Detection of COVID-19 Disease from Chest X-Ray Images: A Deep Transfer Learning Framework

World economy as well as public health have been facing a devastating effect caused by the disease termed as Coronavirus (COVID-19). A significant step of COVID-19 affected patients treatment is the faster and accurate detection of the disease which is the motivation of this study. In this paper, implementation of a deep transfer learning-based framework using a pre-trained network (ResNet-50) for detecting COVID-19 from the chest X-rays was done. Our dataset consists of 2905 chest X-ray images of three categories: COVID-19 affected (219 cases), Viral Pneumonia affected (1345 cases), and Normal Chest X-rays (1341 cases). The implemented neural network demonstrates significant performance in classifying the cases with an overall accuracy of 96%. Most importantly, the model has shown a significantly good performance over the current research-based methods in detecting the COVID-19 cases in the test dataset (Precision = 1.00, Recall = 1.00, F1-score = 1.00 and Specificity = 1.00). Therefore, our proposed approach can be adapted as a reliable method for faster and accurate COVID-19 affected case detection.

Dealing with Covid-19 infections in Kolkata, India: A GIS based risk analysis and implications for future scenarios

The Covid-19 pandemic has inherently affected daily lives of people around the world. We, the people, have already started to live in a new-normal way, such as wearing masks, keeping a safe distance from others and maintaining higher level of health & hygiene standards. However, because of this pandemic, the global economy has taken a major blow. Many people have lost their livelihood, and now facing challenges in getting better healthcare facilities and food for survival. Given the serious challenges to the problem, this paper analyzes demographic data to predict vulnerable areas in Kolkata metropolitan city that houses nearly one-third of its population in slums and one-fifth below poverty level, under compromised living conditions. The analysis revealed that the highest risk areas are located in the east and west of the city, the area to a great extent overlapped with wards containing larger share of population below poverty level and are also living in slums. The analysis of publicly accessible Covid-19 case records and containment zones data indicate the hardest hit areas lies in the central Kolkata and several wards along the eastern and northeastern border of the Kolkata Municipal Corporation. The data further revealed that the virus infections have extended to the south Kolkata with increasing number of broad-based containment zones with heightened cases. The analysis of demographic characteristics of the hardest hit wards revealed that not a single variables are directly associated with the increase in the number of containments for a particular ward. The ranking of wards based on four intervention criterion have suggested that the lack of social awareness along with lack of social distancing have dominantly contributed to the increasing number of containments of Covid-19 cases in Kolkata. Determination of optimized ranking and Spearmans rank correlation coefficient of the wards based on four intervention criterion provided a basis for the policy makers to assess ward-based interventions criterion to control further spread of the disease and/or prevent second wave of infections. Given that the effective antiviral drugs is far away from common publics reach, the application of our study approach would benefit saving lives of many vulnerable populations.

COVID-19 trend in Bangladesh: deviation from epidemiological model and critical analysis of the possible factors

BackgroundSince its first report on March 08, COVID-19 positive cases and number of deaths are increasing in Bangladesh. In the first month of COVID-19 infection, incidence of daily positive cases did follow the susceptible, infected and recovered (SIR) based predictions we reported in April, but started to deviate in the following months. COVID-19 transmission and disease progression depends on multifaceted determinants e.g. viral genetics, host immunity, social distancing, co-morbidity, socio-demographic and environmental parameters. Therefore deviation in confirmed cases from predicted model may appear and warrant thorough investigation. MethodsIn this short report, we compared real data with SIR model and analyzed the possible factors associated with the deviation which included preventive intervention strategies, socioeconomic capabilities, climatic and meteorological indexes, acquired immunity of Bangladeshi population, demographic characteristics, health indicators and food habits. ResultsThe key factor responsible for the observed deviation was found to be the number of tests performed. Having population with low median age, young age groups are being mostly infected. Low prevalence of non-communicable diseases among them and strong immunity compared to the elderly might have kept most of them asymptomatic with silent recovery. Warm temperature, humidity and UV index of Bangladesh during this summer period might have contributed to the slow progression of infection. Longer daylight mediated immunity, fresh air circulations and ventilation, less population density in rural areas and certain food habits perhaps helped the large number of populations to restrict the infection up to a level. ConclusionDespite all these helpful determinants in Bangladesh, person to person contact is still the leading risk factor for COVID-19 transmission. Infection may increase rapidly if safe distance and preventive measures are not strictly followed while resuming the normal social and work life. Expanding test capacity, strong collaborative action plans, strategies and implementation are needed immediately to prevent catastrophe. HighlightsO_LILimited number of tests compared to large population was the key reason for possible low daily positive cases reported in Bangladesh. C_LIO_LIControlled interventions viz. official leave; transport ban and social distancing had helped initially to slow down the transmission. C_LIO_LIWarm weather, high humidity and UV index, sunlight mediated immunity, fresh air circulations, low pollutions, food habit and heterologous immunity might have reduced the transmission capabilities of SARS-CoV-2. C_LIO_LIHaving large number of young people with strong immunity might have kept most of the infected asymptomatic who recovered silently. C_LIO_LIPerson to person contact still remain as key risk factor in COVID-19 transmission, so strict health measures should be in place even after reopening social activities to contain further transmission. C_LI

Lifestyle acquired immunity, decentralized intelligent infrastructures and revised healthcare expenditures may limit pandemic catastrophe: a lesson from COVID-19

Human race has often faced pandemic with substantial number of fatalities. As COVID-19 pandemic reached and endured in every corner on earth, countries with moderate to strong healthcare support and expenditure seemed to struggle in containing disease transmission and casualties. COVID-19 affected countries have variability in demographic, socioeconomic and life style health indicators. At this context it is important to find out at what extent these parametric variations are actually modulating disease outcomes. To answer this, we have selected demographic, socioeconomic and health indicators e.g. population density, percentage of urban population, median age, health expenditure per capita, obesity, diabetes prevalence, alcohol intake, tobacco use, case fatality of non communicable diseases (NCDs) as independent variables. Countries were grouped according to these variables and influence on dependent variables e.g. COVID-19 test positive, case fatality and case recovery rates were statistically analyzed. The results suggest that countries with variable median age has significantly different outcome on test positive rate (P<0.01). Both median age (P=0.0397) and health expenditure per capita (P=0.0041) has positive relation with case recovery. Increasing number of test per 100K population showed positive and negative relation with number of positives per 100K population (P=0.0001) and percentage of test positives (P<0.0001) respectively. Alcohol intake per capita in liter (P=0.0046), diabetes prevalence (P=0.0389) and NCDs mortalities (P=0.0477) also showed statistical relation with case fatality rate. Further analysis revealed that countries with high healthcare expenditure along with high median age and increased urban population showed more case fatality but also had better recovery rate. Investment in health sector alone is insufficient in controlling pandemic severity. Intelligent and sustainable healthcare both in urban and rural settings and healthy lifestyle acquired immunity may reduce disease transmission and comorbidity induced fatalities respectively.

Impact of control strategies on COVID-19 pandemic and the SIR model based forecasting in Bangladesh.

BackgroundCOVID-19 is transmitting worldwide drastically and infected nearly two and half million of people so far. Till date 2144 cases of COVID-19 is confirmed in Bangladesh till 18th April though the stage-3/4 transmission is not validated yet. MethodsTo project the final infection numbers in Bangladesh we used the SIR mathematical model. Confirmed cases of infection data were obtained from Institute of Epidemiology, Disease Control and Research (IEDCR) of Bangladesh ResultsThe confirmed cases in Bangladesh follow our SIR model prediction cases. By the end of April the predicted cases of infection will be 17450 to 21616 depending on the control strategies. Due to large population and socio-economic characteristics, we assumed 60% social distancing and lockdown can be possible. Assuming that, the predicated final size of infections will be 3782558 on the 92th day from the first infections and steadily decrease to zero infection after 193 days ConclusionTo estimate the impact of social distancing we assumed eight different scenarios, the predicted results confirmed the positive impact of this type of control strategies suggesting that by strict social distancing and lockdown, COVID-19 infection can be under control and then the infection cases will steadily decrease down to zero.