SLR - August 2023 - Madeline Walkner DPM, MPH
Title: Advancements in Artificial Intelligence for Foot and Ankle Surgery: A Systematic ReviewReference: Advancements in Artificial Intelligence for Foot and Ankle Surgery: A Systematic Review.
Gupta P, Kingston KA, O'Malley M, Williams RJ, Ramkumar PN.Foot Ankle Orthop. 2023 Feb 13;8(1):24730114221151079. doi: 10.1177/24730114221151079. eCollection 2023 Jan.PMID: 36817020
Level of Evidence: Level III
Reviewed By: Madeline Walkner DPM, MPH
Residency Program: Lenox Hill Hospital, New York, NY
Podiatric Relevance: Artificial intelligence (AI) has many applications in foot and ankle surgery and is currently in the forefront of healthcare. Applications of AI in healthcare range from interpretation of diagnostic imaging such as radiographs, ultrasound, CT and MRI, to predicting clinical outcomes. The uses for AI in foot and ankle surgery have only just started to be explored as increased applications are being developed. This study aims to review the use of AI in foot and ankle surgery by evaluating the performance and validity of current AI models.
Methods: PubMed/MEDLINE and Embase databases were used to conduct a systematic literature review. The database search was performed in December 2022. A variety of different foot and ankle anatomy terms were combined with either artificial intelligence, machine learning or deep learning to conduct the search. The primary outcomes evaluated were subject area, performance, and validity. Other variables such as number of participants, male to female ratio, average age, and outcomes of interest were also evaluated. Studies were then grouped into 3 categories: clinical predictions, image interpretation, and “other”. Area under the receiver operating characteristic curve (AUC) was used as a performance metric for AI in clinical predictions and image interpretation.
Results: Thirty-one studies were included, the study was broken down into three groups: AI use in image interpretation, AI use for making clinical predictions, and other. Fourteen articles were evaluated for image interpretation, 13 for clinical predictions, and 4 for 'other’. Most studies observed AI applications pertaining to ankle and calcaneus fractures, Lisfranc malalignment, hallux valgus, Achilles tendon pathologies, plantar fasciitis, and sports injuries. Area under the receiver operating characteristic curve ranged from 0.64 (poor) to 0.99 (excellent). Zero studies evaluating AI and clinical predictions were externally validated and only 2/31 studies in this review were externally validated.
Conclusion: This study concluded there is currently wide variation in the performance of current AI models in foot and ankle surgery. Most studies involved AI uses in interpreting imaging and making clinical predictions. AI is still very novel in foot and ankle surgery compared to other orthopedic areas such as hip and knee joints. Most studies reviewed in this article had small sample sizes and only two studies were externally validated. There is a clear need for more large-scale AI studies in foot and ankle surgery to help validate these AI models. The results from this study advocate for use of validated AI models to create more precise imaging analysis and surgical intervention. These AI models are importance for the advancement of foot and ankle surgery as they may allow surgeons to accurately predict long term clinical outcomes for patients who suffer from foot and ankle pathology. The field of medicine is everchanging and it is important for foot and ankle specialists to provide the most accurate, effective, and innovative approaches to their patients.