Evaluation of Machine Learning Approaches for Predicting Warfarin Discharge Dose in Cardiac Surgery Patients: Retrospective Algorithm Development and Validation Study.

BACKGROUND: Warfarin dosing in cardiac surgery patients is complicated by a heightened sensitivity to the drug, predisposing patients to adverse events. Predictive algorithms are therefore needed to guide warfarin dosing in cardiac surgery patients. OBJECTIVE: This study aimed to develop and validate an algorithm for predicting the warfarin dose needed to attain a therapeutic international normalized ratio (INR) at the time of discharge in cardiac surgery patients. METHODS: We abstracted variables influencing warfarin dosage from the records of 1031 encounters initiating warfarin between April 1, 2011, and November 29, 2019, at St Michael's Hospital in Toronto, Ontario, Canada. We compared the performance of penalized linear regression, k-nearest neighbors, random forest regression, gradient boosting, multivariate adaptive regression splines, and an ensemble model combining the predictions of the 5 regression models. We developed and validated separate models for predicting the warfarin dose required for achieving a discharge INR of 2.0-3.0 in patients undergoing all forms of cardiac surgery except mechanical mitral valve replacement and a discharge INR of 2.5-3.5 in patients receiving a mechanical mitral valve replacement. For the former, we selected 80% of encounters (n=780) who had initiated warfarin during their hospital admission and had achieved a target INR of 2.0-3.0 at the time of discharge as the training cohort. Following 10-fold cross-validation, model accuracy was evaluated in a test cohort comprised solely of cardiac surgery patients. For patients requiring a target INR of 2.5-3.5 (n=165), we used leave-p-out cross-validation (p=3 observations) to estimate model performance. For each approach, we determined the mean absolute error (MAE) and the proportion of predictions within 20% of the true warfarin dose. We retrospectively evaluated the best-performing algorithm in clinical practice by comparing the proportion of cardiovascular surgery patients discharged with a therapeutic INR before (April 2011 and July 2019) and following (September 2021 and May 2, 2022) its implementation in routine care. RESULTS: Random forest regression was the best-performing model for patients with a target INR of 2.0-3.0, an MAE of 1.13 mg, and 39.5% of predictions of falling within 20% of the actual therapeutic discharge dose. For patients with a target INR of 2.5-3.5, the ensemble model performed best, with an MAE of 1.11 mg and 43.6% of predictions being within 20% of the actual therapeutic discharge dose. The proportion of cardiovascular surgery patients discharged with a therapeutic INR before and following implementation of these algorithms in clinical practice was 47.5% (305/641) and 61.1% (11/18), respectively. CONCLUSIONS: Machine learning algorithms based on routinely available clinical data can help guide initial warfarin dosing in cardiac surgery patients and optimize the postsurgical anticoagulation of these patients.

Integration of a clinical pharmacist into a Canadian, urban emergency department: a prospective observational study.

OBJECTIVE: To evaluate the clinical and cost implications generated by a newly integrated ED pharmacist in a Canadian urban, university-affiliated tertiary care hospital. METHODS: A pharmacist documented all interventions that took place over a 5-week period. Interventions were assessed by a review panel for clinical significance and probability of harm had the intervention not occurred. Direct medication cost and cost avoidance as a result of interventions were calculated. KEY FINDINGS: The ED pharmacist made 421 interventions during the study period, 204 (48%) interventions were accepted at the time they were presented to the prescriber. After review, 53.9% of interventions were considered significant, and 52.9% were given a probability of patient harm of ≥50% had the intervention not occurred. Interventions resulted in an increase in direct medication costs of $1270, but generated a cost avoidance of $160 709. The projected direct medication cost estimate for one year was $13 208 with a cost avoidance of over $1.6 million. CONCLUSION: The integration of a pharmacist into a Canadian ED resulted in patient care interventions that were assessed as clinically significant, with a substantial projected cost avoidance.

MicroSPECT/CT imaging of co-expressed HER2 and EGFR on subcutaneous human tumor xenografts in athymic mice using ¹¹¹In-labeled bispecific radioimmunoconjugates.

Epidermal growth factor receptors (EGFR) form heterodimers with HER2 in breast cancer, and increased EGFR expression has been found in HER2-positive tumors resistant to trastuzumab (Herceptin). Our objective was to synthesize bispecific radioimmunoconjugates (bsRICs) that recognize HER2 and EGFR and evaluate their ability to image tumors in athymic mice that express one or both receptors by microSPECT/CT. Bispecific radioimmunoconjugates were constructed by conjugating maleimide-derivatized trastuzumab Fab fragments that bind HER2 to a thiolated form of EGF with an intervening 24 mer polyethylene glycol (PEG24) spacer. Bispecific radioimmunoconjugates were derivatized with diethylenetriaminepentaacetic acid for labeling with (111)In. The ability of (111)In-bsRICs to bind HER2 or EGFR was determined in competition assays using cells expressing one or both receptors. Tumor and normal tissue uptake were examined in CD1 athymic mice bearing subcutaneous tumor xenografts that expressed HER2, EGFR, or both receptors, with or without pre-administration of Fab or EGF to determine specificity. HER2 and EGFR binding and displacement of binding by competitors were found for (111)In-bsICs. The highest uptake of (111)In-bsRICs [7.3 ± 3.5 %ID/g] in 231-H2N human breast cancer xenografts (HER2+/EGFR+) occurred at 48 h post-injection. Pre-administration of trastuzumab Fab decreased uptake in SK-OV-3 (HER2+/EGFR-) human ovarian cancer xenografts from 7.1 ± 1.2 to 2.4 ± 1.5 %ID/g. Pre-administration of excess EGF decreased uptake in MDA-MB-231 (HER2-/EGFR+) human breast cancer xenografts from 5.9 ± 0.5 to 2.0 ± 0.1 %ID/g. All tumors were imaged by microSPECT/CT. We conclude that (111)In-bsRICs composed of trastuzumab Fab and EGF exhibited specific binding in vitro to tumor cells displaying HER2 or EGFR, and were taken up specifically in vivo in tumors expressing one or both receptors, permitting tumor visualization by microSPECT/CT. These agents may ultimately be useful for imaging heterodimerized HER2-EGFR complexes since their bivalent properties permit more avid binding to these complexes.