RESUMEN
Encouraged by the success of pretrained Transformer models in many natural language processing tasks, their use for International Classification of Diseases (ICD) coding tasks is now actively being explored. In this study, we investigated two existing Transformer-based models (PLM-ICD and XR-Transformer) and proposed a novel Transformer-based model (XR-LAT), aiming to address the extreme label set and long text classification challenges that are posed by automated ICD coding tasks. The Transformer-based model PLM-ICD, which currently holds the state-of-the-art (SOTA) performance on the ICD coding benchmark datasets MIMIC-III and MIMIC-II, was selected as our baseline model for further optimisation on both datasets. In addition, we extended the capabilities of the leading model in the general extreme multi-label text classification domain, XR-Transformer, to support longer sequences and trained it on both datasets. Moreover, we proposed a novel model, XR-LAT, which was also trained on both datasets. XR-LAT is a recursively trained model chain on a predefined hierarchical code tree with label-wise attention, knowledge transferring and dynamic negative sampling mechanisms. Our optimised PLM-ICD models, which were trained with longer total and chunk sequence lengths, significantly outperformed the current SOTA PLM-ICD models, and achieved the highest micro-F1 scores of 60.8 % and 50.9 % on MIMIC-III and MIMIC-II, respectively. The XR-Transformer model, although SOTA in the general domain, did not perform well across all metrics. The best XR-LAT based models obtained results that were competitive with the current SOTA PLM-ICD models, including improving the macro-AUC by 2.1 % and 5.1 % on MIMIC-III and MIMIC-II, respectively. Our optimised PLM-ICD models are the new SOTA models for automated ICD coding on both datasets, while our novel XR-LAT models perform competitively with the previous SOTA PLM-ICD models.
Asunto(s)
Clasificación Internacional de Enfermedades , Memoria , Procesamiento de Lenguaje NaturalRESUMEN
BACKGROUND: The narrative free-text data in electronic medical records (EMRs) contain valuable clinical information for analysis and research to inform better patient care. However, the release of free text for secondary use is hindered by concerns surrounding personally identifiable information (PII), as protecting individuals' privacy is paramount. Therefore, it is necessary to deidentify free text to remove PII. Manual deidentification is a time-consuming and labor-intensive process. Numerous automated deidentification approaches and systems have been attempted to overcome this challenge over the past decade. OBJECTIVE: We sought to develop an accurate, web-based system deidentifying free text (DEFT), which can be readily and easily adopted in real-world settings for deidentification of free text in EMRs. The system has several key features including a simple and task-focused web user interface, customized PII types, use of a state-of-the-art deep learning model for tagging PII from free text, preannotation by an interactive learning loop, rapid manual annotation with autosave, support for project management and team collaboration, user access control, and central data storage. METHODS: DEFT comprises frontend and backend modules and communicates with central data storage through a filesystem path access. The frontend web user interface provides end users with a user-friendly workspace for managing and annotating free text. The backend module processes the requests from the frontend and performs relevant persistence operations. DEFT manages the deidentification workflow as a project, which can contain one or more data sets. Customized PII types and user access control can also be configured. The deep learning model is based on a Bidirectional Long Short-Term Memory-Conditional Random Field (BiLSTM-CRF) with RoBERTa as the word embedding layer. The interactive learning loop is further integrated into DEFT to speed up the deidentification process and increase its performance over time. RESULTS: DEFT has many advantages over existing deidentification systems in terms of its support for project management, user access control, data management, and an interactive learning process. Experimental results from DEFT on the 2014 i2b2 data set obtained the highest performance compared to 5 benchmark models in terms of microaverage strict entity-level recall and F1-scores of 0.9563 and 0.9627, respectively. In a real-world use case of deidentifying clinical notes, extracted from 1 referral hospital in Sydney, New South Wales, Australia, DEFT achieved a high microaverage strict entity-level F1-score of 0.9507 on a corpus of 600 annotated clinical notes. Moreover, the manual annotation process with preannotation demonstrated a 43% increase in work efficiency compared to the process without preannotation. CONCLUSIONS: DEFT is designed for health domain researchers and data custodians to easily deidentify free text in EMRs. DEFT supports an interactive learning loop and end users with minimal technical knowledge can perform the deidentification work with only a shallow learning curve.
RESUMEN
Electronic Medical Records (EMRs) contain clinical narrative text that is of great potential value to medical researchers. However, this information is mixed with Personally Identifiable Information (PII) that presents risks to patient and clinician confidentiality. This paper presents an end-to-end de-identification framework to automatically remove PII from Australian hospital discharge summaries. Our corpus included 600 hospital discharge summaries which were extracted from the EMRs of two principal referral hospitals in Sydney, Australia. Our end-to-end de-identification framework consists of three components: (1) Annotation: labelling of PII in the 600 hospital discharge summaries using five pre-defined categories: person, address, date of birth, individual identification number, phone/fax number; (2) Modelling: training six named entity recognition (NER) deep learning base-models on balanced and imbalanced datasets; and evaluating ensembles that combine all six base-models, the three base-models with the best F1 scores and the three base-models with the best recall scores respectively, using token-level majority voting and stacking methods; and (3) De-identification: removing PII from the hospital discharge summaries. Our results showed that the ensemble model combined using the stacking Support Vector Machine (SVM) method on the three base-models with the best F1 scores achieved excellent results with a F1 score of 99.16% on the test set of our corpus. We also evaluated the robustness of our modelling component on the 2014 i2b2 de-identification dataset. Our ensemble model, which uses the token-level majority voting method on all six base-models, achieved the highest F1 score of 96.24% at strict entity matching and the highest F1 score of 98.64% at binary token-level matching compared to two state-of-the-art methods. The end-to-end framework provides a robust solution to de-identifying clinical narrative corpuses safely. It can easily be applied to any kind of clinical narrative documents.
Asunto(s)
Aprendizaje Profundo , Alta del Paciente , Humanos , Australia , Registros Electrónicos de Salud , Hospitales , Procesamiento de Lenguaje NaturalRESUMEN
International Classification of Diseases (ICD) coding plays an important role in systematically classifying morbidity and mortality data. In this study, we propose a hierarchical label-wise attention Transformer model (HiLAT) for the explainable prediction of ICD codes from clinical documents. HiLAT firstly fine-tunes a pretrained Transformer model to represent the tokens of clinical documents. We subsequently employ a two-level hierarchical label-wise attention mechanism that creates label-specific document representations. These representations are in turn used by a feed-forward neural network to predict whether a specific ICD code is assigned to the input clinical document of interest. We evaluate HiLAT using hospital discharge summaries and their corresponding ICD-9 codes from the MIMIC-III database. To investigate the performance of different types of Transformer models, we develop ClinicalplusXLNet, which conducts continual pretraining from XLNet-Base using all the MIMIC-III clinical notes. The experiment results show that the F1 scores of the HiLAT + ClinicalplusXLNet outperform the previous state-of-the-art models for the top-50 most frequent ICD-9 codes from MIMIC-III. Visualisations of attention weights present a potential explainability tool for checking the face validity of ICD code predictions.
Asunto(s)
Clasificación Internacional de Enfermedades , Redes Neurales de la Computación , Codificación Clínica/métodos , Bases de Datos Factuales , Humanos , Alta del Paciente , Reproducibilidad de los ResultadosRESUMEN
BACKGROUND/SIGNIFICANCE: The 10-point Adverse Childhood Experiences (ACEs) score measures childhood exposure to traumatic events. An ACEs score of 4 or higher has been associated with long-term physical and mental health problems, and increased mortality. It is unclear if the rates of undergraduate nursing students experiencing ACEs is different from the general population, and what impact ACEs has on nursing students' depression, anxiety and stress. PURPOSE: The purpose of this study was to assess the rate of ACEs in nursing students and to examine the relationship between ACEs and perceived depression, anxiety, and stress among undergraduate nursing students. METHODS: This is a descriptive correlational study to determine Adverse Childhood Experiences (ACEs) scores and Modified Depression Anxiety and Stress Scale (DASS-21) scores for undergraduate pre-licensure students at a private mid-western college. Students' ACEs scores were correlated with DASS-21 scores and demographic data. RESULTS: 409 students volunteered to participate in the study. Seventeen percent of students had high ACEs scores (≥4). Those students with high ACEs scores also scored statically significantly higher on the DASS-21 items relating to depression, anxiety, and stress. CONCLUSION/IMPLICATIONS: Results from this study demonstrated that many nursing students have experienced ACEs and nursing students with ACEs scores of 4 or higher had higher rates of depression, anxiety, and stress. These findings should drive faculty in nursing programs to acknowledge the prevalence of ACEs among their nursing students and recognize that impact on the mental health of students.
Asunto(s)
Experiencias Adversas de la Infancia , Bachillerato en Enfermería , Estudiantes de Enfermería , Ansiedad/epidemiología , Depresión/epidemiología , HumanosRESUMEN
OBJECTIVE: Data quality (DQ) must be consistently defined in context. The attributes, metadata, and context of longitudinal real-world data (RWD) have not been formalized for quality improvement across the data production and curation life cycle. We sought to complete a literature review on DQ assessment frameworks, indicators and tools for research, public health, service, and quality improvement across the data life cycle. MATERIALS AND METHODS: The review followed PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines. Databases from health, physical and social sciences were used: Cinahl, Embase, Scopus, ProQuest, Emcare, PsycINFO, Compendex, and Inspec. Embase was used instead of PubMed (an interface to search MEDLINE) because it includes all MeSH (Medical Subject Headings) terms used and journals in MEDLINE as well as additional unique journals and conference abstracts. A combined data life cycle and quality framework guided the search of published and gray literature for DQ frameworks, indicators, and tools. At least 2 authors independently identified articles for inclusion and extracted and categorized DQ concepts and constructs. All authors discussed findings iteratively until consensus was reached. RESULTS: The 120 included articles yielded concepts related to contextual (data source, custodian, and user) and technical (interoperability) factors across the data life cycle. Contextual DQ subcategories included relevance, usability, accessibility, timeliness, and trust. Well-tested computable DQ indicators and assessment tools were also found. CONCLUSIONS: A DQ assessment framework that covers intrinsic, technical, and contextual categories across the data life cycle enables assessment and management of RWD repositories to ensure fitness for purpose. Balancing security, privacy, and FAIR principles requires trust and reciprocity, transparent governance, and organizational cultures that value good documentation.
Asunto(s)
Exactitud de los Datos , Mejoramiento de la Calidad , Animales , Estadios del Ciclo de VidaRESUMEN
There has been no empirical evidence about the health informatics workforce in Australia produced in the last ten years. This study reports the findings from an analysis of a subset of the 2018 Australian Health Informatics Workforce Census data. Analysing 420 responses that were identified as the occupational group Health Informatics, the results indicate that whilst most of the workforce is classified as aged (>45 years), many respondents are still relatively early in their health informatics careers. Furthermore, most do not possess any formal education in health informatics and almost a quarter undertake their health informatics role alongside another health-related role. The broad range of position titles and functions demonstrates the breadth within this workforce. Ongoing monitoring of this occupational group is required to inform workforce reform and renewal.
Asunto(s)
Censos , Informática Médica , Australia , Fuerza Laboral en Salud , Recursos HumanosRESUMEN
BACKGROUND: Fluid administration in critically ill surgical patients must be closely monitored to avoid complications. Resuscitation guided by invasive methods are not consistently associated with improved outcomes. As such, there has been increased use of focused ultrasound and Arterial Pulse Waveform Analysis (APWA) to monitor and aid resuscitation. An assessment of these methods using the Grading of Recommendations Assessment, Development, and Evaluation (GRADE) framework is presented. METHODS: A subsection of the Surgical Critical Care Task Force of the Practice Management Guideline Committee of EAST conducted two systematic reviews to address the use of focused ultrasound and APWA in surgical patients being evaluated for shock. Six population, intervention, comparator, and outcome (PICO) questions were generated. Critical outcomes were prediction of fluid responsiveness, reductions in organ failures or complications and mortality. Forest plots were generated for summary data and GRADE methodology was used to assess for quality of the evidence. Reviews are registered in PROSPERO, the International Prospective Register of Systematic Reviews (42015032402 and 42015032530). RESULTS: Twelve focused ultrasound studies and 20 APWA investigations met inclusion criteria. The appropriateness of focused ultrasound or APWA-based protocols to predict fluid responsiveness varied widely by study groups. Results were mixed in the one focused ultrasound study and 9 APWA studies addressing reductions in organ failures or complications. There was no mortality advantage of either modality versus standard care. Quality of the evidence was considered very low to low across all PICO questions. CONCLUSION: Focused ultrasound and APWA compare favorably to standard methods of evaluation but only in specific clinical settings. Therefore, conditional recommendations are made for the use of these modalities in surgical patients being evaluated for shock. LEVEL OF EVIDENCE: Systematic Review, level II.
Asunto(s)
Enfermedad Crítica , Fluidoterapia , Choque Quirúrgico/diagnóstico , Choque Traumático/diagnóstico , Ecocardiografía , Humanos , Guías de Práctica Clínica como Asunto , Análisis de la Onda del Pulso , Resucitación , Choque Quirúrgico/terapia , Choque Traumático/terapiaRESUMEN
BACKGROUND: There is a need to develop a national census of the health information workforce to inform the planning and forecasting of the workforce and inform education and training needs. Yet, this raises a number of issues that need to be incorporated into the planning stage. This paper discusses the issues and recommendations for the development of a national health information workforce census and an Expert Panel's recommendations for how we should proceed. METHODS: Nominated participants from Australia and New Zealand participated as Expert Panel members in focus groups to identify and discuss the issues. Recommendations were identified during the focus groups and documented for participant verification. These were then grouped into themes. RESULTS AND DISCUSSION: The themes of the issues and considerations identified in the focus groups were: Aim of the census; Census advertisement; Census delivery; Participants; Longitudinal study; Types of data elements; Data development and standards; Data ownership, access and governance; Global census; Dissemination of results.
Asunto(s)
Censos , Personal de Salud , Informática Médica , Australia , Grupos Focales , Humanos , Estudios Longitudinales , Nueva ZelandaRESUMEN
The aim of this paper is to summarise common issues and challenges for health information systems (HIS) in Pacific Island Countries and Territories (PICTs) as identified by Pacific participants at two meetings held by the HIS Knowledge Hub in 2009 and provide suggestions for future action. The global agenda and drivers of HIS were discussed at both meetings to provide a clearer understanding of how Pacific Island countries are positioned within the larger international agenda. The two meetings provided the opportunity for participants to highlight suggestions for future action. Many of the solutions proposed highlighted the potential for regional solutions to progress the issue. This suggests an urgent need for national health authorities and regional partners to agree on strategies and programs to derive maximum benefit from regional HIS resources.
Asunto(s)
Desarrollo de Programa/métodos , Informática en Salud Pública/organización & administración , Congresos como Asunto , Islas del PacíficoRESUMEN
BACKGROUND: An intensivist-directed Intensive Care Unit is a closed-model unit in which a physician formally trained in critical care plays a leadership role in patient management. In the last decade, there has been a move toward closed Intensive Care Units. The purpose of this evaluation was to assess the association of changes in the use of intensivists to a closed-model with mortality outcomes in injured patients seen in a long-established urban Level I Trauma Center. METHODS: This analysis used data from the Scottsdale Healthcare Osborn Medical Center trauma registry from January 1, 2002-December 31, 2008. Mortality prior to hospital discharge was compared in the pre-intensivist (intensivists were not employed and did not provide care), partial intensivist (intensivists were employed and provided care during some Intensive Care Unit shifts) and full-time intensivist (intensivists were employed and provided care in the Intensive Care Unit full time) periods. Multiple logistic regression analysis was used to estimate odds ratios for mortality adjusting for patient characteristics and injury severity for the partial intensivist and full-time intensivist periods compared with the pre-intensivist period. RESULTS: Of 18,918 patients, 365 (1.9%) died before hospital discharge. After adjustment for demographic factors and injury severity score, for all patients, odds ratios comparing the partial intensivist and full-intensivist periods with the pre-intensivist period were 0.84 (95% confidence interval 0.64-1.11) and 0.99 (95% confidence interval 0.69-1.41). In patients with an injury severity score 16-24, the adjusted OR for death was 0.20 (95% CI 0.07-0.58) comparing the partial-intensivist with the pre-intensivist period and 0.30 (95% CI 0.11-0.88) comparing the full-time intensivist period with the pre-intensivist period. For patients age 65 + years, compared with the pre-intensivist period, odds ratio were 0.51 (95% confidence interval 0.31-0.84) and 0.61 (95% confidence interval 0.32-1.16) for the partial and full-time intensivist periods respectively. CONCLUSIONS: In our setting, a change to a closed Intensive Care Unit model was associated with improved mortality outcomes in patients with less severe injuries and patients age 65+ years.
RESUMEN
Opportunity is missed by most people because it is dressed in overalls and looks like work. (Thomas Edison).
Asunto(s)
Accidentes de Tránsito/prevención & control , Teléfono Celular , Administración de la Seguridad/organización & administración , Accidentes de Tránsito/legislación & jurisprudencia , Accidentes de Tránsito/psicología , Accidentes de Tránsito/estadística & datos numéricos , Conducción de Automóvil/legislación & jurisprudencia , Conducción de Automóvil/psicología , Teléfono Celular/legislación & jurisprudencia , Teléfono Celular/estadística & datos numéricos , Humanos , Internet , Factores de Riesgo , Asunción de Riesgos , Estados UnidosRESUMEN
To probe the binding of a peptide agonist to a G-protein coupled receptor in native membranes, the spin-labeled amino acid analogue 4-amino-4-carboxy-2,2,6,6-tetramethylpiperidino-1-oxyl (TOAC) was substituted at either position 4 or 9 within the substance P peptide (RPKPQQFFGLM-NH2), a potent agonist of the neurokinin-1 receptor. The affinity of the 4-TOAC analog is comparable to the native peptide while the affinity of the 9-TOAC derivative is approximately 250-fold lower. Both peptides activate receptor signaling, though the potency of the 9-TOAC peptide is substantially lower. The utility of these modified ligands for reporting conformational dynamics during the neurokinin-1 receptor activation was explored using EPR spectroscopy, which can determine the real-time dynamics of the TOAC nitroxides in solution. While the binding of both the 4-TOAC substance P and 9-TOAC substance P peptides to isolated cell membranes containing the neurokinin-1 receptor is detected, a bound signal for the 9-TOAC peptide is only obtained under conditions that maintain the receptor in its high-affinity binding state. In contrast, 4-TOAC substance P binding is observed by solution EPR under both low- and high-affinity receptor states, with evidence of a more strongly immobilized peptide in the presence of GDP. In addition, to better understand the conformational consequences of TOAC substitution into substance P as it relates to receptor binding and activation, atomistic models for both the 4- and 9-TOAC versions of the peptide were constructed, and the molecular dynamics calculated via simulated annealing to explore the influence of the TOAC substitutions on backbone structure.
Asunto(s)
Sustancia P/análogos & derivados , Sustancia P/química , Animales , Células CHO , Simulación por Computador , Cricetinae , Cricetulus , Espectroscopía de Resonancia por Spin del Electrón , Receptores de Neuroquinina-1/efectos de los fármacos , Receptores de Neuroquinina-1/metabolismo , Sustancia P/efectos de los fármacosRESUMEN
Nitric oxide (NO) has a number of physiological and pathophysiological effects in the nervous system. One target of NO is the mitochondrion, where it inhibits respiration and ATP synthesis, which may contribute to NO-mediated neuronal injury. Our recent studies suggested that impaired mitochondrial function impairs mitochondrial trafficking, which could also contribute to neuronal injury. Here, we studied the effects of NO on mitochondrial movement and morphology in primary cultures of forebrain neurons using a mitochondrially targeted enhanced yellow fluorescent protein. NO produced by two NO donors, papa non-oate and diethylamine/NO complex, caused a rapid cessation of mitochondrial movement but did not alter morphology. Movement recovered after removal of NO. The effects of NO on movement were associated with dissipation of the mitochondrial membrane potential. Increasing cGMP levels using 8-bromoguanosine 3',5'-cyclic monophosphate, did not mimic the effects on mitochondrial movement. Furthermore, 1H-[1,2,4]oxadiazolo[4,3-a]quinoxalin-1-one (ODQ), an inhibitor of NO-induced activation of soluble guanylate cyclase, did not block the effects of NO. Thus, neither increasing nor decreasing cGMP levels had an effect on mitochondrial movement. Based on these data, we conclude that NO is a novel modulator of mitochondrial trafficking in neurons, which may act through the inhibition of mitochondrial function.
Asunto(s)
Potenciales de la Membrana/efectos de los fármacos , Mitocondrias/efectos de los fármacos , Neuronas/efectos de los fármacos , Neuronas/ultraestructura , Óxido Nítrico/farmacología , Prosencéfalo/citología , Animales , Células Cultivadas , GMP Cíclico/análogos & derivados , GMP Cíclico/farmacología , Relación Dosis-Respuesta a Droga , Interacciones Farmacológicas , Embrión de Mamíferos , Inhibidores Enzimáticos/farmacología , Ácido Glutámico/farmacología , Hidrazinas/farmacología , Movimiento/efectos de los fármacos , Donantes de Óxido Nítrico/farmacología , Óxidos de Nitrógeno/farmacología , Oxadiazoles/farmacología , Quinoxalinas/farmacología , Ratas , Factores de TiempoRESUMEN
To probe the molecular nature of the binding pocket of a G protein-coupled receptor and the events immediately following the binding and activation, we have modified the substance P peptide, a potent agonist for the neurokinin-1 receptor, with a nitroxide spin probe specifically attached at Lys-3. The agonist properties and binding affinity of the spin-labeled substance P are similar to the native peptide. Using electron paramagnetic resonance (EPR) spectroscopy, the substance P analogue is capable of reporting the microenvironment found in the binding pocket of the receptor. The EPR spectrum of bound peptide indicates that the Lys-3 portion of the agonist is highly flexible. In addition, we detect a slight increase in the mobility of the bound peptide in the presence of a non-hydrolyzable analogue of GTP, indicative of the alternate conformational states described for this class of receptor. The down-regulation of neurokinin-tachykinin receptors is accomplished by a rapid internalization of the activated protein. Thus, it was also of interest to establish whether spin-labeled substance P could serve as a real time reporter for endocytosis. Our findings show the receptor agonist is efficiently endocytosed and the loss of EPR signal upon internalization provides a real time monitor of endocytosis. The rapid loss of signal suggests that endosomal trafficking vesicles maintain a reductive environment. Whereas the reductive capacity of the lysosome has been established, our findings indicate this capacity in early endosomes as well.