Andrew R. Post

The Analytic Information Warehouse (AIW): A platform for analytics using electronic health record data

OBJECTIVE To create an analytics platform for specifying and detecting clinical phenotypes and other derived variables in electronic health record (EHR) data for quality improvement investigations. MATERIALS AND METHODS We have developed an architecture for an Analytic Information Warehouse (AIW). It supports transforming data represented in different physical schemas into a common data model, specifying derived variables in terms of the common model to enable their reuse, computing derived variables while enforcing invariants and ensuring correctness and consistency of data transformations, long-term curation of derived data, and export of derived data into standard analysis tools. It includes software that implements these features and a computing environment that enables secure high-performance access to and processing of large datasets extracted from EHRs. RESULTS We have implemented and deployed the architecture in production locally. The software is available as open source. We have used it as part of hospital operations in a project to reduce rates of hospital readmission within 30days. The project examined the association of over 100 derived variables representing disease and co-morbidity phenotypes with readmissions in 5years of data from our institutions clinical data warehouse and the UHC Clinical Database (CDB). The CDB contains administrative data from over 200 hospitals that are in academic medical centers or affiliated with such centers. DISCUSSION AND CONCLUSION A widely available platform for managing and detecting phenotypes in EHR data could accelerate the use of such data in quality improvement and comparative effectiveness studies.

SHARE: system design and case studies for statistical health information release

OBJECTIVES We present SHARE, a new system for statistical health information release with differential privacy. We present two case studies that evaluate the software on real medical datasets and demonstrate the feasibility and utility of applying the differential privacy framework on biomedical data. MATERIALS AND METHODS SHARE releases statistical information in electronic health records with differential privacy, a strong privacy framework for statistical data release. It includes a number of state-of-the-art methods for releasing multidimensional histograms and longitudinal patterns. We performed a variety of experiments on two real datasets, the surveillance, epidemiology and end results (SEER) breast cancer dataset and the Emory electronic medical record (EeMR) dataset, to demonstrate the feasibility and utility of SHARE. RESULTS Experimental results indicate that SHARE can deal with heterogeneous data present in medical data, and that the released statistics are useful. The Kullback-Leibler divergence between the released multidimensional histograms and the original data distribution is below 0.5 and 0.01 for seven-dimensional and three-dimensional data cubes generated from the SEER dataset, respectively. The relative error for longitudinal pattern queries on the EeMR dataset varies between 0 and 0.3. While the results are promising, they also suggest that challenges remain in applying statistical data release using the differential privacy framework for higher dimensional data. CONCLUSIONS SHARE is one of the first systems to provide a mechanism for custodians to release differentially private aggregate statistics for a variety of use cases in the medical domain. This proof-of-concept system is intended to be applied to large-scale medical data warehouses.

An evaluation of feature sets and sampling techniques for de-identification of medical records

De-identification of text medical records is of critical importance in any health informatics system in order to facilitate research and sharing of medical records. While statistical learning based techniques have shown promising results for de-identification purposes, few such systems are publicly available. It remains a challenge for practitioners to build an accurate and efficient system as it involves a significant amount of feature engineering, i.e. creation and examination of new features used in the system. A comprehensive evaluation is needed to thoroughly understand the effects of different feature sets and potential impacts of sampling and their trade-offs between the often conflicting goals of precision (or positive predictive value), recall (or sensitivity), and efficiency. In this paper, we present the Health Information DE-identification (HIDE) framework and evaluate the open- source software. We present an evaluation of various types of features used in HIDE, and introduce a window sampling technique (only the terms within a specified distance from personal health information are used to train the classifier) and evaluate its effect on both quality and efficiency. Our results show that the context features (previous and next terms) are particularly important and the sampling technique can be used to increase recall with minimal impact on precision. We obtained token-level label precision of 0.967, recall of 0.986 and F-Score of 0.977 when not including true negatives. The overall HIDE system achieves token-level precision of .998, recall of .999, and f-score of .999 on the previous i2b2 challenge task.

Detection of Conflicts and Inconsistencies in Taxonomy-Based Authorization Policies

The values of data elements stored in biomedical databases often draw from biomedical ontologies. Authorization rules can be defined on these ontologies to control access to sensitive and private data elements in such databases. Authorization rules may be specified by different authorities at different times for various purposes. Since such policy rules can conflict with each other, access to sensitive information may inadvertently be allowed. Another problem in biomedical data protection is inference attacks, in which a user who has legitimate access to some data elements is able to infer information related to other data elements. We propose and evaluate two strategies, one for detecting policy inconsistencies to avoid potential inference attacks and the other for detecting policy conflicts.

Efficient regression testing of ontology-driven systems

To manage and integrate information gathered from heterogeneous databases, an ontology is often used. Like all systems, ontology-driven systems evolve over time and must be regression tested to gain confidence in the behavior of the modified system. Because rerunning all existing tests can be extremely expensive, researchers have developed regression-test-selection (RTS) techniques that select a subset of the available tests that are affected by the changes, and use this subset to test the modified system. Existing RTS techniques have been shown to be effective, but they operate on the code and are unable to handle changes that involve ontologies. To address this limitation, we developed and present in this paper a novel RTS technique that targets ontology-driven systems. Our technique creates representations of the old and new ontologies, compares them to identify entities affected by the changes, and uses this information to select the subset of tests to rerun. We also describe in this paper OntoRetest, a tool that implements our technique and that we used to empirically evaluate our approach on two biomedical ontology-driven database systems. The results of our evaluation show that our technique is both efficient and effective in selecting tests to rerun and in reducing the overall time required to perform regression testing.

Radiology Trainee vs Faculty Radiologist Fluoroscopy Time for Imaging-Guided Procedures: A Retrospective Study of 17,966 Reports Over a 5.5-Year Period

To evaluate differences in fluoroscopy time (FT) for common vascular access and gastrointestinal procedures performed by radiology trainees vs faculty radiologists. Report information was extracted for all 17,966 index fluoroscopy services performed by trainees or faculty, or both from 2 university hospitals over 66 months. Various vascular access procedures (eg, peripherally inserted central catheters [PICCs] and ports) and gastrointestinal fluoroscopy procedures (eg, upper gastrointestinal and contrast enema studies) were specifically targeted. Statistical analysis was performed. FT was recorded in 17,549 of 17,966 reports (98%) The 1393 procedures performed by nonphysician providers or transitional year interns were excluded. Residents, fellows, and faculty were primary operators in 5066, 6489, and 4601 procedures, respectively. Average FT (in seconds) for resident and fellow services, respectively, was less than that of faculty only for PICCs (75 and 101 vs 148, P < 0.01). For all other procedures, average FT of trainee services was greater than that for faculty. This was statistically significant (P < 0.05) for fellows vs faculty port placement (121 vs 87), resident vs faculty small bowel series (130 vs 96), and both resident and fellow vs faculty esophagram procedures (143 and 183 vs 126 ). FT for residents was significantly less than that for fellows only for PICCs (75 vs 101, P < 0.01). For most, but not all, fluoroscopy procedures commonly performed by radiology trainees, FT is greater than that for procedures performed by faculty radiologists. Better awareness and understanding of such differences may aid training programs in developing benchmarks, protocols, and focused teaching in the safe use of fluoroscopy for patients and operators.