson ER, Levy JJ. Applications of artificial intelligence in orthopaedic surgery. Front Med Technol. 2022;4:995526.
2. 2. Vaish A, Migliorini F, Vaishya R. Artificial intelligence in foot and ankle surgery: current concepts. Orthopadie (Heidelb). 2023;52(12):1011-6.
3. 3. A proposal for the dartmouth summer research project on artificial intelligence. Front Med Technol. 1955;2:34.
4. 4. Oettl FC, Pareek A, Winkler PW, Zsidai B, Pruneski JA, Senorski EH, et al. ESSKA Artificial Intelligence Working Group. A practical guide to the implementation of AI in orthopaedic research, Part 6: How to evaluate the performance of AI research? J Exp Orthop. 2024;11(3):12039.
5. 5. Khan A, Laghari AA, Ahmed Awan S. Machine learning in computer vision: a review. Endorsed Scal Inf Syst. 2021;8(32):4.
6. 6. Liu XS, Nie R, Duan AW, Yang L, Li X, Zhang LT, et al. YOLOX-SwinT algorithm improves the accuracy of AO/OTA classification of intertrochanteric fractures by orthopedic trauma surgeons. Chin J Traumatol. 2024;12(24):51-8.
7. 7. Bhatnagar A, Kekatpure AL, Velagala VR, Kekatpure A. A Review on the use of artificial intelligence in fracture detection. Cureus. 2024;16(4):58364.
8. 8. Dell'Aria A, Tack D, Saddiki N, Makdoud S, Alexiou J, De Hemptinne FX, et al. Radiographic detection of post-traumatic bone fractures: contribution of artificial intelligence software to the analysis of senior and junior radiologists. J Belg Soc Radiol. 2024;108(1):44.
9. 9. Yuh WT, Khil EK, Yoon YS, Kim B, Yoon H, Lim J, et al. Deep learning-assisted quantitative measurement of thoracolumbar fracture features on lateral radiographs. Neurospine. 2024;21(1):30-43.
10. 10. Wilson NA, Jehn M, York S, Davis CM. Revision total hip and knee arthroplasty implant identification: implications for use of Unique Device Identification 2012 AAHKS member survey results. J Arthroplasty. 2014;29(2):251-5.
11. 11. Krogue JD, Cheng KV, Hwang KM, Toogood P, Meinberg EG, Geiger EJ, et al. Automatic hip fracture identification and functional subclassification with deep learning. Radiol Artif Intell. 2020;2(2):190023.
12. 12. Zhao XL, Shen JJ, Feng KH, Chen ZW, Si YL, Zhang X, et al. Comparison of the efficacy of TiRobot orthopaedic robot assisted F screw technique and inverted triangle parallel nail internal fixation in the treatment of unstable femoral neck fractures. Zhongguo Gu Shang. 2024;37(2):129-34.
13. 13. Jiang SH, Zhang CK, Jia FT, Chen Q, Xu M, Yang PL, et al. Orthopaedic robot assisted closed reduction and cannulated screw internal fixation for the treatment of femoral neck fractures. Zhongguo Gu Shang. 2024;37(2):119-23.
14. 14. Rakhshankhah N, Abbaszadeh M, Kazemi A, Rezaei SS, Roozpeykar S, Arabfard M. Deep learning approach to femoral AVN detection in digital radiography: differentiating patients and pre-collapse stages. BMC Musculoskelet Disord. 2024;25(1):547.
15. 15. Paik S, Park J, Hong JY, Han SW. Deep learning application of vertebral compression fracture detection using mask R-CNN. Sci Rep. 2024;14(1):16308.
16. 16. Lambrechts A, Wirix-Speetjens R, Maes F, Van Huffel S. Artificial intelligence-based patient-specific preoperative planning algorithm for total knee arthroplasty. Front Robot AI. 2022;8(9):840282.
17. 17. Fontana MA, Lyman S, Sarker GK, Padgett DE, MacLean CH. Can machine learning algorithms predict which patients will achieve minimally clinically important differences from total joint arthroplasty? Clin Orthop Relat Res. 2019;477(6):1267-79.
18. 18. Oude Nijhuis KD, Dankelman LHM, Wiersma JP, Barvelink B, IJpma FFA, Verhofstad MHJ, et al. Machine learning consortium. AI for detection, classification and prediction of loss of alignment of distal radius fractures; a systematic review. Eur J Trauma Emerg Surg. 2024;68:24-57.
19. 19. Knee CJ, Campbell RJ, Graham DJ, Handford C, Symes M, Sivakumar BS. Examining the role of ChatGPT in the management of distal radius fractures: insights into its accuracy and consistency. ANZ J Surg. 2024;19:143.
20. 20. Silva G, Ashford R. Using Artificial Intelligence to predict outcomes of operatively managed neck of femur fractures. Br J Hosp Med. 2024;85(6):1-12.
21. 21. Yao W, Wang Y, Zhao X, He M, Wang Q, Liu H, Zhao J. Automatic diagnosis of pediatric supracondylar humerus fractures using radiomics-based machine learning. Medicine (Baltimore). 2024;103(23):38503.
22. 22. Kasapovic A, Ali T, Babasiz M, Bojko J, Gathen M, Kaczmarczyk R, Roos J. Does the information quality of ChatGPT meet the requirements of orthopedics and trauma surgery? Cureus. 2024;16(5):60318.
23. 23. Kaarre J, Feldt R, Zsidai B, Senorski EH, Rydberg EM, Wolf O, et al. ChatGPT can yield valuable responses in the context of orthopaedic trauma surgery. J Exp Orthop. 2024;11(3):12047.
24. 24. Dai A, Liu H, Shen P, Feng Y, Zhong Y, Ma M, et al. Incorporating preoperative frailty to assist in early prediction of postoperative pneumonia in elderly patients with hip fractures: an externally validated online interpretable machine learning model. BMC Geriatr. 2024;24(1):472.
25. 25. Whiteside LA, Roy ME. Use of an artificial intelligence device for evaluating blood loss in complex major orthopaedic surgery procedures. J Arthroplasty. 2024;8(1):53-8.
26. 26. Gutierrez-Naranjo JM, Moreira A, Valero-Moreno E, Bullock TS, Ogden LA, Zelle BA. A machine learning model to predict surgical site infection after surgery of lower extremity fractures. Int Orthop. 2024;48(7):1887-96.
27. 27. Lai CH, Mok PK, Chau WW, Law SW. Application of machine learning models on predicting the length of hospital stay in fragility fracture patients. BMC Med Inform Decis Mak. 2024;24(1):26.
28. 28. Youssef Y, De Wet D, Back DA, Scherer J. Digitalization in orthopaedics: a narrative review. Front Surg. 2024;10:1325423.
29. 29. Shariatzadeh H, Dashtbozorg A, Gorjizadeh N. Association of distal radial fracture with comorbidities: model development and validation. Injury. 2024;55(7):111607.
30. 30. Chalhoub R, Mouawad A, Aoun M, Daher M, El-Sett P, Kreichati G, Kharrat K, et al. Will ChatGPT be able to replace a spine surgeon in the clinical setting? World Neurosurg. 2024;185:648-52.
31. 31. Lex JR, Di Michele J, Koucheki R, Pincus D, Whyne C, Ravi B. Artificial intelligence for hip fracture detection and outcome prediction: A systematic review and meta-analysis. JAMA Netw Open. 2023;6(3):233-391.