Free-text Narratives Research Articles

Electronic cigarette exposures reported to the British Columbia Drug and Poison Information Centre: an observational case series.

Electronic nicotine delivery systems (ENDSs), including electronic cigarettes (e-cigarettes), are rapidly gaining popularity. The aim of this study was to use poison centre data to describe epidemiological trends in ENDS-related exposures. We conducted an observational case series study using records containing both coded fields and free-text narratives from the British Columbia Drug and Poison Information Centre for all calls involving exposure to ENDS received from 2012 to 2017. We described trends in exposures and exposed people, as well as clinical effects. A total of 243 calls were recorded for 186 unique exposures to ENDS devices, e-juice, e-cigarette cartridges and other associated paraphernalia over the study period. Calls related to ENDS exposures increased nearly sixfold between 2013 and 2014 and did not decline subsequently. Exposures were most frequently documented in children aged 4 years or less (81 [43.5%]), with 58 (31.0%) in 1- and 2-year-olds. Seventy-two exposures (89%) in children aged 4 years or less were due to accidental ingestion, whereas adults aged 25 years or more called the poison centre following ENDS malfunctions (7 [23%], spills (4 [13%]) and exposure to e-juice mistaken for other substances (4 [13%]). Of the 186 exposed people, 87 (46.8%) reported symptoms. British Columbia experienced a sixfold increase in ENDS-related calls to the provincial poison centre between 2012 and 2017, driven by ingestions in young children. Regulatory approaches aimed at minimizing children's access to ENDS, clear labelling of nicotine concentration, and packaging that reduces the likelihood of spills, product confusion and malfunction should be considered.

Encountering complexity in collaborative learning activities: an exploratory case study with undergraduate health professionals

ABSTRACT An emerging challenge for educators of the health professions in undergraduate programmes is appropriately and adequately preparing students for the complexity they will encounter in practice. New graduate health professionals report increasing challenges transitioning to practice in contemporary health care environments due to the complexity encountered. Although literature on complexity science in health professional education is growing, research evidence on learning opportunities for students related to complexity remains sparse. This exploratory case study examined first-year health care students’ encounters with complexity in a student-driven collaborative learning activity taking place over one academic semester outside the classroom. Rather than reproducing complexity in a scripted simulation activity, these novices were stimulated to adapt and respond to authentic complexity through a challenging amount of independence and definition of only minimal enabling constraints i.e. basic ground rules of the learning activity and goals to be achieved. 67 students were assigned to small groups with four to five members to form 15 interprofessional collectives (laboratory technology n = 9; medicine n = 13; midwifery n = 1; nursing n = 28; physiotherapy n = 7; radiography n = 4; speech language therapy n = 5). Data was collected using a self-reflection form with narrative free text hand-written answers. Open coding and an inductive-deductive approach to qualitative content analysis of transcribed responses resulted in identification of two key themes: a) reflecting on interprofessional collaboration, and b) evaluating the interprofessional collectives. Our findings showed that rather than perceiving the learning activity in its cognitive and collaborative complexity and feeling overwhelmed, first-year health care students described rising to the challenge and stretching outside their comfort zones. Complexity was tacitly experienced but perceived as a stimulating challenge and successfully navigated within the interprofessional collectives as they collaborated together over time.

Incentivising use of structured language in biological descriptions: Author-driven phenotype data and ontology production.

Phenotypes are used for a multitude of purposes such as defining species, reconstructing phylogenies, diagnosing diseases or improving crop and animal productivity, but most of this phenotypic data is published in free-text narratives that are not computable. This means that the complex relationship between the genome, the environment and phenotypes is largely inaccessible to analysis and important questions related to the evolution of organisms, their diseases or their response to climate change cannot be fully addressed. It takes great effort to manually convert free-text narratives to a computable format before they can be used in large-scale analyses. We argue that this manual curation approach is not a sustainable solution to produce computable phenotypic data for three reasons: 1) it does not scale to all of biodiversity; 2) it does not stop the publication of free-text phenotypes that will continue to need manual curation in the future and, most importantly, 3) It does not solve the problem of inter-curator variation (curators interpret/convert a phenotype differently from each other). Our empirical studies have shown that inter-curator variation is as high as 40% even within a single project. With this level of variation, it is difficult to imagine that data integrated from multiple curation projects can be of high quality. The key causes of this variation have been identified as semantic vagueness in original phenotype descriptions and difficulties in using standardised vocabularies (ontologies). We argue that the authors describing phenotypes are the key to the solution. Given the right tools and appropriate attribution, the authors should be in charge of developing a project’s semantics and ontology. This will speed up ontology development and improve the semantic clarity of phenotype descriptions from the moment of publication. A proof of concept project on this idea was funded by NSF ABI in July 2017. We seek readers input or critique of the proposed approaches to help achieve community-based computable phenotype data production in the near future. Results from this project will be accessible through https://biosemantics.github.io/author-driven-production.

Deep Learning Facilitates Rapid Cohort Identification Using Human and Veterinary Clinical Narratives

Background: In public health research, there is currently a need to close the gap between care delivery and cohort identification. We need dedicated tagging staff to allocate a considerable amount of effort to assigning clinical codes after reading patient summaries. Machine learning automation can facilitate the classification of these clinical narratives, but sufficient availability of electronic medical records is still a bottleneck. Veterinary medical records represent a largely untapped data source that could be used to benefit both human and non-human patients. Very few approaches utilizing veterinary data sources currently exist. Methods: In this retrospective cross-sectional and chart review study, we trained separate long short-term memory (LSTM) Recurrent Neural Networks (RNNs) on 52,722 human records and 89,591 veterinary records, tested the models' efficacy in a standard train-test split setup, and probed the portability of these models across species domains. We trained versions of our models using first the free-text clinical narratives, and then only using extracted clinically relevant terms from MetaMap Lite, a natural language processing tool intended for this purpose. Findings: We show that our LSTM approach correctly classifies across toplevel codes in the veterinary records (F1 score =0·83), and identifies top-level neoplasia records in veterinary records (F1 score = 0·93). The model trained with veterinary data can be ported over to identify neoplasia records in the human records (F1 score = 0·70). Interpretation: Our findings suggest that free-text clinical narratives can be used to learn classification models that allow the rapid identification of patient cohorts. Ultimately, this effort can lead to new insights that can address emerging public health concerns. Digitization of health information will continue to be a reality in both human and veterinary data; our approach serves as first proof-of-concept regarding how these two domains can learn from, and inform, one another. Funding Statement: Stanford University & The Chan Zuckerberg Biohub Investigator Award Declaration of Interests: CDB is Principal and Chairman of CDB Consulting LTD. He has advised Imprimed, Embark Vet and Etalon DX as a member of their respective Scientific Advisory Boards, and is a Director of Etalon DX. The remaining authors declare no conflicts of interest. Ethics Approval Statement: This research was reviewed and approved by Stanford’s Institutional Review Board (IRB), which provided a nonhuman subject determination under eProtocol 46979. Consent was not required.

A text mining analysis of medication quality related event reports from community pharmacies.

Medication errors are estimated to cost $42 billion in annual global treatment costs. Pharmacy-based Patient Safety Organizations (PSO) are tasked with collecting and analyzing incidents, near misses, and unsafe condition reports as one way of engaging pharmacies in quality improvement efforts. Collectively, these reports are referred to as quality related events (QREs). Large-scale analysis of typed narratives from QRE reports across organizations has been a missing component of quality improvement programs. To identify topics within the components of a proposed medication safety event framework contained in the free-text narrative of QRE reports. A retrospective, observational analysis of data from a PSOs voluntary reporting system, from January 1, 2011 to December 31, 2014. The dataset contained structured and unstructured data elements. A structural topic model extracted themes from the free-text narrative component of the report. These topics were assigned a human label and mapped onto constructs of the medication safety event framework. A total of 531,555 QREs were analyzed from 1660 pharmacies. 90.6% were near miss and unsafe condition reports. There were 40 topics generated. There were 29 topics identified as QRE types, 3 were identified as contributing factors, and 5 were related to signals/alerts that an incident or near miss had occurred. One topic each was identified as a recovery step and a quality improvement strategy. One topic was not assigned a human label. Examples of topics labeled included incorrect tapering directions, needing to double-check work, and attention-related contributing factor. The free-text narrative provided novel information compared to the structured fields of the reports. Topics were mapped onto a proposed medication safety event framework to advance knowledge of medication QREs and identify ways to improve medication safety in community pharmacy. Future work may focus on communicating these topics to the pharmacies to improve medication safety efforts.

Statistics students’ identification of inferential model elements within contexts of their own invention

Statistical thinking partially depends upon an iterative process by which essential features of a problem setting are identified and mapped onto an abstract model or archetype, and then translated back into the context of the original problem setting (Wild and Pfannkuch 1999). Assessment in introductory statistics often relies on tasks that present students with data in context and expects them to choose and describe an appropriate model. This study explores post-secondary student responses to an alternative task that prompts students to clearly identify a sample, population, statistic, and parameter using a context of their own invention. The data include free text narrative responses of a random sample of 500 students from a sample of more than 1600 introductory statistics students. Results suggest that students' responses often portrayed sample and population accurately. Portrayals of statistic and parameter were less reliable and were associated with descriptions of a wide variety of other concepts. Responses frequently attributed a variable of some kind to the statistic, or a study design detail to the parameter. Implications for instruction and research are discussed, including a call for emphasis on a modeling paradigm in introductory statistics.

Structured reporting in petrous bone MRI examinations: impact on report completeness and quality

To investigate whether structured reports (SRs) provide benefits regarding the completeness and the clarity of reports, as well as regarding the satisfaction of the referring physicians compared to narrative freetext reports (NRs) of MRI examinations of the petrous bone. After sample size calculation, 32 patients with clinically indicated MRI examinations of the petrous bone were included in this retrospective study. The already existing NRs were taken from the radiologic information system. The corresponding SRs were retrospectively generated by two radiologists using an online-based application. All 64 reports (one NR and one SR per patient) were evaluated by two head and neck physicians using a questionnaire. While 41% of the SRs showed no missing report key feature, all NRs exhibited at least one missing key feature (p < 0.001). SRs achieved significantly higher satisfaction rates regarding the linguistic quality and overall report quality compared to NRs: Using a 6-point Likert scale (1 = insufficient to 6 = excellent), SRs were rated with a median value of 6 [interquartile range (IQR): 1] for linguistic as well as overall quality, and NRs were rated with a median of 5 (IQR: 0) for linguistic as well as overall quality (p < 0.001). Structured reporting of petrous bone MRI examinations may positively influence the completeness and quality of radiologic reports. Due to the easier readability and facilitation of information extraction, SRs improve the satisfaction level of the referring physicians.

Automatic classification of radiological reports for clinical care.

Radiological reporting generates a large amount of free-text clinical narratives, a potentially valuable source of information for improving clinical care and supporting research. The use of automatic techniques to analyze such reports is necessary to make their content effectively available to radiologists in an aggregated form. In this paper we focus on the classification of chest computed tomography reports according to a classification schema proposed for this task by radiologists of the Italian hospital ASST Spedali Civili di Brescia. The proposed system is built exploiting a training data set containing reports annotated by radiologists. Each report is classified according to the schema developed by radiologists and textual evidences are marked in the report. The annotations are then used to train different machine learning based classifiers. We present in this paper a method based on a cascade of classifiers which make use of a set of syntactic and semantic features. The resulting system is a novel hierarchical classification system for the given task, that we have experimentally evaluated.

Effect of a Complementary/Integrative Medicine Treatment Program on Taxane-Induced Peripheral Neuropathy: A Brief Report.

Peripheral neuropathy is a common complication of cancer treatment impairing quality of life and function. This study explored the impact of a complementary and integrative medicine (CIM) program on taxane-induced peripheral neuropathy (TIPN). Taxane-treated female patients with breast and gynecological cancer reporting TIPN-related symptoms were referred to an integrative physician, followed by patient-tailored CIM treatments (acupuncture with/without other modalities). Assessment of study outcomes at 6 to 12 weeks was conducted using the Measure Yourself Concerns and Wellbeing, which documented free-text narratives about patients' experience during the CIM treatment process. Content was analyzed using ATLAS.Ti software. Of the 125 patients treated with taxanes, 69 had been referred for CIM treatment of TIPN-associated symptoms. Multidisciplinary narrative analysis identified 2 groups of CIM-treated patients: those with an apparently moderate improvement in symptoms (n = 35) and those with either only an apparent mild or no improvement at all. For 10 patients, assessment of their response to treatment was unclear. The 2 identified groups had similar demographic, cancer-related, and quality of life-related parameters at baseline. Content analysis of patients with an apparent moderate improvement suggested a short-term (24-48 hours) effect with acupuncture treatment, either alone or combined with manual, mind-body, and anthroposophic music therapies. Symptoms showing improvement included paresthesia and numbness. Acupuncture and other CIM therapies may result in a short-term and transitory reduction in TIPN-related symptoms.

Next generation phenotyping using narrative reports in a rare disease clinical data warehouse

BackgroundSecondary use of data collected in Electronic Health Records opens perspectives for increasing our knowledge of rare diseases. The clinical data warehouse (named Dr. Warehouse) at the Necker-Enfants Malades Children’s Hospital contains data collected during normal care for thousands of patients. Dr. Warehouse is oriented toward the exploration of clinical narratives. In this study, we present our method to find phenotypes associated with diseases of interest.MethodsWe leveraged the frequency and TF-IDF to explore the association between clinical phenotypes and rare diseases. We applied our method in six use cases: phenotypes associated with the Rett, Lowe, Silver Russell, Bardet-Biedl syndromes, DOCK8 deficiency and Activated PI3-kinase Delta Syndrome (APDS). We asked domain experts to evaluate the relevance of the top-50 (for frequency and TF-IDF) phenotypes identified by Dr. Warehouse and computed the average precision and mean average precision.ResultsExperts concluded that between 16 and 39 phenotypes could be considered as relevant in the top-50 phenotypes ranked by descending frequency discovered by Dr. Warehouse (resp. between 11 and 41 for TF-IDF). Average precision ranges from 0.55 to 0.91 for frequency and 0.52 to 0.95 for TF-IDF. Mean average precision was 0.79. Our study suggests that phenotypes identified in clinical narratives stored in Electronic Health Record can provide rare disease specialists with candidate phenotypes that can be used in addition to the literature.ConclusionsClinical Data Warehouses can be used to perform Next Generation Phenotyping, especially in the context of rare diseases. We have developed a method to detect phenotypes associated with a group of patients using medical concepts extracted from free-text clinical narratives.

Providing diabetes services on-site: Bridging the gap between the HMO and employer in a medically underserved area

Medication errors are estimated to cost $42 billion in annual global treatment costs. Pharmacy-based Patient Safety Organizations (PSO) are tasked with collecting and analyzing incidents, near misses, and unsafe condition reports as one way of engaging pharmacies in quality improvement efforts. Collectively, these reports are referred to as quality related events (QREs). Large-scale analysis of typed narratives from QRE reports across organizations has been a missing component of quality improvement programs.To identify topics within the components of a proposed medication safety event framework contained in the free-text narrative of QRE reports.A retrospective, observational analysis of data from a PSOs voluntary reporting system, from January 1, 2011 to December 31, 2014. The dataset contained structured and unstructured data elements. A structural topic model extracted themes from the free-text narrative component of the report. These topics were assigned a human label and mapped onto constructs of the medication safety event framework.A total of 531,555 QREs were analyzed from 1660 pharmacies. 90.6% were near miss and unsafe condition reports. There were 40 topics generated. There were 29 topics identified as QRE types, 3 were identified as contributing factors, and 5 were related to signals/alerts that an incident or near miss had occurred. One topic each was identified as a recovery step and a quality improvement strategy. One topic was not assigned a human label. Examples of topics labeled included incorrect tapering directions, needing to double-check work, and attention-related contributing factor.The free-text narrative provided novel information compared to the structured fields of the reports. Topics were mapped onto a proposed medication safety event framework to advance knowledge of medication QREs and identify ways to improve medication safety in community pharmacy. Future work may focus on communicating these topics to the pharmacies to improve medication safety efforts.

A health technology network approach to quality improvement – Optimizing E-prescribing

Medication errors are estimated to cost $42 billion in annual global treatment costs. Pharmacy-based Patient Safety Organizations (PSO) are tasked with collecting and analyzing incidents, near misses, and unsafe condition reports as one way of engaging pharmacies in quality improvement efforts. Collectively, these reports are referred to as quality related events (QREs). Large-scale analysis of typed narratives from QRE reports across organizations has been a missing component of quality improvement programs.To identify topics within the components of a proposed medication safety event framework contained in the free-text narrative of QRE reports.A retrospective, observational analysis of data from a PSOs voluntary reporting system, from January 1, 2011 to December 31, 2014. The dataset contained structured and unstructured data elements. A structural topic model extracted themes from the free-text narrative component of the report. These topics were assigned a human label and mapped onto constructs of the medication safety event framework.A total of 531,555 QREs were analyzed from 1660 pharmacies. 90.6% were near miss and unsafe condition reports. There were 40 topics generated. There were 29 topics identified as QRE types, 3 were identified as contributing factors, and 5 were related to signals/alerts that an incident or near miss had occurred. One topic each was identified as a recovery step and a quality improvement strategy. One topic was not assigned a human label. Examples of topics labeled included incorrect tapering directions, needing to double-check work, and attention-related contributing factor.The free-text narrative provided novel information compared to the structured fields of the reports. Topics were mapped onto a proposed medication safety event framework to advance knowledge of medication QREs and identify ways to improve medication safety in community pharmacy. Future work may focus on communicating these topics to the pharmacies to improve medication safety efforts.

Assessing Provider-Generated Free-Text Quality in EHR-Integrated Handoff Notes.

Handoff notes are increasingly integrated within electronic health record (EHR) systems and often contain data automatically generated from the EHR and free-text narratives. We examined the quality of data entered by providers in the free-text portion of our institutional EHR handoff tool. Overall, 65% of handoff notes contained at least one error (average 1.7 errors per note). Most errors were omissions in information around patient plan/management or assessment/diagnosis rather than entry of false data. Factors associated with increased error rate were increasing hospital day number; weekend note; medical (vs. surgical) service team; and authorship by a medical student, first or fourth year resident physician, or attending physician. Our findings suggest that errors are common in handoff notes, and while these errors are not completely false data, they may provide individuals caring for patients an inaccurate understanding of patient status.

ADEPt, a semantically-enriched pipeline for extracting adverse drug events from free-text electronic health records.

Adverse drug events (ADEs) are unintended responses to medical treatment. They can greatly affect a patient’s quality of life and present a substantial burden on healthcare. Although Electronic health records (EHRs) document a wealth of information relating to ADEs, they are frequently stored in the unstructured or semi-structured free-text narrative requiring Natural Language Processing (NLP) techniques to mine the relevant information. Here we present a rule-based ADE detection and classification pipeline built and tested on a large Psychiatric corpus comprising 264k patients using the de-identified EHRs of four UK-based psychiatric hospitals. The pipeline uses characteristics specific to Psychiatric EHRs to guide the annotation process, and distinguishes: a) the temporal value associated with the ADE mention (whether it is historical or present), b) the categorical value of the ADE (whether it is assertive, hypothetical, retrospective or a general discussion) and c) the implicit contextual value where the status of the ADE is deduced from surrounding indicators, rather than explicitly stated. We manually created the rulebase in collaboration with clinicians and pharmacists by studying ADE mentions in various types of clinical notes. We evaluated the open-source Adverse Drug Event annotation Pipeline (ADEPt) using 19 ADEs specific to antipsychotics and antidepressants medication. The ADEs chosen vary in severity, regularity and persistence. The average F-measure and accuracy achieved by our tool across all tested ADEs were 0.83 and 0.83 respectively. In addition to annotation power, the ADEPT pipeline presents an improvement to the state of the art context-discerning algorithm, ConText.

Lessons Learned From the VA Radiotherapy Incident Reporting and Analysis System (RIRAS)

The intranet Web-based RIRAS was developed and deployed in the Veterans Health Administration (VHA) System to collect adverse events and good catch data; analyze the causes and contributing factors; and find ways to prevent future occurrences at its 39 radiation oncology practices nationwide. 286 reports were electronically submitted to the RIRAS between 5/11/2014 and 1/02/2017. Each reported event included information on the incident chronology, general information specific to the treatment modality, site and intent and free text narrative. All reports were carefully analyzed by a radiation oncology subject matter expert (SME) and assigned to one of the eight discrete radiation oncology process steps. Steps relate to where the event was discovered and/or occurred. As part of the analysis, error type, medical and dosimetric severity, interventions and potential causes were recorded for each report. Patient safety work products (PSWP) created for these reports listed corrective, preventive and learning actions. For learning purposes, the anonymized version of each PSWP is publicly available on the RIRAS website, which enhances the safety culture. We completed causal analysis and developed corrective action plans for 100% of the reported incidents. Out of a total of 286 reports, 35% were actual events of which 3% met the criteria of VHA misadministration classification but with a low medical severity. 61% were good catches/ near misses, 4% were unsafe conditions. 32% of the reported events occurred in the treatment planning phase, 28% in the treatment delivery phase and 7% in the imaging for treatment verification phase. 31% of the events were discovered by the radiation physicist, 25% by the radiation therapist and 19% by the attending radiation oncologist. 32% of the events were related to organizational management issues like non-existent or inadequate policy and procedures. 31% were related to procedural issues such as distractions / loss of attention and poor documentation. 24% were human behavior related to poor judgement and lack of vigilance. 9% were based on technical issues such as lack of IT / biomedical attention. 36 reports were due to documentation errors, 11 based on previous RT treatment not taken into account for retreatment and 12 due to inconsistent patient setup instructions. Based on the analyses of reported incidents, we were able to make individualized recommendations for radiotherapy process improvement to enhance patient safety at each reporting facility. While RIRAS utilization is a marker of a positive safety culture within VHA, it is important to have a concrete plan for following up, learning and documenting process improvements from the reported events. We found that the ability to learn is significantly enhanced when the reporter is self-identified and there is an ability to request additional information on the event.

Extracting Drug-Drug Interactions with Word and Character-Level Recurrent Neural Networks.

Drug-drug interactions (DDIs) are known to be responsible for nearly a third of all adverse drug reactions. Hence several current efforts focus on extracting signal from EMRs to prioritize DDIs that need further exploration. To this end, being able to extract explicit mentions of DDIs in free text narratives is an important task. In this paper, we explore recurrent neural network (RNN) architectures to detect and classify DDIs from unstructured text using the DDIExtraction dataset from the SemEval 2013 (task 9) shared task. Our methods are in line with those used in other recent deep learning efforts for relation extraction including DDI extraction. However, to our knowledge, we are the first to investigate the potential of character-level RNNs (Char-RNNs) for DDI extraction (and relation extraction in general). Furthermore, we explore a simple but effective model bootstrapping method to (a). build model averaging ensembles, (b). derive confidence intervals around mean micro-F scores (MMF), and (c). assess the average behavior of our methods. Without any rule based filtering of negative examples, a popular heuristic used by most earlier efforts, we achieve an MMF of 69.13. By adding simple replicable heuristics to filter negative instances we are able to achieve an MMF of 70.38. Furthermore, our best ensembles produce micro F-scores of 70.81 (without filtering) and 72.13 (with filtering), which are superior to metrics reported in published results. Although Char-RNNs turnout to be inferior to regular word based RNN models in overall comparisons, we find that ensembling models from both architectures results in nontrivial gains over simply using either alone, indicating that they complement each other.

For the greater good? Patient and public attitudes to use of medical free text data in research

ABSTRACTObjectivesElectronic health records (EHRs) contain rich information for understanding health conditions and their treatment. A large proportion of clinical information in EHRs is stored in narrative free text. This text is currently under-utilised due to privacy concerns, as it is harder to remove patient identifiers from text than from structured data. Automated de-identification of clinical text is now possible using heuristic or machine-learning-based systems. We conducted a review of the literature on patient and public understanding and attitudes towards the use of patients’ medical data for research, particularly seeking views on free text. The aim was to inform and develop a governance framework for the de-identification and use of medical free text for research, and to instigate a wider discussion on the topic. ApproachWe undertook a systematic search in Web of Science and ScienceDirect with terms such as “public attitudes” and “electronic health records”. 3480 results were sifted by title, abstract and full text. Forty-two articles were retained for review, these reported on studies of patient and public perceptions, understanding and attitudes towards the use of patients’ medical data in research. ResultsResearch participants were positively inclined towards information in records being used in research “for the greater good”. However, no clear patterns by age, ethnicity, education level or SES emerged as to who was more favourable to data use. Participants generally trusted health care professionals and public sector researchers with de-identified medical data, whereas government health agencies and commercial entities were not trusted. No explicitly feared harms associated with data use were articulated. However the general objections appeared to be a dislike of personal data being exploited for commercial gain, and a dislike of personal data being moved around and used without personal knowledge or consent. Notably the use of EHR medical text for research did not emerge as a specific patient/public concern. De-identification was important to participants but text was not identified as a distinct privacy risk.ConclusionThis review demonstrates that transparency about data usage, and working “for the greater good” rather than financial gain, appear to be the most important public concerns to be addressed when using patients’ medical data. Governance frameworks for using EHRs must now be enhanced to provide for the use of medical text. This will involve informing both regulators and the public about the current capabilities of automated de-identification, and developing other assurances to safeguard patients’ privacy.

Structured reporting of MRI of the shoulder - improvement of report quality?

To evaluate the effect of structured reports (SRs) in comparison to non-structured narrative free text (NRs) shoulder MRI reports and potential effects of both types of reporting on completeness, readability, linguistic quality and referring surgeons' satisfaction. Thirty patients after trauma or with suspected degenerative changes of the shoulder were included in this study (2012-2015). All patients underwent shoulder MRI for further assessment and possible surgical planning. NRs were generated during clinical routine. Corresponding SRs were created using a dedicated template. All 60 reports were evaluated by two experienced orthopaedic shoulder surgeons using a questionnaire that included eight questions. Eighty per cent of the SRs were fully complete without any missing key features whereas only 45% of the NRs were fully complete (p < 0.001). The extraction of information was regarded to be easy in 92% of the SRs and 63% of the NRs. The overall quality of the SRs was rated better than that of the NRs (p < 0.001). Structured reporting of shoulder MRI improves the readability as well as the linguistic quality of radiological reports, and potentially leads to a higher satisfaction of referring physicians. • Structured MRI reports of the shoulder improve readability. • Structured reporting facilitates information extraction. • Referring physicians prefer structured reports to narrative free text reports. • Structured MRI reports of the shoulder can reduce radiologist re-consultations.

Assessing the completeness of coded and narrative data from the Victorian Emergency Minimum Dataset using injuries sustained during fitness activities as a case study.

BackgroundInjury surveillance systems support the ongoing systematic collection, analysis and interpretation of health information vital to the prevention, planning and evaluation of injury prevention strategies. One key measure of the success of such systems is their reliability. Data completeness is a major component of system reliability, and is an indicator of a system’s data quality. The Victorian Emergency Minimum Dataset (VEMD) is a state-wide record of injury presentations to emergency departments in Victoria, Australia. For each case, it provides information on the injury cause, place of occurrence, activity at time of injury, body region affected and nature of injury, as well as a free-text narrative of the injury event. The aim of this study was to assess the completeness of data in the VEMD using injuries sustained in fitness facilities as a case study.MethodsAnalysis of VEMD coded parent injury variables (nature of injury, injured body region, cause of injury, place where injury occurred, activity at time of injury) and detailed narratives were reviewed for completeness over the ten-year period July 2003 to June 2012, inclusive. Narratives were text analysed manually to determine which items of injury information they contained and compared to the parent injury variables.ResultsThere were 2936 identified cases related to injuries sustained during fitness activities. Two percent of cases had all coded injury variables unspecified. Overall, 95.8 % of narratives had at least one piece of injury information missing. The nature of injury and body region variables were coded in 92.6 and 96.6 % of cases, yet were only mentioned in 27.1 and 75.4 % of narratives, respectively. The cause variable was allocated a specified code in 47.7 % of cases and was mentioned in 45.9 % of narratives. The cause was missing in both in 42.8 % of cases. In approximately half of all cases, the activity and place were specified in both the coded injury variable and narrative; they were missing in both in 7.4 and 13.6 % of cases, respectively.ConclusionsThe reliability of the VEMD as an injury surveillance system, varied depending on the injury variable being examined.

Natural Language Processing in Oncology

Natural language processing (NLP) has the potential to accelerate translation of cancer treatments from the laboratory to the clinic and will be a powerful tool in the era of personalized medicine. This technology can harvest important clinical variables trapped in the free-text narratives within electronic medical records. Natural language processing can be used as a tool for oncological evidence-based research and quality improvement. Oncologists interested in applying NLP for clinical research can play pivotal roles in building NLP systems and, in doing so, contribute to both oncological and clinical NLP research. Herein, we provide an introduction to NLP and its potential applications in oncology, a description of specific tools available, and a review on the state of the current technology with respect to cancer case identification, staging, and outcomes quantification. More automated means of leveraging unstructured data from daily clinical practice is crucial as therapeutic options and access to individual-level health information increase. Research-minded oncologists may push the avenues of evidence-based research by taking advantage of the new technologies available with clinical NLP. As continued progress is made with applying NLP toward oncological research, incremental gains will lead to large impacts, building a cost-effective infrastructure for advancing cancer care.