Targeting repetitive laboratory testing with electronic health records-embedded predictive decision support: A pre-implementation study

Probabilistic Prediction of Laboratory Test Information Yield

Low-yield repetitive laboratory diagnostics burden patients and inflate cost of care. In this study, we assess whether stability in repeated laboratory diagnostic measurements is predictable with uncertainty estimates using electronic health record data available before the diagnostic is ordered. We use probabilistic regression to predict a distribution of plausible values, allowing use-time customization for various definitions of "stability" given dynamic ranges and clinical scenarios. After converting distributions into "stability" scores, the models achieve a sensitivity of 29% for white blood cells, 60% for hemoglobin, 100% for platelets, 54% for potassium, 99% for albumin and 35% for creatinine for predicting stability at 90% precision, suggesting those fractions of repetitive tests could be reduced with low risk of missing important changes. The findings demonstrate the feasibility of using electronic health record data to identify low-yield repetitive tests and offer personalized guidance for better usage of testing while ensuring high quality care.

Clinical and Analytic Accuracy of Simultaneously Acquired Hemoglobin Measurements: A Multi-Institution Cohort Study to Minimize Redundant Laboratory Usage

OBJECTIVES

Frequent diagnostic blood sampling contributes to anemia among critically ill children. Reducing duplicative hemoglobin testing while maintaining clinical accuracy can improve patient care efficacy. The objective of this study was to determine the analytical and clinical accuracy of simultaneously acquired hemoglobin measurements with different methods.

DESIGN

Retrospective cohort study.

SETTING

Two U.S. children's hospitals.

PATIENTS

Children (< 18 yr old) admitted to the PICU.

INTERVENTIONS

None.

MEASUREMENTS AND MAIN RESULTS

We identified hemoglobin results from complete blood count (CBC) panels paired with blood gas (BG) panels and point-of-care (POC) devices. We estimated analytic accuracy by comparing hemoglobin distributions, correlation coefficients, and Bland-Altman bias. We measured clinical accuracy with error grid analysis and defined mismatch zones as low, medium, or high risk-based on deviance from unity and risk of therapeutic error. We calculated pairwise agreement to a binary decision to transfuse based on a hemoglobin value. Our cohort includes 49,004 ICU admissions from 29,926 patients, resulting in 85,757 CBC-BG hemoglobin pairs. BG hemoglobin was significantly higher (mean bias, 0.43-0.58 g/dL) than CBC hemoglobin with similar Pearson correlation ( R2 ) (0.90-0.91). POC hemoglobin was also significantly higher, but of lower magnitude (mean bias, 0.14 g/dL). Error grid analysis revealed only 78 (< 0.1%) CBC-BG hemoglobin pairs in the high-risk zone. For CBC-BG hemoglobin pairs, at a BG hemoglobin cutoff of greater than 8.0 g/dL, the "number needed to miss" a CBC hemoglobin less than 7 g/dL was 275 and 474 at each institution, respectively.

CONCLUSIONS

In this pragmatic two-institution cohort of greater than 29,000 patients, we show similar clinical and analytic accuracy of CBC and BG hemoglobin. Although BG hemoglobin values are higher than CBC hemoglobin values, the small magnitude is unlikely to be clinically significant. Application of these findings may reduce duplicative testing and decrease anemia among critically ill children.

DEPLOYR: a technical framework for deploying custom real-time machine learning models into the electronic medical record

OBJECTIVE

Heatlhcare institutions are establishing frameworks to govern and promote the implementation of accurate, actionable, and reliable machine learning models that integrate with clinical workflow. Such governance frameworks require an accompanying technical framework to deploy models in a resource efficient, safe and high-quality manner. Here we present DEPLOYR, a technical framework for enabling real-time deployment and monitoring of researcher-created models into a widely used electronic medical record system.

MATERIALS AND METHODS

We discuss core functionality and design decisions, including mechanisms to trigger inference based on actions within electronic medical record software, modules that collect real-time data to make inferences, mechanisms that close-the-loop by displaying inferences back to end-users within their workflow, monitoring modules that track performance of deployed models over time, silent deployment capabilities, and mechanisms to prospectively evaluate a deployed model's impact.

RESULTS

We demonstrate the use of DEPLOYR by silently deploying and prospectively evaluating 12 machine learning models trained using electronic medical record data that predict laboratory diagnostic results, triggered by clinician button-clicks in Stanford Health Care's electronic medical record.

DISCUSSION

Our study highlights the need and feasibility for such silent deployment, because prospectively measured performance varies from retrospective estimates. When possible, we recommend using prospectively estimated performance measures during silent trials to make final go decisions for model deployment.

CONCLUSION

Machine learning applications in healthcare are extensively researched, but successful translations to the bedside are rare. By describing DEPLOYR, we aim to inform machine learning deployment best practices and help bridge the model implementation gap.