Pulmonary abnormality screening on chest x-rays from different machine specifications: a generalized AI-based image manipulation pipeline.

BACKGROUND: Chest x-ray is commonly used for pulmonary abnormality screening. However, since the image characteristics of x-rays highly depend on the machine specifications, an artificial intelligence (AI) model developed for specific equipment usually fails when clinically applied to various machines. To overcome this problem, we propose an image manipulation pipeline. METHODS: A total of 15,010 chest x-rays from systems with different generators/detectors were retrospectively collected from five institutions from May 2020 to February 2021. We developed an AI model to classify pulmonary abnormalities using x-rays from a single system. Then, we externally tested its performance on chest x-rays from various machine specifications. We compared the area under the receiver operating characteristics curve (AUC) of AI models developed using conventional image processing pipelines (histogram equalization [HE], contrast-limited histogram equalization [CLAHE], and unsharp masking [UM] with common data augmentations) with that of the proposed manipulation pipeline (XM-pipeline). RESULTS: The XM-pipeline model showed the highest performance for all the datasets of different machine specifications, such as chest x-rays acquired from a computed radiography system (n = 356, AUC 0.944 for XM-pipeline versus 0.917 for HE, 0.705 for CLAHE, 0.544 for UM, p [Formula: see text] 0.001, for all) and from a mobile x-ray generator (n = 204, AUC 0.949 for XM-pipeline versus 0.933 for HE, p = 0.042, 0.932 for CLAHE (p = 0.009), 0.925 for UM (p = 0.001). CONCLUSIONS: Applying the XM-pipeline to AI training increased the diagnostic performance of the AI model on the chest x-rays of different machine configurations. RELEVANCE STATEMENT: The proposed training pipeline would successfully promote a wide application of the AI model for abnormality screening when chest x-rays are acquired using various x-ray machines. KEY POINTS: • AI models developed using x-rays of a specific machine suffer from generalization. • We proposed a new image processing pipeline to address the generalization problem. • AI models were tested using multicenter external x-ray datasets of various machines. • AI with our pipeline achieved the highest diagnostic performance than conventional methods.

Mobile Health and Nutrition Team Service Implementation in Somali and Afar Regions of Ethiopia: A Qualitative Implementation Science Study.

Background: Ethiopia has been implementing Mobile Health and Nutrition Teams (MHNTs) to improve the accessibility of essential healthcare services for unreached populations with poor infrastructure to ensure health for all target by Universal health coverage (UHC). However, the current implementation status of this MNHT is not assessed. Objective: This study aimed to capture the current implementation status of the MHNT from the program managers, supporting partners and decision makers at each level of the health system structure in Afar and Somali regions. Methods: We conducted qualitative study with phenomenological study design. The data was collected from RHB MHNT coordinator, woreda health office MHNT coordinator, MHNT leader and representatives from implementing partners. The interview guides were developed using the CFIR framework. Results: Out of the 17 respondents, 13 responded all the standard service packages a MHNT is expected to deliver (76.5%). Overall, the KIIs mentioned that the MHNTs are effective in ensuring access and quality of health services. MHNT strategy has high demand and acceptability by the community and the service providers. The main barrier to program implementation is the gap in service integration within and across sectors. Inadequate staffing of the MHNT, gaps in ensuring proper professional mix, frequent turnover of contract health workers, and skill gaps hamper effective and sustainable implementation of the program. Conclusion: MHNT establishment, effectiveness, acceptability and sustainably in the implementing woredas of Afar and Somali Regions is very promising. The culture of MHNT documentation and reporting needs some improvement. Besides, community engagement and government ownership are good drivers for sustainability of MHNT. Standardizing and adding additional professionals with capacity building is crucial to ensuring service quality. Furthermore, community mobilization and woreda leadership commitment boosting will be needed for granting sustainability. Finally, national scale up of this alternative strategy is recommended through standardized implementation modality.

Physicians' Attitude towards Electronic Medical Record Systems: An Input for Future Implementers.

BACKGROUND: Electronic medical record (EMR) systems offer the potential to improve health care quality by allowing physicians real-time access to patient healthcare information. The endorsement and usage of EMRs by physicians have a significant influence on other user groups in the healthcare system. As a result, the purpose of this study was to examine physicians' attitudes regarding EMRs and identify the elements that may influence their attitudes. METHOD: An institutional-based cross-sectional study design supplemented with a qualitative study was conducted from March 1 to April 30, 2018, among a total of 403 physicians. A self-administered questionnaire was used to collect quantitative data. The validity of the prediction bounds for the dependent variable and the validity of the confidence intervals and P values for the parameters were measured with a value of less than 0.05 and 95 percent of confidence interval. For the supplementary qualitative study, data were collected using semistructured in-depth interviews from 11 key informants, and the data were analyzed using thematic analysis. RESULT: Physicians' computer literacy (CI: 0.264, 0.713; P: 0001) and computer access at work (CI: 0.141, 0.533, P: 0.001) were shown to be favorable predictors of their attitude towards EMR system adoption. Another conclusion from this study was the inverse relationship between physicians' prior EMR experience and their attitude about the system (CI: -0.517, -0.121; P: 0.002). CONCLUSION: According to the findings of this study, physicians' attitudes regarding EMR were found moderate in the studied region. There was a favorable relationship between computer ownership, computer literacy, lack of EMR experience, participation in EMR training, and attitude towards EMR. Improving the aforementioned elements is critical to improving physicians' attitudes regarding EMR.

The impact of being of the female gender for household head on the prevalence of food insecurity in Ethiopia: a systematic-review and meta-analysis.

BACKGROUND: Ethiopia signed both for Millennium Development Goals (MDGs) previously and Sustainable Development Goals (SDGs) currently to improve food security through gender equality and empowerment of women by positioning them as household leader. However, there is no concrete evidence about the impact of being of the female gender for household head on the prevalence of food insecurity at the national level, the authors' intention being to fill this gap. METHODS: Preferred Reporting Items for Systematic Reviews and Meta-Analyses protocol (PRISMA-P) guideline was followed. All major databases such as PubMed/MEDLINE, WHOLIS, Cochrane Library, Embase, PsycINFO, ScienceDirect, Web of science, and reference lists were used to identify published articles, whereas shelves, author contact, Google, and Google Scholar were also searched to identify unpublished studies. Joanna Briggs Institute Meta-Analysis of Statistical Assessment and Review Instrument (JBI-MAStARI) was used for critical appraisal of studies. Meta-analysis was conducted using the STATA software version 14. The random effect model was used to estimate the pooled prevalence of food insecurity at 95% confidence level, while subgroup analysis and meta-regression were employed to identify the possible source of heterogeneity and the associated factors respectively. Moreover, Begg's test was used to check publication bias. RESULTS: A total of 143 articles were identified, of which 15 studies were included in the final model with a total sample size of 2084 female-headed households. The pooled estimate of food insecurity among female-headed households was 66.11% (95% confidence level (CL) 54.61, 77.60). Female-headed households had 1.94 (95% CL 1.26, 3.01) times the odds of developing food insecurity as compared with male-headed households in Ethiopia. However, considerable heterogeneity across studies was also exhibited (I 2 = 92.5%, p value < 0.001). CONCLUSION: This review found that severity of food insecurity among female-headed households in Ethiopia was a more pronounced issue as compared with the general national estimate of food insecurity. Food insecurity among them was two-fold increased as compared with their men counterparts.So that, the government of Ethiopia needs to outlook how cultural and social restriction of women's involvement in every aspect of activity affects their level of household food security. Beyond this, previous success and current gap of food insecurity among female-headed households should be explored in future research to run in accordance with Sustainable Development Goals (SDGs) specially with goals 2 and 5.

Prevalence of anemia and its associated factors in human immuno deficiency virus infected adult individuals in Ethiopia. A systematic review and meta-analysis.

BACKGROUND: Anemia is a common hematologic disorder among human Immunodeficiency virus (HIV) infected adult Individuals. However, there is no concrete scientific evidence established at national level in Ethiopia. Hence, this review gave special emphasis on Ethiopian HIV infected adult individuals to estimate pooled prevalence of anemia and its associated factors at national level. METHODS: Studies were retrieved through search engines in PUBMED/Medline, Cochrane Library, and the web of science, Google and Google scholar following the Preferred Reporting Items for Systematic Review and Meta-Analysis Protocols (PRISMA-P). Joanna Briggs Institute Meta-Analysis of Statistical Assessment and Review Instrument (JBI-MAStARI) was used for critical appraisal of the included studies. Random effects meta-analysis was used to estimate the pooled prevalence of anemia and associated factors at 95% Confidence interval with its respective odds ratio (OR). Meta regression was also carried out to identify the factors. Moreover, Sub-group analysis, begs and egger test followed by trim-and-fill analysis were employed to assess heterogeneity and publication bias respectively. RESULT: A total of 532 articles were identified through searching of which 20 studies were included in the final review with a total sample size of 8079 HIV infected adult individuals. The pooled prevalence of anemia was 31.00% (95% CI: 23.94, 38.02). Cluster of Differentiation 4 (CD4) count <= 200 cells/µl with OR = 3.01 (95% CI: 1.87, 4.84), World Health Organization (WHO) clinical stage III&IV with OR = 2.5 (95% CI: 1.29, 4.84), opportunistic infections (OIs) with OR = 1.76 (95% CI: 1.07, 2.89) and body mass index (BMI) < 18.5 kg/M2 with OR = 1.55 ((95% CI: 1. 28, 1.88) were the associated factors. CONCLUSION: This review demonstrates high prevalence of anemia among HIV infected adults. Low CD4 count, WHO clinical stage III&IV, OIs and low level of BMI were found to have significant association with the occurrence of anemia. Therefore, the responsible stockholders including anti retro viral treatment (ART) clinics should strengthen the system and procedures for the early diagnosis of opportunistic infection and screening of underlying problems. There should be also early screening for OIs and under nutrition with strict and frequent monitoring of HIV infected individuals CD4 count.

A Comparison of Electronic Medical Record Data to Paper Records in Antiretroviral Therapy Clinic in Ethiopia: What is affecting the Quality of the Data?

BACKGROUND: Anti-Retroviral Therapy (ART) care is a lifelong treatment, which needs accurate and reliable data collected for long period of time. Poor quality of medical records data remains a challenge and is directly related to the quality of care of patients. To improve this, there is an increasing trend to implement electronic medical record (EMR) in hospitals. However, there is little evidence on the impact of EMR on the quality of health data in low- resource setting hospitals like Ethiopia. This comparative study aims to fill this evidence gap by assessing the completeness and reliability of paper-based and electronic medical records and explore the challenges of ensuring data quality at the Anti-Retroviral Therapy (ART) clinic at the University of Gondar Referral Hospital in Northwest Ethiopia. METHODS: An institution-based comparative cross-sectional study, supplemented with a qualitative approach was conducted from February 1 to March 30, 2017 at the ART clinic of the University of Gondar Hospital. A total of 250 medical records having both electronic and paper-based versions were collected and assessed. A national ART registration form which consists of 40 ART data elements was used as a checklist to assess completeness and reliability dimensions of data quality on medical records of patients on HIV care. Kappa statistics were computed to describe the level of data agreement between paper-based and electronic records across patient characteristics. In-depth interviews were conducted using semi-structured questionnaires with ten key informants to explore the challenges related with the quality of medical records. Responses of the key informant interviews were analyzed using thematic analysis. RESULTS: The overall completeness of medical records was 78% with 95% CI (70.8% - 85.1%) in paper-based and 76% with 95%CI (67.8% - 83.2%) EMR. The data reliability measured in Kappa statistics shows strong agreements on the socio-demographic data such as educational status 0.93 (0.891, 0.963), WHO staging 0.86 (0.808, 0.906); general appearance 0.83 (0.755, 0.892) and patient referral record 0.87 (0.795, 0.932). The major challenges hindering good data quality was the current side by side dual data documentation practice (the need to document both on the paper and the EMR for a single record), patient overload and low data documentation practice of health workers. CONCLUSION: The overall completeness of ART medical records was still slightly better in paper-based records than EMR. The main reason affecting the EMR data quality was the current dual documentation practice both on the paper and electronic for each patient in the hospital and the high load of patients in the clinic. The hospital management need to decide to use either the paper or the electronic system and build the capacity of health workers to improve data quality in the hospital.

How can the use of data within the immunisation programme be increased in order to improve data quality and ensure greater accountability in the health system? A protocol for implementation science study.

BACKGROUND: Immunisation remains one of the most important and cost-effective interventions to reduce vaccine-preventable child morbidity, disability and mortality. Health programmes like the Expanded Program of Immunization rely on complex decision-making and strong local level evidence is important to effectively and efficiently utilise limited resources. Lack of data use for decision-making at each level of the health system remains the main challenge in most developing countries. While there is much evidence on data quality and how to improve it, there is a lack of sufficient evidence on why the use of data for decision-making at each level of the health system is low. Herein, we describe a comprehensive implementation science study that will be conducted to identify organisational, technical and individual level factors affecting local data use at each level of the Ethiopian health system. METHODS: We will apply a mixed methods approach using key informant interviews and document reviews. The qualitative data will be gathered through key informant interviews using a semi-structured guide with open- and closed-ended questions with four categories of respondents, namely decision-makers, data producers, data users and community representatives at the federal, regional, zonal, woreda and community levels of the health system. The document review will be conducted on selected reports and feedback documented at different levels of the health system. Data will be collected from July 2017 to March 2018. Descriptive statistics will be analysed for the quantitative study using SPSS version 20 software and thematic content analysis will be performed for the qualitative part using NVivo software. DISCUSSION: Appropriate and timely use of health and health-related information for decision-making is an essential element in the process of transforming the health sector. The findings of the study will inform stakeholders at different levels on the institutionalisation of evidence-based practice in immunisation programmes.