Introduction:
Anemia is common in patients suffering from chronic kidney disease (CKD). The multifactorial pathophysiology of CKD anemia and differences in patient population characteristics add challenges to anemia management. To assist clinical decision making, Fresenius Medical Care (FMC) developed the Anemia Control Model (ACM), an Artificial Intelligence (AI)-based medical device designed to suggest Erythropoietin Stimulating agents (ESA)/Iron dosage optimized for personalized patient characteristics. Since 2014, ACM has been utilized by the European FMC Nephrocare network and has well-documented experience (Barbieri et al, 2016; Garbelli et al, 2024). In recent years, ACM was successfully implemented in FMC clinics in Singapore. The objective of this study is to evaluate the clinical efficacy of ACM and physician’s response to ACM suggestion in the Singapore cohort.
Methods:
This is a retrospective study of patients receiving Hemodialysis (HD) at FMC clinics in Singapore between March 2023 and March 2024. We analyzed percentages of patients reaching the haemoglobin (Hb) target when ACM’s dosage suggestions were accepted and not accepted. We also analyzed outcomes of patients based on difference between ESA prescription and ESA administration. The prescribed monthly dose is based on the total ESA dosage specified for the month. The administered monthly dose, which may differ from the prescribed dose due to various clinical factors, is the sum of all ESA doses actually administered during the month. The target Hb levels are 10-12 g/dL for patients treated with ESA and > 12 g/dL for patients not treated with ESA.
Results:
For the 12-month study period (March 2023 and February 2024), the ACM returned an average of 78.2% (± 3.75%) suggestions of the total monthly triggers. The average percentage of suggestions receiving validation by a physician was 72.3% (± 12.9%) of the monthly ACM suggestions. The average percentage of suggestions accepted by the physician was 51.2% (± 3.3%) out of the monthly validated suggestions. For the study period between April 2023 and March 2024), the percentages of patients achieving their Hb target were significantly higher (p < 0.05) in ACM-accepted-suggestions (76.7% (± 2.2%)) in comparison with ACM non-accepted suggestions (67.9% (± 2.8%)) (Figure 1). ACM suggestions, validated ACM suggestions, and ACM acceptance rate based on prescription showed an increase over time during the evaluation period. Importantly, when the ACM suggested dose matched the administered dose, the Hb target achievement was at its highest at 85%.
Conclusions:
Our results demonstrate an increasing acceptance of ACM as clinical decision aid among Singapore nephrologists, leading to improved Hb target achievement over time. Further studies in Asian populations are warranted to delineate the role of ACM in the anemia management of dialysis patients in Singapore.
I have potential conflict of interest to disclose.
The work is funded by Fresenius Medical Care.
I did not use generative AI and AI-assisted technologies in the writing process