CogStack - Experiences Of Deploying Integrated Information Retrieval And Extraction Services In A Large National Health Service Foundation Trust Hospital

Abstract

Background: Traditional health information systems are generally devised to support clinical data collection at the point of care. However, as the significance of the modern information economy expands in scope and permeates the healthcare domain, there is an increasing urgency for healthcare organisations to offer information systems that address the expectations of clinicians, researchers and the business intelligence community alike. Amongst other emergent requirements, the principal unmet need might be defined as the 3R principle (right data, right place, right time) to address deficiencies in organisational data flow while retaining the strict information governance policies that apply within the UK National Health Service (NHS). Here, we describe our work on creating and deploying a low cost structured and unstructured information retrieval and extraction architecture within King's College Hospital, the management of governance concerns and the associated use cases and cost saving opportunities that such components present. Results: To date, our CogStack architecture has processed over 300 million lines of clinical data, making it available for internal service improvement projects at King's College London. On generated data designed to simulate real world clinical text, our de-identification algorithm achieved up to 94% precision and up to 96% recall. Conclusion:We describe a toolkit which we feel is of huge value to the UK (and beyond) healthcare community. It is the only open source, easily deployable solution designed for the UK healthcare environment, in a landscape populated by expensive proprietary systems. Solutions such as these provide a crucial foundation for the genomic revolution in medicine.

Bleeding in cardiac patients prescribed antithrombotic drugs: Electronic health record phenotyping algorithms, incidence, trends and prognosis

“…Two clinicians (blinded to the ICD-10 and OPCS-4 codes recorded) reviewed the entire hospital record (charts, referral letters, discharge letters, imaging reports) for 283 patient hospital episodes from two large NHS Trusts (University College London Hospitals NHS Foundation Trust and Kings College Hospital NHS Foundation Trust). The hospital record corpus (14,364,947 words in total) was made available as a single text files per patient, through the use of CogStack(39), method of enterprise-wide retrieval and extraction architecture for structured and unstructured information which integrates data across multiple EHR systems in a hospital. Patient consent for reviewing these records was provided from the NIHR funded SIGNUM study of stroke patients.…”

UK phenomics platform for developing and validating EHR phenotypes: CALIBER

“…Bleeding assignments from the clinicians review was compared with those from the phenotyping algorithm and we estimated the PPV, NPV, sensitivity and specificity using the case review data as the "gold standard". We extracted hospital data (14,364,947 words) using CogStack [57] from the consented Stroke InvestiGation Network-Understanding Mechanisms (SIGNUM) study.…”

SemEHR: A General-purpose Semantic Search System to Surface Semantic Data from Clinical Notes for Tailored Care, Trial Recruitment and Clinical Research

“…data retrieval, information extraction and semantic indexing. CogStack [14], a data harmonisation and enterprise search toolkit for EHRs, is adopted in the data retrieval step to provide a unified interface to unstructured EHR data, which is often very heterogeneous in format and distributed in storage. Each document that flows out from medical history, laboratory results); the continuous learning subsystem (to be described in next subsection) learns the contexts from user assessed annotations (see Supplementary Material 1 for detail).…”

“…To realise a general-purpose biomedical information extraction (IE) system on EHRs, there are at least three fundamental challenges: a) syntactic heterogeneity: how to effectively access multi-modal/multisource EHR data that are almost certainly heterogeneous in formats, data models and access interfaces; b) knowledge coverage: how to cover all possible biomedical concepts that are required by potential use cases; c) context capturing: how to represent and capture the contexts associated with extracted concepts, and which are critical to understanding the clinical domain. To address these challenges, SemEHR architects a production infrastructure that integrates our previous work in the CogStack pipeline [14] to harmonise and cleanse heterogeneous records, using them to identify contextualised 4 mentions (negation, temporality and experiencer) of a wide range of biomedical concepts including SNOMED CT 1 , ICD-10 2 , LOINC 3 and Drug Ontology 4 . In addition, SemEHR automatically associates semantic types of annotations and their clinical contexts (derived from containing documents or sections) with dedicated extraction rules, which enables better IE capabilities such as populating the structured vital sign data from observation notes.…”