Topic Modeling mediante Machine Learning no supervisado de artículos científicos sobre Salud Laboral y Servicios de Atención de Salud a Domicilio / Topic Modeling through unsupervised Machine Learning of scientific articles on Occupational Health and Home Care Services

Objetivo: Identificar de manera no supervisada mediante topic modeling los temas de mayor in-terés en el campo de la Salud Laboral y los Ser-vicios de Atención a Domicilio de los artículos científicos publicados en la materia.Método: Este estudio empleó el algoritmo de Machine Learning no supervisado Asignación Latente de Dirichlet para el topic modeling y el lexicón NRC para la realización del análisis de sentimientos del corpus de las fichas docu-mentales obtenidas de MEDLINE (vía PubMed) usando los descriptores “Salud Laboral” y “Ser-vicios de Atención de Salud a Domicilio”.Resultados: Del total de 70 fichas documenta-les analizadas, se obtuvo que la intensidad de las emociones en los textos era baja (oscilando en valores de 5 a 10), teniendo una mayor re-presentación los sentimientos positivos frente a los negativos en una relación de 60/40. No hubo una variación de las proporciones de las emo-ciones con respecto al período del estudio. Se identificaron los cuatro temas de mayor interés en los artículos analizados: cuidado domicilia-rio y satisfacción de los cuidadores, período de lactancia, programas de rehabilitación, y activi-dad física para mitigación del dolor.Conclusiones: Se ha podido constatar que las metodologías del procesado de lenguaje natu-ral pueden ser una gran herramienta de apoyo al análisis de artículos científicos. Concretamen-te, se ha logrado determinar de manera clara y no supervisada los temas de mayor interés en el campo de la Salud Laboral y la Atención de Salud a Domicilio. (AU)

Objective: To identify in an unsupervised man-ner through topic modeling the topics of great-est interest in the field of Occupational Health and Home Care Services from the scientific arti-cles published on the subject.Method: The study used the unsupervised Ma-chine Learning algorithm Dirichlet Latent Assign-ment for topic modeling and the NRC lexicon to carry out the sentiment analysis of the corpus of document files obtained from MEDLINE (via PubMed) using the descriptors “Occupational Health” and “Home Care Services”.Results: Of the total of 70 documentary files analyzed, it was obtained that the intensity of the emotions in the texts was low (ranging in val-ues from 5 to 10), with positive feelings having a greater representation compared to negative ones in a ratio of 60/ 40. There was no variation in the proportions of emotions with respect to the study period. The four topics of greatest interest were identified in the articles analyzed: home care and caregiver satisfaction, breastfeeding period, rehabilitation programs, and physical activity to mitigate pain.Conclusions: It has been confirmed that natural language processing methodologies can be a great support tool for the analysis of scientific articles. Specifically, it has been possible to de-termine in a clear and unsupervised manner the topics of greatest interest in the field of Occupa-tional Health and Home Care Services. (AU)

Occupational Health Applied Infodemiological Studies of Nutritional Diseases and Disorders: Scoping Review with Meta-Analysis.

(1) Objective: to identify and review existing infodemiological studies on nutritional disorders applied to occupational health and to analyse the effect of the intervention on body mass index (BMI) or alternatively body weight (BW); (2) Methods: This study involved a critical analysis of articles retrieved from MEDLINE (via PubMed), Embase, Cochrane Library, PsycINFO, Scopus, Web of Science, Latin American, and Caribbean Health Sciences Literature (LILACS) and Medicina en Español (MEDES) using the descriptors "Nutrition Disorders, "Occupational Health" and "Infodemiology", applying the filters "Humans" and "Adult: 19+ years". The search was conducted on 29 May 2021; (3) Results: a total of 357 references were identified from the bibliographic database searches; after applying the inclusion and exclusion criteria, a total of 11 valid studies were obtained for the review. Interventions could be categorised into (1) interventions related to lifestyle, physical activity, and dietary changes through education programmes, (2) interventions associated with lifestyle, physical activity, and dietary changes through the use of telemonitoring systems or self-help applications, (3) interventions tied to lifestyle, physical activity, and dietary changes through control and/or social network support groups, and (4) interventions linked to changes in the work environment, including behavioural change training and work environment training tasks. The meta-analysis demonstrated that the heterogeneity present when analysing the results for BMI was 72% (p < 0.01), which decreased to 0% (p = 0.57) when analysing the outcomes for weight, in which case the null hypothesis of homogeneity could be accepted. In all instances, the final summary of the effect was on the decreasing side for both BMI and BW; (4) Conclusions: Despite the high heterogeneity of the results reported, the trend shown in all cases indicates that the intervention methodologies implemented by empowering individuals through Web 2.0 technologies are positive in terms of the problem of overweight. Further implementation of novel strategies to support individuals is needed to overcome obesity, and, at least in the early studies, these strategies seem to be making the necessary change.

Identification of high-risk patients for referral through machine learning assisting the decision making to manage minor ailments in community pharmacies.

Background: Data analysis techniques such as machine learning have been used for assisting in triage and the diagnosis of health problems. Nevertheless, it has not been used yet to assist community pharmacists with services such as the Minor Ailment Services These services have been implemented to reduce the burden of primary care consultations in general medical practitioners (GPs) and to allow a better utilization of community pharmacists' skills. However, there is a need to refer high-risk patients to GPs. Aim: To develop a predictive model for high-risk patients that need referral assisting community pharmacists' triage through a minor ailment service. Method: An ongoing pragmatic type 3 effectiveness-implementation hybrid study was undertaken at a national level in Spanish community pharmacies since October 2020. Pharmacists recruited patients presenting with minor ailments and followed them 10 days after the consultation. The main outcome measured was appropriate medical referral (in accordance with previously co-designed protocols). Nine machine learning models were tested (three statistical, three black box and three tree models) to assist pharmacists in the detection of high-risk individuals in need of referral. Results: Over 14'000 patients were included in the study. Most patients were female (68.1%). With no previous treatment for the specific minor ailment (68.0%) presented. A percentage of patients had referral criteria (13.8%) however, not all of these patients were referred by the pharmacist to the GP (8.5%). The pharmacists were using their clinical expertise not to refer these patients. The primary prediction model was the radial support vector machine (RSVM) with an accuracy of 0.934 (CI95 = [0.926,0.942]), Cohen's kappa of 0.630, recall equal to 0.975 and an area under the curve of 0.897. Twenty variables (out of 61 evaluated) were included in the model. radial support vector machine could predict 95.2% of the true negatives and 74.8% of the true positives. When evaluating the performance for the 25 patient's profiles most frequent in the study, the model was considered appropriate for 56% of them. Conclusion: A RSVM model was obtained to assist in the differentiation of patients that can be managed in community pharmacy from those who are at risk and should be evaluated by GPs. This tool potentially increases patients' safety by increasing pharmacists' ability to differentiate minor ailments from other medical conditions.

Tendencias temporales de los patrones de búsqueda sobre Servicios de Atención de Salud a Domicilio antes y después del COVID-19 / Temporal trends in Home Care Services search patterns before and after COVID-19

Objetivo: Analizar las tendencias temporales de los patrones de búsqueda de información, sobre los Servicios de Atención de Salud a Domicilio a través de Google Trends antes y después de la aparición del COVID-19. Método: Estudio ecológico y correlacional. Los datos se obtuvieron de la consulta directa en la herramienta «Google Trends». Término de búsqueda: “Servicio de asistencia sanitaria domiciliaria”. Fecha de consulta 01-09-2021. Resultados: El máximo de búsquedas se alcanzó en enero de 2020, coincidiendo con el inicio de la pandemia del COVID-19. Durante el período pre-COVID se observó una estacionalidad en el interés de la población (ADF: - 0.49; *p > 0.05) que desapareció con la irrupción del COVID-19 (ADF: -8.55; p < 0.05). La comparación de las medianas mostró diferencias estadísticamente significativas antes y después del COVID (KW: 31.15; *** p-valor < 0.001). Conclusiones: Se ha demostrado que la aparición del COVID-19 ha supuesto un hito significativo respecto al interés de la población general sobre los Servicios de Atención de Salud a Domicilio.(AU)

Objetive: To analyze the temporal trends of the information search patterns on Home Care Services through Google Trends before and after the appearance of COVID-19. Method: Ecological and correlational study. The data were obtained from direct queries in the «Google Trends» tool. Search term: “Home Care Services.” Date of consultation 01-09-2021. Results: The maximum number of searches was reached in January 2020, coinciding with the start of the COVID-19 pandemic. During the pre-COVID period, a seasonality was observed in the interest of the population (ADF: - 0.49; * p> 0.05) that disappeared with the emergence of COVID-19 (ADF: -8.55; p < 0.05). The compari-son of the medians showed statistically significant differences before and after COVID (KW: 31.15; *** p-value < 0.001). Conclusions: It has been shown that the appearance of COVID-19 has been a significant milestone regarding the general population’s interest in Home Care Services.(AU)

Trends and Seasonality of Information Searches Carried Out through Google on Nutrition and Healthy Diet in Relation to Occupational Health: Infodemiological Study.

This study aimed to analyze and relate the population interest through information search trends on Nutrition and Healthy Diet (HD) with the Occupational Health (OH). Ecological and correlational study of the Relative Search Volume (RSV) obtained from Google Trends query, segmented in two searched periods concerning antiquity; date of query: 20 April 2021. The RSV trends for the analyzed three Topics were: Nutrition (R2 = 0.02), HD (R2 = 0.07) and OH (R2 = -0.72). There was a good positive correlation between Nutrition and OH (R = 0.56, p < 0.001) and a moderate one between HD and OH (R = 0.32, p < 0.001). According to seasons, differences were verified between RSV means in the Topics HD (p < 0.01) and OH (p < 0.001). Temporal dependence was demonstrated on Nutrition searches (Augmented Dickey-Fuller = -2.35, p > 0.05). There was only a significant relationship between the RSV Topic HD (p < 0.05) for the Developing and Least Developed countries. The data on the analyzed RSV demonstrated diminishing interest in the search information on HD and OH as well as a clearly positive trend change in recent years for Nutrition. A good positive correlation was observed between the RSV of nutrition and OH whereas the correlation between HD and OH was moderate. There were no milestones found that may report a punctual event leading to the improvement of information searches. Temporal dependence was corroborated in the RSV on Nutrition, but not in the other two Topics. Strangely, only an association was found on HD searches between the Developing and Least Developed Countries. The study of information search trends may provide useful information on the population's interest in the disease data, as well as would gradually allow the analysis of differences in popularity, or interest even between different countries. Thus, this information might be used as a guide for public health approaches regarding nutrition and a healthy diet at work.

Perception of risk of exposure in the management of hazardous drugs in home hospitalization and hospital units.

OBJECTIVE: To assess the perception of risk of exposure in the management of hazardous drugs (HDs) through home hospitalization and hospital units. MATERIAL AND METHODS: A questionnaire was released, at the national level, to health professionals with HD management expertise. Questionnaire included 21 questions that were scored using a Likert scale: 0 (null probability) to 4 (very high probability). The internal consistency and reliability of the questionnaire were calculated using Cronbach's alpha and the intraclass correlation coefficient, respectively. RESULTS: 144 questionnaires (response rate 70.2%) were obtained: 65 (45.1%) were nurses, 42 (28.9%) occupational physicians, and 37 (26.1%) were pharmacists. Cronbach's alpha was 0.93, and intraclass correlation coefficient was 0.94 (95% CI 0.91-0.97; p-value < 0.001). The mean probability was 1.95 ± 1.02 (median 1.9; minimum: 0.05; 1st quartile 1.1; 3rd quartile 2.6; and maximum 4). Differences were observed in scoring among professional groups (occupational physicians versus nurses (1.6/2.1, p = 0.044); pharmacists versus nurses (1.7/2.1, p = 0.05); and occupational physicians versus pharmacists (1.6/1.7, p = 0.785), due mainly to the administration stage (p = 0.015). CONCLUSIONS: The perception of risk of exposure was moderate, being higher for nurses. It would be advisable to integrate HDs into a standardized management system (risk management model applicable to any healthcare center) to improve the safety of health professionals.