*Purpose: Transplant immunosuppressive regimens include multiple drugs with varying pharmacokinetics, metabolic pathways, and drug-drug interactions. This study aimed to use CURATE.AI, an artificial intelligence platform, to make personalized tacrolimus dose recommendations based on each individual’s tacrolimus serum trough level profiles, while accounting for multi-drug regimen changes (e.g. changes in fluconazole or valganciclovir doses) and their corresponding interactions with trough levels using a powerful correlation termed phenotypic response surface. CURATE.AI has previously been clinically validated and has optimized combination therapy in applications as diverse as oncology, tuberculosis, and HIV treatment.

*Methods: IRB approval was obtained for this retrospective study of 40 liver transplant patients. CURATE.AI calibrated individualized tacrolimus dose-trough level profiles and population derived multidimensional drug-drug interaction profiles, which identified and minimized the patient-specific recalibration shifts from regimen changes.

*Results: CURATE.AI multi-dimensional drug-drug interaction profiles identified prednisone-fluconazole and valganciclovir-sulfamethoxazole dose changes corresponding to recalibration shifts in the sample population (Figure 1).

Integrating both individualized profiles and regimen change recalibration shifts, CURATE.AI demonstrated the robust identification of corresponding trough levels with clinically administered tacrolimus doses (Figure 2A). CURATE.AI-guided and clinically-titrated doses were different, and trough level managements were also compared (Figure 2B).

*Conclusions: CURATE.AI demonstrated consistent identification of measured trough levels corresponding to clinically administered tacrolimus dosing and dynamic trough level management by guiding tacrolimus dosing using only the individuals’ previous immunosuppression regimen doses and trough levels.

« Back to 2019 American Transplant Congress