# The accuracy of artificial intelligence in 3D preoperative planning for total hip arthroplasty: A systematic review and meta‐analysis

## TL;DR

This study shows that AI improves accuracy in 3D preoperative planning for hip replacement surgery compared to traditional 2D methods.

## Contribution

The study introduces a meta-analysis confirming AI's superior accuracy in predicting implant sizes for total hip arthroplasty.

## Key findings

- AI predicted acetabular cup size with an odds ratio of 3.85 compared to traditional methods.
- AI predicted femoral stem size with an odds ratio of 3.28, showing significant improvement.
- AI predictions within one standard deviation showed odds ratios of 3.49 and 5.35 for cup and stem sizes, respectively.

## Abstract

This systematic review and meta‐analysis compare AI‐assisted 3‐dimensional (3D) preoperative planning in total hip arthroplasty (THA) to traditional 2‐dimensional (2D) templating.

PubMed, Scopus, and Embase were searched from inception until October 2024 for studies on the accuracy of 3D preoperative planning in THA. Statistical analysis was performed using R (v4.3.3) with a random‐effects model due to high heterogeneity. Odds ratios with 95% confidence intervals were calculated for dichotomous outcomes. Heterogeneity was assessed using the I² statistic, and publication bias was evaluated through funnel plots and Egger's test. The primary outcome was the accuracy of detecting acetabular cup and femoral stem size. This meta‐analysis followed PRISMA guidelines for systematic reviews.

Eight studies with 1371 participants from China were analysed. The mean age was 54.48 ± 12.98 years, and the mean BMI was 24.63 ± 3.73 kg/m². The Newcastle–Ottawa Scale (NOS) scores ranged from 6 to 9. The AI model effectively predicted acetabular cup and femoral stem sizes, with an odds ratio (OR) of 3.85 for the exact cup size (95% CI: 2.79–5.32; p < 0.0001) and an OR of 3.49 for predictions within one standard deviation (95% CI: 1.21–10.13; p = 0.0212). Heterogeneity was 42% and 81%, respectively. For the femoral stem, the AI achieved an OR of 3.28 for exact size predictions (95% CI: 2.56–4.22; p < 0.0001) and an OR of 5.35 for size within one standard deviation (95% CI: 3.84–7.45; p < 0.0001), with no significant heterogeneity (I² = 0%).

This meta‐analysis confirms that AI‐assisted 3D preoperative planning in THA provides better accuracy for predicting the acetabular cup and femoral stem sizes than traditional 2D templating methods. Further studies with larger sample sizes and more extended follow‐up periods across multiple countries are warranted to validate our findings.

Level III.