Artificial neural networks (ANNs) are powerful tools for data analysis and are particularly suitable for modeling relationships between variables for best prediction of an outcome. While these models can be used to answer many important research questions, their utility has been critically limited because the interpretation of the "black box" model is difficult. Clinical investigators usually employ ANN models to predict the clinical outcomes or to make a diagnosis; the model however is difficult to interpret for clinicians. To address this important shortcoming of neural network modeling methods, we describe several methods to help subject-matter audiences (e.g., clinicians, medical policy makers) understand neural network models. Garson's algorithm describes the relative magnitude of the importance of a descriptor (predictor) in its connection with outcome variables by dissecting the model weights. The Lek's profile method explores the relationship of the outcome variable and a predictor of interest, while holding other predictors at constant values (e.g., minimum, 20th quartile, maximum). While Lek's profile was developed specifically for neural networks, partial dependence plot is a more generic version that visualize the relationship between an outcome and one or two predictors. Finally, the local interpretable model-agnostic explanations (LIME) method can show the predictions of any classification or regression, by approximating it locally with an interpretable model. R code for the implementations of these methods is shown by using example data fitted with a standard, feed-forward neural network model. We offer codes and step-by-step description on how to use these tools to facilitate better understanding of ANN.
BackgroundThe diagnosis of diabetes has been described as a lifelong psychological burden on the patient and his or her family. Social support plays a pivotal role in patients with diabetes and is important in enabling them to cope effectively with the disease. There is a dearth of research on social support and coping in patients with diabetes in South Africa.ObjectivesThe aim of this study was to explore whether patients with poor perceived social support have lower levels of well-being and coping than patients with good social support.MethodA cross-sectional study was conducted at both public and private facilities on the north coast of KwaZulu-Natal, South Africa. The Diabetes Care Profile (DCP), the General Health Questionnaire (GHQ), the Social Support Questionnaire (SSQ) and the WHO-5 Well-being Index (WHO-5) were administered to 401 participants.ResultsThe findings indicate that there is an inverse relationship between social support and coping, which suggests that an increase in social support is associated with a decrease in emotional distress. There was a small positive correlation between the SSQ and the WHO-5, which suggests participants who had good support endorsed better levels of well-being. Although participants indicated that they were satisfied with their level of support, they had poor coping as indicated by the high mean score on the GHQ and the high HbA1c level. There was a small positive correlation between GHQ and HbA1c. There was no association between social support and HbA1c.ConclusionSocial support is important in helping the patient with diabetes cope with the disease and to improve adherence to treatment. Health care providers should take cognisance of psychosocial factors in the treatment regime of the patient. Family members should be educated about diabetes, the importance of adherence and long-term complications of the disease.
Background Discordant genotypic/phenotypic rifampicin susceptibility testing in Mycobacterium tuberculosis is a significant challenge, yet there are limited data on its prevalence and how best to manage such patients. Whether to treat isolates with rpoB mutations not conferring phenotypic resistance as susceptible or multidrug-resistant tuberculosis (MDR-TB) is unknown. We describe phenotypic and genotypic characteristics of discordant isolates and clinical characteristics and treatment outcomes of affected patients in KwaZulu-Natal, South Africa. Methods We analyzed clinical isolates showing rifampicin resistance on GenoType MTBDR plus while susceptible on 1% agar proportion method. We measured rifampicin minimum inhibitory concentrations (MICs) using Middlebrook 7H10 agar dilution and BACTEC MGIT 960. Sensititre MYCOTB plates were used for drug-susceptibility testing, and rpoB gene sequencing was performed on all isolates. Local MDR-TB program data were reviewed for clinical information and patient outcomes. Results Discordant isolates constituted 4.6% (60) of 1302 rifampicin-resistant cases over the study period. Of these, 62% remained susceptible to isoniazid and 98% remained susceptible to rifabutin. Rifampicin MICs were close to the critical concentration of 1 µg/mL (0.5–2 µg/mL) for 83% of isolates. The most frequent rpoB mutations were Q513P (25.3%), D516V (19.2%), and D516Y (13.3%). Whereas 70% were human immunodeficiency virus infected, the mean CD4 count was 289 cells/mm 3 and 87% were receiving antiretroviral therapy. Standard therapy for MDR-TB was used and 53% achieved successful treatment outcomes. Conclusions Rifampicin-discordant TB is not uncommon and sequencing is required to confirm results. The high susceptibility to rifabutin and isoniazid and poor treatment outcomes with the current regimen suggest a potential utility for rifabutin-based therapy.
BackgroundAcute respiratory tract infections contribute significantly to morbidity and mortality among young children in resource-poor countries. However, studies on the viral aetiology of acute respiratory infections, seasonality and the relative contributions of comorbidities such as immune deficiency states to viral respiratory tract infections in children in these countries are limited.MethodsA retrospective analysis of laboratory test results of upper or lower respiratory specimens of children between 0 and 5 years of age collected between 1st January 2011 and 31st July 2015 from hospitals in KwaZulu-Natal, South Africa. Respiratory specimens were tested for viral respiratory pathogens using multiplex polymerase chain reaction (PCR), HIV testing was performed either by serological or PCR methods. Cytomegalovirus (CMV) respiratory infection was determined using the CMV R-gene PCR kit.ResultsIn total 2172 specimens were analysed, of which 1175 (54.1%) were from males. The median age was 3.0 months (interquartile range [IQR] 1–7). Samples from the lower respiratory tract accounted for 1949 (89.7%) of all specimens. Respiratory multiplex PCR results were positive in 834 (45.7%) specimens. Respiratory syncytial virus (RSV) was the most commonly detected virus in 316 (32.1%) patients, followed by adenovirus (ADV) in 215 (21.8%), human rhinovirus (Hrhino) in 152 (15.4%) and influenza A (FluA) in 50 (5.1%). A seasonal time series pattern was observed for ADV (winter peak), enterovirus (EV) (autumn), human bocavirus (HBoV) (summer), and parainfluenza viruses 1 and 3 (PIV1 and 3) (spring). Stationary or untrended seasonal variation was observed for FluA (winter peak) and RSV (summer). HIV results were available for 1475 (67.9%) specimens; of these 348 (23.6%) were positive. CMV results were available for 714 (32.9%) specimens, of which 416 (58.3%) were positive. There was a statistically significant association between the coinfection of HIV and CMV with ADV.ConclusionsIn this study, we identified the most common respiratory viral pathogens detected among hospitalized children in KwaZulu-Natal. The coinfection between HIV and CMV was found to be associated with an increased risk of only adenovirus infection. Most viral pathogens showed a seasonal trend of occurrence. Our data has implications for the rational design of public health programmes.
Background: Sustained injudicious and indiscriminate use of antimicrobials has exerted selection pressure for developing antimicrobial resistance (AMR), requiring behaviour change from healthcare professionals (HCPs) based on their knowledge, attitudes and practices (KAP) on antimicrobials, AMR and antimicrobial stewardship (AMS). Methods:A cross-sectional online questionnaire-based survey was conducted nationally amongst doctors, pharmacists and nurses from November 2017 to January 2018. The questionnaire comprised demographic information and KAP questions.Results: Respondents comprised of 1120 doctors, 744 pharmacists and 659 nurses. Antimicrobial resistance was considered a severe problem globally and nationally by majority of HCPs. Self-assessment of knowledge revealed gaps in understanding of antimicrobials, AMR and AMS. Confidence scores in prescribing by doctors, pharmacists and nurses were 57.82%, 32.88% and 45.28%, respectively. Doctors, 441 (45.2%) indicated no confidence in using combination therapy. Prescribing correctly showed a confidence level of 33.99% from 436 doctors, 41.88% from nine pharmacists and 35.23% from 107 nurses. Healthcare professionals ( 1600[91.22%]) stated educational campaigns would combat AMR. Only 842 (40.13%) HCPs attended training on these topics and 1712 (81.60%) requesting more education and training. Conclusion:This is the first comparative survey on KAP of practising doctors, pharmacists and nurses in South Africa. Doctors had the highest knowledge score followed by nurses and pharmacists. Practice scores did not corroborate knowledge and the higher attitude scores. Gaps in KAP were evident. Healthcare professionals indicated the need for more education and training, thus requiring a review of pre-service and in-service education and training in addition to continued professional development programmes for practising HCPs.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.