top of page
Search

Democratizing Measure Development with Federated Computing

Updated: May 6

Generative AI image.


Democratizing Measure Development with Federated Computing


Modern healthcare systems rely on quality measures to ensure that patients receive the best possible care and that healthcare providers are motivated to deliver high-quality services. Professional societies and healthcare providers develop these quality measures using centralized data repositories. However, accessing these repositories can be challenging because of privacy concerns and regulatory constraints. In contrast, Rhino Health's Federated Computing Platform (Rhino FCP) can transform this process by enabling access to data without typical privacy and regulatory constraints. This can democratize measure development to institutions without access to data to participate, marking a significant shift in healthcare technology. 


Federated Biostatistics and Analytics


By leveraging the Rhino FCP, Rhino Health empowers measure developers to conduct comprehensive analyses of proposed healthcare quality measures without ever taking possession of the data. This capability was recently demonstrated in a successful project by Rhino Health, showcasing the potential of Federated Computing in healthcare quality measurement. The Rhino Health team recreated the measure development effort for AHRQ Inpatient Quality Indicator 21: Cesarean Delivery Rate using publicly available clinical data, instilling confidence in the Rhino FCP’s capabilities. 


Such an analysis required encoding the quality measure across eight hospitals and performing statistical tests used in the measure development process. The Rhino SDK, an easy-to-use Python package, provides the necessary functionality. It equips quality measure developers with a comprehensive suite of statistical tools, instilling confidence in their analyses. The Rhino SDK enables healthcare data analysts and researchers to perform statistical analyses on federated data - a \capability that was not commercially available previously. 


Hypothesis Testing. The Rhino SDK includes parametric and non-parametric methods, such as t-tests, chi-square tests, ANOVA, and Wilcoxon tests.

Statistical Correlations. The SDK includes functions to calculate Pearson, Intraclass, and Spearman correlations, facilitating the analysis of various data types and both normal and skewed distributions.

Risk Adjustment Methods. Developers can employ advanced statistical models, such as logistic regression, linear regression, Poisson regression, Kaplan-Meier analysis, and Cox Proportional Hazards models to adjust for risk factors.


Biostatistics & Federated Biostatistics


Federated Biostatistics comes into play as it involves the application of biostatistical methods within the Federated Learning environment of the Rhino FCP. This application, which is part of the Federated Biostatistics and Analytics suite, allows the Rhino Health team to apply complex statistical techniques directly to the Federated Distributed Datasets solution workflow. These methods also provide a robust framework for assessing health interventions and clinical outcomes without necessitating the movement of sensitive data, thus ensuring patient privacy and compliance with stringent privacy laws.


By creating a Python notebook and using the Rhino SDK to perform their measure development analyses, measure developers can reduce the effort necessary to develop an auditable, transparent measure development analysis that can be reviewed by approving bodies like the Center for Medicare and Medicaid Services (CMS). Such notebooks can be exported as PDFs and transmitted to approving bodies as part of applications for novel quality measures.


Rhino Health AI Program Lead demonstrating the use of the Rhino SDK for Federated Biostatistics in healthcare, analyzing quality measures across multiple hospitals. The video-demo highlights the application of statistical methods like chi-squared tests, t-tests, and correlation analyses in evaluating healthcare quality indicators related to patient outcomes.


Conclusion


The democratization of measure development through federated data represents significant progress in healthcare analytics. Rhino Health's Federated Computing Platform and Federated Biostatistics offer a scalable, privacy-preserving solution that can transform how we access and analyze healthcare data. Rhino Health is paving the way for a new era of innovation in healthcare quality improvement by providing measure developers with the tools and technologies to navigate the complexities of healthcare data..


Transform healthcare initiatives with Rhino Health’s Federated Biostatistics — where security meets healthcare innovation.


Daniel Feller, PhD

AI Program Lead


Chris Laws

Chief Operating Officer


Lili Lau, MSF, MSIB

Director of Product Marketing


72 views0 comments

Comments


bottom of page