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

Objective Unlocking the data contained within both structured and unstructured components of Electronic Health Records (EHRs) has the potential to provide a step change in data available forsecondary research use, generation of actionable medical insights, hospital management and trial recruitment. To achieve this, we implemented SemEHR - a semantic search and analytics, open source tool for EHRs. Methods SemEHR implements a generic information extraction (IE) and retrieval infrastructure by identifying contextualised mentions of a wide range of biomedical concepts within EHRs. Natural Language Processing (NLP) annotations are further assembled at patient level and extended with EHR-specific knowledge to generate a timeline for each patient. The semantic data is serviced via ontology-based search and analytics interfaces. Results SemEHR has been deployed to a number of UK hospitals including the Clinical Record Interactive Search (CRIS), an anonymised replica of the EHR of the UK South London and Maudsley (SLaM) NHS Foundation Trust, one of Europes largest providers of mental health services. In two CRIS-based studies, SemEHR achieved 93% (Hepatitis C case) and 99% (HIV case) F-Measure results in identifying true positive patients. At King’s College Hospital in London, as part of the CogStack programme (github.com/cogstack), SemEHR is being used to recruit patients into the UK Dept of Health 100k Genome Project (genomicsengland.co.uk). The validation study suggests that the tool can validate previously recruited cases and is very fast in searching phenotypes - time for recruitment criteria checking reduced from days to minutes. Validated on an open intensive care EHR data - MIMICIII, the vital signs extracted by SemEHR can achieve around 97% accuracy. Conclusion Results from the multiple case studies demonstrate SemEHR’s efficiency - weeks or months of work can be done within hours or minutes in some cases. SemEHR provides a more comprehensive view of a patient, bringing in more and unexpected insight compared to study-oriented bespoke information extraction systems. SemEHR is open source available at https://github.com/CogStack/SemEHR.

[1]  Simon Lin,et al.  Application of clinical text data for phenome-wide association studies (PheWASs) , 2015, Bioinform..

[2]  Clement J. McDonald,et al.  Research and applications: Combining structured and unstructured data to identify a cohort of ICU patients who received dialysis , 2014, J. Am. Medical Informatics Assoc..

[3]  Angus Roberts,et al.  Extracting antipsychotic polypharmacy data from electronic health records: developing and evaluating a novel process , 2015, BMC Psychiatry.

[4]  William Pao,et al.  CUSTOM-SEQ: a prototype for oncology rapid learning in a comprehensive EHR environment , 2016, J. Am. Medical Informatics Assoc..

[5]  Leo Anthony Celi,et al.  Beyond Open Big Data: Addressing Unreliable Research , 2014, Journal of medical Internet research.

[6]  Peter Szolovits,et al.  MIMIC-III, a freely accessible critical care database , 2016, Scientific Data.

[7]  Markus Krötzsch,et al.  Wikidata , 2014, Commun. ACM.

[8]  G. Hartvigsen,et al.  Secondary Use of EHR: Data Quality Issues and Informatics Opportunities , 2010, Summit on translational bioinformatics.

[9]  K. Luyckx,et al.  Data integration of structured and unstructured sources for assigning clinical codes to patient stays , 2015, J. Am. Medical Informatics Assoc..

[10]  Michael Ball,et al.  TextHunter - A User Friendly Tool for Extracting Generic Concepts from Free Text in Clinical Research , 2014, AMIA.

[11]  Betsy L. Humphreys,et al.  The unified medical language system (UMLS) and computer-based patient records , 1992 .

[12]  Gabriela Vazquez-Benitez,et al.  Predicting neutropenia risk in patients with cancer using electronic data , 2017, J. Am. Medical Informatics Assoc..

[13]  Zina M. Ibrahim,et al.  Identification of Adverse Drug Events from Free Text Electronic Patient Records and Information in a Large Mental Health Case Register , 2015, PloS one.

[14]  Olivier Bodenreider,et al.  The Unified Medical Language System (UMLS): integrating biomedical terminology , 2004, Nucleic Acids Res..

[15]  Jens Lehmann,et al.  DBpedia: A Nucleus for a Web of Open Data , 2007, ISWC/ASWC.

[16]  Pedro Gullón,et al.  Population cardiovascular health and urban environments: the Heart Healthy Hoods exploratory study in Madrid, Spain , 2016, BMC Medical Research Methodology.

[17]  Zina M. Ibrahim,et al.  Encoding Medication Episodes for Adverse Drug Event Prediction , 2016, SGAI Conf..

[18]  Aziz Sheikh,et al.  Six ways for governments to get value from health IT , 2016, The Lancet.

[19]  R. Dobson,et al.  Natural language processing to extract symptoms of severe mental illness from clinical text: the Clinical Record Interactive Search Comprehensive Data Extraction (CRIS-CODE) project , 2017, BMJ Open.

[20]  David W. Bates,et al.  Ten key considerations for the successful optimization of large-scale health information technology , 2017, J. Am. Medical Informatics Assoc..

[21]  David W. Baker,et al.  Use of electronic health record data to evaluate overuse of cervical cancer screening , 2012, J. Am. Medical Informatics Assoc..

[22]  Graham Thornicroft,et al.  The South London and Maudsley NHS Foundation Trust Biomedical Research Centre (SLAM BRC) case register: development and descriptive data , 2009, BMC psychiatry.

[23]  Tudor Groza,et al.  CogStack - experiences of deploying integrated information retrieval and extraction services in a large National Health Service Foundation Trust hospital , 2017, bioRxiv.