AI for Early Acute Kidney Injury Prediction During Surgery

A recent review conducted by Hanfei Zhang and team shows the potential of artificial intelligence for early acute kidney injury (AKI) prediction during surgery. The findings of this study were published in BMC Nephrology.

In a variety of surgical scenarios, acute kidney damage is independently linked to morbidity and death. Artificial intelligence is now being utilized more and more to enhance perioperative AKI early diagnosis and management due to the growing usage of electronic health records (EHR), improvements in patient information retrieval, and cost savings in clinical informatics. However, there isn't a quantitative analysis of how well these strategies function. In order to determine the sensitivity and specificity of artificial intelligence for the forecasting of acute kidney damage during the perioperative phase, this systematic review and meta-analysis was carried out.

Up till October 2, 2021, Embase, Pubmed, and Cochrane Library were searched. Studies demonstrating the early diagnosis of perioperative acute renal damage using artificial intelligence were considered. To aggregate specificity and sensitivity with 95% CIs, true positives, false positives, true negatives, and false negatives were combined. The findings were displayed in forest plots. The PROBAST method was used to evaluate the risk of bias in the qualifying studies.

Key Findings

• There were 304,076 patients in 19 trials total.
• The Rutter and Gatsonis hierarchical summary receiver operating characteristics (HSROC) model was used to perform a quantitative random-effects meta-analysis.
• The results showed that the diagnostic odds ratio, pooled sensitivity, and specificity were all 0.77 (95% CI: 0.73 to 0.81), 0.75 (95% CI: 0.71 to 0.80), and 10.7 (95% CI: 8.5 to 13.5), respectively.
• There was no proof of publishing bias, and the threshold effect was discovered to be the only source of heterogeneity.

In conclusion, none of the pre-specified subgroups had an effect on the predicted accuracy, according to this study. It was argued that without technical innovation, the advancement of artificial intelligence may have reached a plateau, making it challenging to further optimize forecast accuracy using current techniques.

Reference

Zhang, H., Wang, A. Y., Wu, S., Ngo, J., Feng, Y., He, X., Zhang, Y., Wu, X., & Hong, D. (2022). Artificial intelligence for the prediction of acute kidney injury during the perioperative period: systematic review and Meta-analysis of diagnostic test accuracy. In BMC Nephrology (Vol. 23, Issue 1). Springer Science and Business Media LLC. https://doi.org/10.1186/s12882-022-03025-w