No abstract

Chest X-rays are the most economically viable diagnostic imaging test for active pulmonary tuberculosis screening despite the high sensitivity and low specificity when interpreted by clinicians or radiologists. Computer aided detection (CAD) algorithms, especially convolution based deep learning architecture, have been proposed to facilitate the automation of radiography imaging modalities. Deep learning algorithms have found success in classifying various abnormalities in lung using chest X-ray. We fine-tuned, validated and tested EfficientNetB4 architecture and utilized the transfer learning methodology for multilabel approach to detect lung zone wise and image wise manifestations of active pulmonary tuberculosis using chest X-ray. We used Area Under Receiver Operating Characteristic (AUC), sensitivity and specificity along with 95% confidence interval as model evaluation metrics. We also utilized the visualisation capabilities of convolutional neural networks (CNN), Gradient-weighted Class Activation Mapping (Grad-CAM) as post-hoc attention method to investigate the model and visualisation of Tuberculosis abnormalities and discuss them from radiological perspectives. EfficientNetB4 trained network achieved remarkable AUC, sensitivity and specificity of various pulmonary tuberculosis manifestations in intramural test set and external test set from different geographical region. The grad-CAM visualisations and their ability to localize the abnormalities can aid the clinicians at primary care settings for screening and triaging of tuberculosis where resources are constrained or overburdened.

Accessibility determines health care utilization among individuals with noncommunicable diseases as they need to visit health facilities frequently. Hence, we aimed to assess the road distance and travel time to the diabetes clinic of Persons with Diabetes (PWDs) seeking care at a public tertiary care facility in South India. PWDs house locations were geocoded using ArcGIS World Geocoding Services, and ArcGIS Pro Business Analyst Geoprocessing extension was used to conduct network analysis. A simple median regression analysis was done to compare the association of sociodemographic variables with distance and time. Of the total 2261 PWDs included, the mean (SD) age was 53.7 (11.5) years, and 49.4% were males and about 66.0% of the PWDs resided in rural areas. The median (IQR) travel distance of PWDs from their home to the diabetes clinic was 30.5 (7.6-78.5) km and the median (IQR) time spent in travelling was 77.9 (16.4-194.7) minutes. About 76% travelled more than 5 km to the diabetes clinic. About 85% of PWDs travelled farther than the nearest available public health facility to avail care from the diabetes clinic. Younger age group, male gender, PWDs from rural areas and the state of Tamil Nadu travelled significantly longer distance compared to their counterparts. To conclude, about three-fourth of the PWDs travelled more than 5 km for care at the diabetes clinic. Also, about 9 out of 10 travelled farther than the nearest available public health facility where diabetes care was available.

No abstract

Background: Hypertension is a global public health issue. Geographic information systems (GIS) are increasingly being used by health-care systems as an emerging tool to address the public health burden of hypertension. Objective: The objective of the study is to describe the geographic distribution of adults with known hypertension residing in the urban field practice area of a tertiary care institution and to assess the factors associated with its control status. Materials and Methods: We conducted a cross-sectional analytical study in an urban health center (UHC) with adults with hypertension (n = 343) seeking care from the NCD clinic of UHC and private clinics were included. Geo-coding was done (n = 343) using digital GPS device by house-to-house visit and average of the three blood pressure recordings using digital sphygmomanometer taken for assessing control status (n = 277) of hypertension. A structured questionnaire was used to collect sociodemographic, risk factors distribution, and medication adherence. Geospatial analysis was done using QGIS 3.0, ArcGIS 10.2 and SPSS version 22 (IBM Corp. Armonk, NY, USA) was used for statistical analysis. Results: The geographic distribution showed clusters and hotspots in the study area. Of the 277 study participants, 57.4% (51.6–63.5) had blood pressure under control and 41% were male. Patients with age ≥60 years (prevalence ratios [PR]: 1.2, 95% CI: 1–1.6), with no comorbidity (PR: 1.3, 95% CI: 1–1.7), high medicine adherence (PR: 7.6, 95% CI: 3.9–14.6) were independent factors associated with control status. Conclusion: The study identified the clustering and hotspot areas of known patients with hypertension. Around three-fifth of known hypertensives had their blood pressure under control.