An efficient and accurate distributed learning algorithm for modeling multi-site zero-inflated count outcomes

Mackenzie J Edmondson,Justine Shults,Mitchell Maltenfort,Yong Chen,Christopher B Forrest,Rui Duan,Jiang Bian,Kenneth Locke,Chongliang Luo,Zhaoyi Chen,Patrick B Ryan

doi:10.1038/s41598-021-99078-2

Abstract

Clinical research networks (CRNs), made up of multiple healthcare systems each with patient data from several care sites, are beneficial for studying rare outcomes and increasing generalizability of results. While CRNs encourage sharing aggregate data across healthcare systems, individual systems within CRNs often cannot share patient-level data due to privacy regulations, prohibiting multi-site regression which requires an analyst to access all individual patient data pooled together. Meta-analysis is commonly used to model data stored at multiple institutions within a CRN but can result in biased estimation, most notably in rare-event contexts. We present a communication-efficient, privacy-preserving algorithm for modeling multi-site zero-inflated count outcomes within a CRN. Our method, a one-shot distributed algorithm for performing hurdle regression (ODAH), models zero-inflated count data stored in multiple sites without sharing patient-level data across sites, resulting in estimates closely approximating those that would be obtained in a pooled patient-level data analysis. We evaluate our method through extensive simulations and two real-world data applications using electronic health records: examining risk factors associated with pediatric avoidable hospitalization and modeling serious adverse event frequency associated with a colorectal cancer therapy. In simulations, ODAH produced bias less than 0.1% across all settings explored while meta-analysis estimates exhibited bias up to 12.7%, with meta-analysis performing worst in settings with high zero-inflation or low event rates. Across both applied analyses, ODAH estimates had less than 10% bias for 18 of 20 coefficients estimated, while meta-analysis estimates exhibited substantially higher bias. Relative to existing methods for distributed data analysis, ODAH offers a highly accurate, computationally efficient method for modeling multi-site zero-inflated count data.

Highlights

Clinical research networks (CRNs), made up of multiple healthcare systems each with patient data from several care sites, are beneficial for studying rare outcomes and increasing generalizability of results
Results from using one-shot distributed algorithm for performing hurdle regression (ODAH) to model serious adverse event (SAE) frequency in colorectal cancer patients using data from OneFlorida are shown in Fig. 7, displayed to the Children’s Hospital of Philadelphia (CHOP) avoidable hospitalizations (AHs) results
For four of the five log odds ratios estimated in the logistic component of the hurdle model, relative biases produced by ODAH were less than 7%

Summary

Introduction

Clinical research networks (CRNs), made up of multiple healthcare systems each with patient data from several care sites, are beneficial for studying rare outcomes and increasing generalizability of results. A one-shot distributed algorithm for performing hurdle regression (ODAH), models zero-inflated count data stored in multiple sites without sharing patient-level data across sites, resulting in estimates closely approximating those that would be obtained in a pooled patient-level data analysis. RWE quality can be substantially improved from analyzing pooled data, patient records aggregated across health systems This is especially true in the context of studying rare outcomes, where outcome prevalence at any single institution may not be large enough to result in an analysis with meaningful conclusions. While useful in certain settings, the computation time required for blockchain-based analyses can be a limitation in healthcare settings where both accurate and efficient solutions are desirable

Methods

Results

Discussion

Conclusion