"Swastha-Shwasa": Utility of Deep Learning for Diagnosis of Common Lung Pathologies from Chest X-Rays

Respiratory diseases are one of the leading causes of death and disability in the world. Integration of AI with existing Chest X-Ray (CXR) diagnostics is currently a hot research topic. On similar lines, we propose a technique termed "Swasta-shwasa" for multi-class classification that associates CXR with one among Tuberculosis, COVID-19, Viral pneumonia, Bacteria Pneumonia, Normal and Lung Opacity ailments based on Deep Learning. The proposed technique which has accomplished an overall 98% test accuracy, 0.9991 AUROC, average Specificity of 99.82% and average Sensitivity of 98.51% involves four stages: Pre-processing, Segmentation, Classification and Saliency map visualization. Further, the trained model is used to predict on unseen real life data of COVID-19 cases from India and a cross-population generalization accuracy of 85% is witnessed. XAI is augmented for model interpretability. We also explore why CLAHE may not be suitable choice for pre-processing of CXRs.

Probing the competitive inhibitor efficacy of frog-skin alpha helical AMPs identified against ACE2 binding to SARS-CoV-2 S1 spike protein as therapeutic scaffold to prevent COVID-19.

In COVID-19 infection, the SARS-CoV-2 spike protein S1 interacts to the ACE2 receptor of human host, instigating the viral infection. To examine the competitive inhibitor efficacy of broad spectrum alpha helical AMPs extracted from frog skin, a comparative study of intermolecular interactions between viral S1 and AMPs was performed relative to S1-ACE2p interactions. The ACE2 binding region with S1 was extracted as ACE2p from the complex for ease of computation. Surprisingly, the Spike-Dermaseptin-S9 complex had more intermolecular interactions than the other peptide complexes and importantly, the S1-ACE2p complex. We observed how atomic displacements in docked complexes impacted structural integrity of a receptor-binding domain in S1 through conformational sampling analysis. Notably, this geometry-based sampling approach confers the robust interactions that endure in S1-Dermaseptin-S9 complex, demonstrating its conformational transition. Additionally, QM calculations revealed that the global hardness to resist chemical perturbations was found more in Dermaseptin-S9 compared to ACE2p. Moreover, the conventional MD through PCA and the torsional angle analyses indicated that Dermaseptin-S9 altered the conformations of S1 considerably. Our analysis further revealed the high structural stability of S1-Dermaseptin-S9 complex and particularly, the trajectory analysis of the secondary structural elements established the alpha helical conformations to be retained in S1-Dermaseptin-S9 complex, as substantiated by SMD results. In conclusion, the functional dynamics proved to be significant for viral Spike S1 and Dermaseptin-S9 peptide when compared to ACE2p complex. Hence, Dermaseptin-S9 peptide inhibitor could be a strong candidate for therapeutic scaffold to prevent infection of SARS-CoV-2.

An opinion study about online learning method conducted during COVID-19 pandemic among agricultural students

Online learning is an educational process which takes place over the internet as a form of distance education where the learners and the instructors are not in the same place. Due to COVID-19 pandemic circumstances, online learning played an indispensable role in education programs, where most of the educational institute shifted towards online learning platform to up keep the academic activities. For a developing country like India, the use of ICT in education process still poses many challenges and it is not clear about its effectiveness. The research study focuses on agricultural students' opinion about online learning method which was conducted using social distancing during COVID-19 pandemic situation. For the purpose, an online survey was conducted among 60 randomly selected undergraduate students who had attended the online classes. The results indicated that majority of the students (91.70%) preferred to use smart phone for online learning. Most of them viewed that live classes with quiz at the end of each class helps in effective learning. Majority of the students opined that more interaction during online classes makes it interesting, whereas network connectivity related issues in rural areas makes it a challenge for students to make use of online learning initiatives. However, in agricultural education system, many courses are practical oriented, conducting practical classes in online mode may not be possible. Hence, there is a need to device a hybrid mode of learning. The results of the study can be helpful in making online learning more effective.

Comparative Analysis of Machine Learning Algorithms for Stock Market Prediction During COVID-19 Outbreak

In the current period of time, when there is a havoc across the world due to COVID-19 virus outbreak, it becomes very important to foresee the impact of this pandemic on the world economy. This has attracted us to analyze and predict the stock market prices of some international IT international companies which provide employment to thousands of people and create revenue for many countries namely Google, Microsoft, Apple and Amazon. In this study, we have implemented algorithms such as SVM and LSTM on stock market data to see if major IT companies see a rise or fall during the COVID-19 pandemic. We have also used ARIMA forecasting method to predict the stocks of above mentioned 4 companies. This paper provides a simple but original statistical analysis of the impact of the COVID-19 pandemic on stock market risk for 4 major IT companies of the world. Results revealed that while some businesses like personal computers from Microsoft, I phone handsets, sale of luxury and fashion goods at Amazon has declined during the pandemic, thus leading to fall of stocks. However, some prominent other segments like online shopping, cloud computing and streaming video from Amazon, oversees Office, Dynamics, Skype, LinkedIn Intelligent Cloud from Microsoft, Google’s ad sales during the crisis and issue of cheap bonds by Apple came out to be the winning corporate strategies to fight the negative economic effect of COVID-19 and to stabilize the situation of stocks in coming months. This study may help investors and companies to sustain the tide of economic fall. © 2021, The Author(s), under exclusive license to Springer Nature Switzerland AG.

Adaptations During COVID-19 and Smart Disease Detection Through ‘Germica’

COVID-19 pandemic has shaken the world and has taught a lesson to maintain personal cleanliness and hygiene. In these tough times, only people with higher degree of cleanliness and hygiene could beat this deadly virus and survive the pandemic. Everyday millions of people are losing their lives due to COVID-19, which is majorly spreading due to lack of sanitization and unhygienic surroundings. Simple remedies to fight corona virus including washing hands with soaps, regular use of hand and air sanitizers, use of face mask etc. In public places such as railway stations, airports, restaurants, hospitals, cafes, shopping malls etc., the maintenance and monitoring of hygiene conditions of visiting people conditions becomes a big challenge. To help the government authorities and local administration in better surveillance and monitoring of Corona positive cases, or people contaminated with any other kind of germs and bacteria, we came up with the prototype design of ‘Germica-A germ detection machine’. It is savvy medical and clinic framework, which has multiple germ detection features installed like UV lights to detect the germ intensity on hands or any other targeted body parts. Another feature of ‘Germica’ is germ detection cameras;which can take the image of different body parts, example nails and based to its colour, it can tell if a person is sick or healthy. In preparing the prototype of ‘Germica’, scientist have vigorously tested and trained the medical data to see its actual application and degree of accuracy. Small portable device like ‘Germica’ can help in maintaining hygiene conditions in hospitals, clinics and other places. ‘Germica’ in long run can help to combat with COVID-19 pandemic and other disease that spread through human carrier. © 2021, The Author(s), under exclusive license to Springer Nature Switzerland AG.

Could Dermaseptin Analogue be a Competitive Inhibitor for ACE2 Towards Binding with Viral Spike Protein Causing COVID19?: Computational Investigation.

Initial phase of COVID-19 infection is associated with the binding of viral spike protein S1 receptor binding domain (RBD) with the host cell surface receptor, ACE2. Peptide inhibitors typically interact with spike proteins in order to block its interaction with ACE2, and this knowledge would promote the use of such peptides as therapeutic scaffolds. The present study examined the competitive inhibitor activity of a broad spectrum antimicrobial peptide, Dermaseptin-S4 (S4) and its analogues. Three structural S4 analogues viz., S4 (K4), S4 (K20) and S4 (K4K20) were modelled by substituting charged lysine for non-polar residues in S4 and subsequently, docked with S1. Further, the comparative analysis of inter-residue contacts and non-covalent intermolecular interactions among S1-S4 (K4), S1-S4 (K4K20) and S1-ACE2 complexes were carried out to explore their mode of binding with S1. Interestingly, S1-S4 (K4) established more inter-molecular interactions compared to S4 (K4K20) and S1-ACE2. In order to substantiate this study, the normal mode analysis (NMA) was conducted to show how the structural stability of the flexible loop region in S1 is affected by atomic displacements in unbound S1 and docked complexes. Markedly, the strong interactions consistently maintained by S1-S4 (K4) complex revealed their conformational transition over the harmonic motion period. Moreover, S1-S4 (K4) peptide complex showed a higher energy deformation profile compared to S1-S4 (K4K20), where the higher energy deformation suggests the rigidity of the docked complex and thus it's harder deformability, which is also substantiated by molecular dynamics simulation. In conclusion, S1-S4 (K4) complex has definitely exhibited a functionally significant dynamics compared to S1-ACE2 complex; this peptide inhibitor, S4 (K4) will need to be considered as the best therapeutic scaffold to block SARS-CoV-2 infection.