Your browser doesn't support javascript.
Mostrar: 20 | 50 | 100
Resultados 1 - 20 de 13.544
Filtrar
4.
Neural Netw ; 121: 132-139, 2020 Jan.
Artículo en Inglés | MEDLINE | ID: mdl-31541881

RESUMEN

Neural networks (NNs) have become the state of the art in many machine learning applications, such as image, sound (LeCun et al., 2015) and natural language processing (Young et al., 2017; Linggard et al., 2012). However, the success of NNs remains dependent on the availability of large labelled datasets, such as in the case of electronic health records (EHRs). With scarce data, NNs are unlikely to be able to extract this hidden information with practical accuracy. In this study, we develop an approach that solves these problems for named entity recognition, obtaining 94.6 F1 score in I2B2 2009 Medical Extraction Challenge (Uzuner et al., 2010), 4.3 above the architecture that won the competition. To achieve this, we bootstrap our NN models through transfer learning by pretraining word embeddings on a secondary task performed on a large pool of unannotated EHRs and using the output embeddings as a foundation of a range of NN architectures. Beyond the official I2B2 challenge, we further achieve 82.4 F1 on extracting relationships between medical terms using attention-based seq2seq models bootstrapped in the same manner.


Asunto(s)
Registros Electrónicos de Salud/clasificación , Aprendizaje Automático/clasificación , Procesamiento de Lenguaje Natural , Recolección de Datos/clasificación , Recolección de Datos/métodos , Humanos
5.
Accid Anal Prev ; 135: 105392, 2020 Feb.
Artículo en Inglés | MEDLINE | ID: mdl-31841865

RESUMEN

Crash Detection is essential in providing timely information to traffic management centers and the public to reduce its adverse effects. Prediction of crash risk is vital for avoiding secondary crashes and safeguarding highway traffic. For many years, researchers have explored several techniques for early and precise detection of crashes to aid in traffic incident management. With recent advancements in data collection techniques, abundant real-time traffic data is available for use. Big data infrastructure and machine learning algorithms can utilize this data to provide suitable solutions for the highway traffic safety system. This paper explores the feasibility of using deep learning models to detect crash occurrence and predict crash risk. Volume, Speed and Sensor Occupancy data collected from roadside radar sensors along Interstate 235 in Des Moines, IA is used for this study. This real-world traffic data is used to design feature set for the deep learning models for crash detection and crash risk prediction. The results show that a deep model has better crash detection performance and similar crash prediction performance than state of the art shallow models. Additionally, a sensitivity analysis was conducted for crash risk prediction using data 1-minute, 5-minutes and 10-minutes prior to crash occurrence. It was observed that is hard to predict the crash risk of a traffic condition, 10 min prior to a crash.


Asunto(s)
Accidentes de Tránsito/prevención & control , Recolección de Datos/métodos , Aprendizaje Profundo , Accidentes de Tránsito/estadística & datos numéricos , Humanos , Modelos Logísticos , Medición de Riesgo/métodos
6.
Accid Anal Prev ; 134: 105319, 2020 Jan.
Artículo en Inglés | MEDLINE | ID: mdl-31706186

RESUMEN

The continuous motorization of traffic has led to a sustained increase in the global number of road related fatalities and injuries. To counter this, governments are focusing on enforcing safe and law-abiding behavior in traffic. However, especially in developing countries where the motorcycle is the main form of transportation, there is a lack of comprehensive data on the safety-critical behavioral metric of motorcycle helmet use. This lack of data prohibits targeted enforcement and education campaigns which are crucial for injury prevention. Hence, we have developed an algorithm for the automated registration of motorcycle helmet usage from video data, using a deep learning approach. Based on 91,000 annotated frames of video data, collected at multiple observation sites in 7 cities across the country of Myanmar, we trained our algorithm to detect active motorcycles, the number and position of riders on the motorcycle, as well as their helmet use. An analysis of the algorithm's accuracy on an annotated test data set, and a comparison to available human-registered helmet use data reveals a high accuracy of our approach. Our algorithm registers motorcycle helmet use rates with an accuracy of -4.4% and +2.1% in comparison to a human observer, with minimal training for individual observation sites. Without observation site specific training, the accuracy of helmet use detection decreases slightly, depending on a number of factors. Our approach can be implemented in existing roadside traffic surveillance infrastructure and can facilitate targeted data-driven injury prevention campaigns with real-time speed. Implications of the proposed method, as well as measures that can further improve detection accuracy are discussed.


Asunto(s)
Aprendizaje Profundo , Dispositivos de Protección de la Cabeza/estadística & datos numéricos , Motocicletas/estadística & datos numéricos , Accidentes de Tránsito/estadística & datos numéricos , Adulto , Ciudades , Recolección de Datos/métodos , Humanos , Masculino , Mianmar
7.
Rev Bras Epidemiol ; 22Suppl 3(Suppl 3): e190010.supl.3, 2019.
Artículo en Inglés | MEDLINE | ID: mdl-31800849

RESUMEN

INTRODUCTION: Brazil presented a high proportion of ill-defined causes of death (IDCD) in 2000, compromising accurate cause-of-death analysis. OBJECTIVE: To analyze specific underlying causes for deaths originally assigned as IDCD in the Mortality Information System (SIM - Sistema de Informação sobre Mortalidade), after investigation activities implemented in country between 2006 and 2017. METHOD: For all IDCD identified in the SIM, municipal health professionals collected information about the final disease obtained from hospital records, autopsies, forms of family health teams, and home investigation. Specific causes among reclassified IDCD after investigation were evaluated according to age groups and four calendar periods. RESULTS: Proportions of IDCD reassigned to other causes after review increased over time, reaching 30.1% in 2017. From a total of 257,367 IDCD reclassified in 2006-2017, neonatal-related conditions, injury, ischemic heart disease and stroke were the leading causes detected in the age groups 0-9 years, 10-29 years, 30-69 years, 70 years and over, respectively. DISCUSSION: The similarity and plausibility of cause-specific proportions derived from the reclassification of IDCD by age group over time indicate the accuracy of the investigation data. CONCLUSION: High proportions of IDCD reassigned to more informative causes after review indicate the success of this approach to correct misclassification in the SIM, an initiative that should be maintained. Training physicians on death certification along with better quality of medical care and access to health services would lead to further improvement.


Asunto(s)
Causas de Muerte , Recolección de Datos/métodos , Sistemas de Información/estadística & datos numéricos , Adolescente , Adulto , Distribución por Edad , Factores de Edad , Anciano , Autopsia/estadística & datos numéricos , Brasil/epidemiología , Niño , Preescolar , Certificado de Defunción , Femenino , Humanos , Lactante , Entrevistas como Asunto/estadística & datos numéricos , Masculino , Registros Médicos/estadística & datos numéricos , Persona de Mediana Edad , Factores de Tiempo , Adulto Joven
9.
Poult Sci ; 98(11): 5234-5240, 2019 Nov 01.
Artículo en Inglés | MEDLINE | ID: mdl-31581757

RESUMEN

Assessing keel bone damage reliably and accurately is a requirement for all research on this topic. Most commonly, assessment is done on live birds by palpation and is therefore prone to bias. A 2-day Training School of the COST Action "Identifying causes and solutions of keel bone damage in laying hens" with 16 participants of variable experience was held where palpation of live hens was followed by consulting corresponding radiographic images of keel bones. We hypothesized that the inter-observer and intra-observer repeatabilities as well as the agreement between palpation and assessment from the radiograph (considered as the accuracy) would increase from day 1 to 2. Repeatability estimates were calculated using the R-package rptR and the change in level of accuracy on day 1 and 2 was analyzed with generalized linear models. As predicted, the inter-observer repeatabilities of the assessments of the fractures and deviations were improved by training, but this improvement differed for fractures and deviations between the cranial, middle, and caudal parts of the keel bone. Intra-observer repeatabilities before training also differed between the different parts of the keel bone and were highest for fractures at the caudal part of the keel bone. The training affected the accuracy of palpation to different degrees for the different parts of the keel bone. A training effect was found for the caudal part of the keel bone in regard to fractures and deviations, but for fractures the training effect was missing for the cranial part and for deviations it was missing for the middle part of the keel bone. In conclusion, the training school involving radiographs improved inter-observer repeatabilities in the diagnosis of fractures and deviations of keel bones and thus had the potential to lead to more comparable results among research groups.


Asunto(s)
Crianza de Animales Domésticos/métodos , Pollos , Recolección de Datos/métodos , Fracturas Óseas/veterinaria , Palpación/veterinaria , Enfermedades de las Aves de Corral/diagnóstico , Radiografía/veterinaria , Esternón/patología , Animales , Femenino , Fracturas Óseas/diagnóstico , Variaciones Dependientes del Observador
10.
Tidsskr Nor Laegeforen ; 139(15)2019 10 22.
Artículo en Noruego | MEDLINE | ID: mdl-31642631

RESUMEN

BACKGROUND: The aim of the public health survey in the Norwegian counties is to obtain information that is useful for public health work. In 2018, two parallel data collection processes were undertaken in Hordaland county. Both samples were drawn randomly from the National Population Register, but one of these was limited to users of the helsenorge.no website. The purpose of this article is to investigate the degree to which limiting users to the helsenorge.no website leads to selection bias beyond the selection that occurs through ordinary non-participation. MATERIAL AND METHOD: Services for Sensitive Data (TSD) was used in the data collection for the sample drawn from the National Population Register (n = 36 000), and the helsenorge.no platform was used in the data collection for the sample limited to users of helsenorge.no (n = 30 000). The response rate was 40.8 % and 41.5 %, respectively. RESULTS: For some outcome measures, the differences between the two datasets were modest (gender distribution, age, education and health habits). For variables that were more directly related to health, the differences were greater. In the helsenorge.no sample a higher proportion reported generally poorer health (29.4 vs. 24.0 %), mental health problems (13.6 vs. 11.6 %), disability pension (10.5 vs. 7.8 %) and long-term illness (13.3 vs. 9.3 %). Analyses of subgroups showed more pronounced differences in the proportion with generally poorer health and mental health problems between those with low education in the helsenorge.no sample and the corresponding group in the sample from the National Population Register. INTERPRETATION: Systematic and pronounced differences between the samples show that limiting recruitment to users of helsenorge.no's services results in further selection problems.


Asunto(s)
Encuestas Epidemiológicas , Selección de Paciente , Adolescente , Adulto , Anciano , Enfermedad Crónica/epidemiología , Recolección de Datos/métodos , Escolaridad , Femenino , Estado de Salud , Encuestas Epidemiológicas/métodos , Encuestas Epidemiológicas/normas , Humanos , Masculino , Salud Mental/estadística & datos numéricos , Persona de Mediana Edad , Noruega/epidemiología , Satisfacción Personal , Salud Pública , Sistema de Registros , Sesgo de Selección , Autoinforme , Seguridad Social/estadística & datos numéricos , Apoyo Social , Uso de Tabaco/epidemiología , Adulto Joven
11.
Rev Bras Epidemiol ; 22Suppl 02(Suppl 02): E190004.SUPL.2, 2019.
Artículo en Portugués, Inglés | MEDLINE | ID: mdl-31596375

RESUMEN

INTRODUCTION: This article aims at describing the National Health Survey (Pesquisa Nacional de Saúde- PNS) methodology of collecting laboratory exams data. METHODOLOGY: A subsample of 25% of the census tracts was selected, according to the stratification of the PNS sample, with a probability inversely proportional to the difficulty of collection. The collection of blood and urine was done in the households by a laboratory agent, among residents selected for individual interview. Due to the difficulties found in the field work, the sample did not reach the minimum expected number in some strata, and a post-stratification procedure was proposed for the data analysis. RESULTS: The collection of biospecimens was performed in 8,952 individuals. Laboratory tests were: glycated hemoglobin; total cholesterol; LDL cholesterol; HDL cholesterol; serology for dengue; red blood cell count (erythrogram) and white series count (leukogram); high performance liquid chromatography (HPLC) for diagnosis of hemoglobinopathies; creatinine. Theexcretion of potassium, salt and sodium and creatinine was estimated in the urine. The database of laboratory exams was weighed and made publicly available on the Oswaldo Cruz Foundation's PNS website and can be accessed without prior authorization. CONCLUSION: The total subsample of laboratory exams is of great value, since it allowed us to establish national reference parameters adequate to sociodemographic and geographic characteristics of the Brazilian population, providing relevant and complementary information for the analysis of the health situation of Brazil.


Asunto(s)
Técnicas de Laboratorio Clínico/métodos , Recolección de Datos/métodos , Bases de Datos Factuales , Encuestas Epidemiológicas/métodos , Adolescente , Adulto , Recolección de Muestras de Sangre/métodos , Brasil , Colesterol/sangre , Cromatografía Líquida de Alta Presión , Dengue/sangre , Recuento de Eritrocitos , Femenino , Humanos , Recuento de Leucocitos , Masculino , Persona de Mediana Edad , Toma de Muestras de Orina/métodos , Adulto Joven
12.
World J Gastroenterol ; 25(37): 5619-5629, 2019 Oct 07.
Artículo en Inglés | MEDLINE | ID: mdl-31602162

RESUMEN

BACKGROUND: Korean National Health Insurance (NHI) claims database provides large-cohort. However, studies regarding accuracy of administrative database for pancreatic cancer (PC) have not been reported. We aimed to identify accuracy of NHI database regarding PC classified by international classification of disease (ICD)-10 codes. AIM: To identify the accuracy and usefulness of administrative database in PC and the accurate ICD codes for PC with location. METHODS: Study and control groups were collected from 2003 to 2016 at Seoul National University Bundang Hospital. Cases of PC were identified in NHI database by international classification of diseases, 10th revision edition (ICD-10 codes) supported with V codes. V code is issued by medical doctors for covering 95% of medical cost by Korean government. According to pathologic reports, definite or possible diagnoses were defined using medical records, images, and pathology. RESULTS: A total of 1846 cases with PC and controls were collected. Among PC, only 410 (22.2%) cases were identified as specific cancer sites including head in 234 (12.7%) cases, tail in 104 (5.6%) cases and body in 72 (3.9%) cases. Among PC, 910 (49.3%) cases were diagnosed by definite criteria. Most of these were adenocarcinoma (98.0%). The rates of definite diagnosis of PC were highest in head (70.1%) followed by body (47.2%) and tail (43.3%). False-positive cases were pancreatic cystic neoplasm and metastasis to the pancreas. In terms of the overall diagnosis of PC, sensitivity, specificity, positive predictive value, and negative predictive value were 99.95%, 98.72%, 98.70%, and 99.95%, respectively. Diagnostic accuracy was similar both in terms of diagnostic criteria and tumor locations. CONCLUSION: Korean NHI claims database collected according to ICD-10 code with V code for PC showed good accuracy.


Asunto(s)
Recolección de Datos/métodos , Bases de Datos Factuales/estadística & datos numéricos , Neoplasias Pancreáticas/diagnóstico , Anciano , Recolección de Datos/estadística & datos numéricos , Sistemas de Administración de Bases de Datos/estadística & datos numéricos , Registros Electrónicos de Salud/estadística & datos numéricos , Femenino , Hospitales Universitarios/estadística & datos numéricos , Humanos , Clasificación Internacional de Enfermedades , Masculino , Persona de Mediana Edad , Páncreas/patología , Neoplasias Pancreáticas/epidemiología , Neoplasias Pancreáticas/patología , República de Corea/epidemiología , Estudios Retrospectivos , Sensibilidad y Especificidad
13.
Prof Inferm ; 72(2): 152-161, 2019.
Artículo en Inglés | MEDLINE | ID: mdl-31550432

RESUMEN

In this first part of the article, we aim to provide health researchers with an understanding of how to design a qualitative health research study, including: topic identification, design selection, and engagement in reflexivity. We offer practical guidance for writing an overarching question using a novel framework that helps develop a clearly articulated question that includes the following elements: emphasis, purposeful sampling, phenomenon of interest, and context. We then expand on specific methodological issues: study designs, contexts, sampling, and recruitment. Finally, we provide examples of qualitative health research studies that illustrate the application of different research designs. In part two of this article, we will discuss how to prepare for going into the field, how to generate, manage and analyse data, and plan for the dissemination of qualitative health research.


Asunto(s)
Investigación sobre Servicios de Salud/organización & administración , Investigación Cualitativa , Proyectos de Investigación , Recolección de Datos/métodos , Humanos
14.
BMC Res Notes ; 12(1): 592, 2019 Sep 18.
Artículo en Inglés | MEDLINE | ID: mdl-31533806

RESUMEN

OBJECTIVE: To investigate medications prescribed for pregnant women and their potential teratogenicity risk in Kemisse General Hospital. RESULT: A total of 263 medical records of pregnant women were reviewed, of which 234 pregnant women were prescribed with a total of 430 prescription drugs. The average numbers of drugs per pregnant women was found to be 1.84. Most pregnant women 166 (63.2%) were in the third trimester and more than half of them (51.3%) were multigravida. The maximum number of drugs were prescribed in the second trimester 162 (37.67%) followed by third trimester 143 (33.26%). Supplemental drugs were the most widely used medications 297 (69.07%) and followed by 82 (19.1%) drugs from category B; 54 (12.6%) drugs from category C; and the rest 7 (1.6%) drugs from category D. There was no any drug from category X. Moreover, approximately one third of the pregnant women encountered with drugs from category B, C and D. However, there were no FDA category C and D drugs prescribed in first trimester.


Asunto(s)
Prescripciones de Medicamentos/estadística & datos numéricos , Hospitales Generales , Atención Prenatal/estadística & datos numéricos , Medicamentos bajo Prescripción/administración & dosificación , Adolescente , Adulto , Recolección de Datos/métodos , Recolección de Datos/estadística & datos numéricos , Prescripciones de Medicamentos/normas , Etiopía , Femenino , Humanos , Persona de Mediana Edad , Embarazo , Medicamentos bajo Prescripción/efectos adversos , Medición de Riesgo/métodos , Medición de Riesgo/estadística & datos numéricos , Factores de Riesgo , Teratogénesis/efectos de los fármacos , Adulto Joven
15.
Nat Commun ; 10(1): 4389, 2019 09 26.
Artículo en Inglés | MEDLINE | ID: mdl-31558713

RESUMEN

Do clashes between ideologies reflect policy differences or something more fundamental? The present research suggests they reflect core psychological differences such that liberals express compassion toward less structured and more encompassing entities (i.e., universalism), whereas conservatives express compassion toward more well-defined and less encompassing entities (i.e., parochialism). Here we report seven studies illustrating universalist versus parochial differences in compassion. Studies 1a-1c show that liberals, relative to conservatives, express greater moral concern toward friends relative to family, and the world relative to the nation. Studies 2a-2b demonstrate these universalist versus parochial preferences extend toward simple shapes depicted as proxies for loose versus tight social circles. Using stimuli devoid of political relevance demonstrates that the universalist-parochialist distinction does not simply reflect differing policy preferences. Studies 3a-3b indicate these universalist versus parochial tendencies extend to humans versus nonhumans more generally, demonstrating the breadth of these psychological differences.


Asunto(s)
Actitud , Cultura , Recolección de Datos/métodos , Empatía , Principios Morales , Política , Recolección de Datos/estadística & datos numéricos , Familia/psicología , Femenino , Amigos/psicología , Humanos , Masculino , Encuestas y Cuestionarios/estadística & datos numéricos
17.
Artículo en Inglés | MEDLINE | ID: mdl-31470544

RESUMEN

Citizen science is a growing method of scientific discovery and community engagement. To date, there is a paucity of data using citizen scientists to monitor community level physical activity, such as bicycling or walking; these data are critical to inform community level intervention. Volunteers were recruited from the local community to make observations at five local greenways. The volunteers picked their location, time to collect data and duration of data collection. Volunteer observations included recording estimated age, race or ethnicity and activity level of each individual they encountered walking, running or bicycling on the greenway. A total of 102 volunteers were recruited to participate in the study, of which 60% completed one or more observations. Average observational time lasted 81 minutes and resulted in recording the demographics and physical activity of a mean of 48 people per session. The majority of adult bicyclists observed were biking at a moderate pace (86%) and were white (72%) males (62%). Similar results were observed for those walking. We demonstrate the feasibility of using citizen scientists to address the current scarcity of data describing community-level physical activity behavior patterns. Future work should focus on refining the citizen science approach for the collection of physical activity data to inform community-specific interventions in order to increase greenway use.


Asunto(s)
Planificación Ambiental , Ejercicio , Adulto , Grupos de Población Continentales , Recolección de Datos/métodos , Femenino , Humanos , Masculino , North Carolina , Recreación , Proyectos de Investigación , Factores Sexuales , Voluntarios
18.
Sensors (Basel) ; 19(17)2019 Aug 29.
Artículo en Inglés | MEDLINE | ID: mdl-31470554

RESUMEN

Integrating speech recondition technology into an electronic health record (EHR) has been studied in recent years. However, the full adoption of the system still faces challenges such as handling speech errors, transforming raw data into an understandable format and controlling the transition from one field to the next field with speech commands. To reduce errors, cost, and documentation time, we propose a dialogue system care record (DSCR) based on a smartphone for nursing documentation. We describe the effects of DSCR on (1) documentation speed, (2) document accuracy and (3) user satisfaction. We tested the application with 12 participants to examine the usability and feasibility of DSCR. The evaluation shows that DSCR can collect data efficiently by achieving 96% of documentation accuracy. Average documentation speed was increased by 15% (P = 0.012) compared to traditional electronic forms (e-forms). The participants' average satisfaction rating was 4.8 using DSCR compared to 3.6 using e-forms on a scale of 1-5 (P = 0.032).


Asunto(s)
Recolección de Datos/métodos , Registros Electrónicos de Salud , Lenguaje , Software de Reconocimiento del Habla , Interfaz Usuario-Computador
19.
Gac Med Mex ; 155(4): 399-405, 2019.
Artículo en Inglés | MEDLINE | ID: mdl-31486784

RESUMEN

Research designs refer to the way information is obtained and are limited by ethical, economic and temporal viability. Research designs are standardized strategies to reduce biases, which in the architectural model of research are identified in the baseline state, the maneuver and the outcome; hence, there are no specific designs for each question. The design with the lowest probability of bias is the clinical trial, followed by cohort and case-control studies and, finally, by cross-sectional surveys. Among the main characteristics that give merit to research designs are the following: population inquiry, which refers to the situation of the population in relation to the clinical course/natural history of the disease; the maneuver, or action that is expected to modify the baseline state, which can be observational or experimental; follow-up, or documented monitoring that is given to each subject, which can be longitudinal or cross-sectional; and directionality, which can prolective or retrolective and refers to the timing of data collection for research purposes. It will always be better having a valuable question, even when answered with a design with higher risk of bias, than a question that is irrelevant or has no applicability.


Asunto(s)
Ensayos Clínicos como Asunto/métodos , Estudios Epidemiológicos , Proyectos de Investigación , Sesgo , Ensayos Clínicos como Asunto/normas , Estudios de Cohortes , Exactitud de los Datos , Recolección de Datos/métodos , Humanos
20.
Accid Anal Prev ; 132: 105286, 2019 Nov.
Artículo en Inglés | MEDLINE | ID: mdl-31487665

RESUMEN

Safety performance functions (SPFs) are generally used to relate exposure to the expected number of crashes aggregated over a long time (e.g. a year) by holding all other risk factors constant, and to identify hotspots that have excessive crashes regardless of different time periods. However, it is highly likely that the relationships of exposure, risk factors and crash occurrence can vary across different times of day. This study aims to establish time-dependent SPFs for urban roads by using large-scale dangerous driving event data captured by smartphones in different times of day. Multivariate conditional autoregressive (MVCAR) models are developed to jointly account for spatial and temporal dependence of crash observations. Results of two-sample Kolmogorov-Smirnov tests affirm the heterogeneity of the safety effects of dangerous driving events in different time periods. Time-dependent hotspots are identified using potential for safety improvement (PSI) metric. The assumption here is that due to the change of traffic conditions and environment across different times of day, safety hotspots for different time periods should be different from each other. According to the results of Wilcoxon signed-rank tests, hotspots identified by times of day are found to be mostly different from each other. The findings of this study provide insights into temporal effects of risk factors and can support the development of time-dependent safety countermeasures. Besides, this study also shows the potential of leveraging anonymized and aggregated dangerous driving data to assess traffic safety issues.


Asunto(s)
Accidentes de Tránsito/estadística & datos numéricos , Conducción de Automóvil/estadística & datos numéricos , Recolección de Datos/métodos , Teléfono Inteligente , Accidentes de Tránsito/prevención & control , Entorno Construido , Humanos , Factores de Riesgo , Seguridad
SELECCIÓN DE REFERENCIAS
DETALLE DE LA BÚSQUEDA