Background: We aimed to develop and validate an automated machine learning (autoML) prediction model for cardiac surgery-associated acute kidney injury (CSA-AKI). Methods: Using 69 preoperative variables, we developed several models to predict post-operative AKI in adult patients undergoing cardiac surgery. Models included autoML and non-autoML types, including decision tree (DT), random forest (RF), extreme gradient boosting (XGBoost), and artificial neural network (ANN), as well as a logistic regression prediction model. We then compared model performance using area under the receiver operating characteristic curve (AUROC) and assessed model calibration using Brier score on the independent testing dataset. Results: The incidence of CSA-AKI was 36%. Stacked ensemble autoML had the highest predictive performance among autoML models, and was chosen for comparison with other non-autoML and multivariable logistic regression models. The autoML had the highest AUROC (0.79), followed by RF (0.78), XGBoost (0.77), multivariable logistic regression (0.77), ANN (0.75), and DT (0.64). The autoML had comparable AUROC with RF and outperformed the other models. The autoML was well-calibrated. The Brier score for autoML, RF, DT, XGBoost, ANN, and multivariable logistic regression was 0.18, 0.18, 0.21, 0.19, 0.19, and 0.18, respectively. We applied SHAP and LIME algorithms to our autoML prediction model to extract an explanation of the variables that drive patient-specific predictions of CSA-AKI. Conclusion: We were able to present a preoperative autoML prediction model for CSA-AKI that provided high predictive performance that was comparable to RF and superior to other ML and multivariable logistic regression models. The novel approaches of the proposed explainable preoperative autoML prediction model for CSA-AKI may guide clinicians in advancing individualized medicine plans for patients under cardiac surgery.

Women with advanced kidney disease are advised to wait until after transplant to pursue pregnancy, but the impact of pregnancy on estimated glomerular filtration rate (eGFR) decline and kidney histology is unclear. We identified a cohort of women aged 18 to 44 years at transplant from 1996 to 2014 at our 3-site program (N = 816) and determined whether they had a pregnancy >20 weeks gestation post-transplant by chart review. Outcomes included rate of change in eGFR after pregnancy, changes in kidney histology before and after pregnancy, graft failure, and 50% reduction in eGFR. There were 37 women with one or more pregnancies lasting longer than 20 weeks gestation post-transplant. Comparing women with and without pregnancy post-transplant, there was a significant increase in the rate of eGFR decline after pregnancy (−2.4 ml/min per 1.73 m² per year vs. −1.9 ml/min per 1.73 m² per year in women with no pregnancy, P < 0.001). Pregnancy did not affect the risk of graft failure, death-censored graft failure, or 50% reduction in eGFR. Pregnancy affects the rate of eGFR decline in the allograft. Postpregnancy biopsy findings revealed an increase in vascular injury, which could be a potential mechanism. We did not find a significant increase in risk of graft failure or reduction in eGFR by 50% owing to pregnancy.

BACKGROUND: Senescence, a mechanism of cellular aging, which is characterized by irreversible proliferation arrest and a proinflammatory secretory phenotype, has been documented in women with preeclampsia. As cellular senescence can persist and progress, we postulated that it is associated with accelerated aging phenotype and accumulation of comorbidities in women with a history of preeclampsia. METHODS: We included a cohort of women with a history of preeclampsia (n=40) age- and parity-matched to a group of referent women with normotensive pregnancies (n=40). Women with prior major cardiovascular events, neurological, or autoimmune conditions were excluded. We collected urine and blood samples to study markers of aging, data on multimorbidity at the time of enrollment, and prospectively followed them for events over the course of 6 years, on average. RESULTS: Women with a history of preeclampsia exhibited unfavorable aging profiles compared with referent women, including decreased urinary α-Klotho ( P =0.018); increased leptin ( P =0.016) and leptin/adiponectin ratio ( P =0.027), and increased extracellular vesicles positive for tissue factor ( P =0.025). Women with a history of preeclampsia likewise had a higher rate of comorbidities at the time of enrollment ( P =0.003) and had a 4× higher risk of developing major cardiovascular events compared with referent women ( P =0.003). CONCLUSIONS: Our data suggest that a history of preeclampsia is associated with accelerated aging as indicated by senescence marker differences and the accumulation of multimorbidity later in life. Targeting cellular senescence may offer novel, mechanism-based approaches for the diagnosis and treatment of adverse health outcomes in women with a history of preeclampsia.