521 CoronaGo Website Integrated with Chatbot for COVID-19 Tracking a a b b b Anil K. Pandey , R. R. Janghel , R. Sujatha , S. Sathish Kumar , T. Sangeeth Kumar , Jyotir Moy c, * Chatterjee a NIT Raipur, Chhattisgarh, India b Vellore Institute of Technology, Vellore, India c * Lord Buddha Education Foundation, Kathmandu, Nepal Abstract The first cases of a typical pneumonia of unidentified ailment were reported on December 30, 2019, from Wuhan, China. After many researches, severe acute respiratory syndrome corona virus 2 (SARS-CoV-2) is found as the main reason of the ailment and the problem has been named as COVID-19. The rapid spread of this virus resulted in the worldwide pandemic state. This global pandemic has made a devastating impact on several domains like education, business and others. There are many problems that the people are facing in this situation. The medical department staff are facing problem in providing medical assistance to the people in need, providing awareness among the people has become difficult, there are many people who need financial help and the list goes on. As of now, there are some websites and mobile applications to help the people to fight these problems. Here in this work, we are proposing a website incorporated with a healthcare chatbot for assistance & tracking the COVID-19 situation. Keywords COVID-19, Website, SARS-CoV-2, Global Pandemic State, Tracking 1. Introduction solve the COVID-19 crisis and help people Recently, an outbreak caused by the virus using technology. named SARS-CoV-2 has impacted the lives of As a collective solution to all the problems, humans very badly across the globe. The very we are proposing a user-friendly, reliable web first occurrence of COVID-19 was enlisted in application that includes a COVID-19 tracker, December 2019 in China. The infection may COVID-19 prediction, a Chatbot, and many outspread from bats to people through another other features which are solutions to some median host and cause extreme respiratory problems faced by people. We are trying to disorder, described by strong man-to-man integrate an efficiently developed Chatbot, transferal through the air [4]. From that which can assist people to surf the website and particular day, there’s a rapid growth in the also accurately answer the COVID-19 related number of cases being registered daily. And the queries they have. many countries were under lockdown for almost On the internet, there are many applications, 3-4 months. During this period, people face websites that are designed and developed to many problems financially, medically. This predict the COVID-19 outbreak. The models global pandemic has even made a negative used various machine learning algorithms, deep impact on the economy of most of the countries. learning algorithms and a few have used some This global pandemic has made a statistical methods to do predictions. All these devastating impact on several domains like models provide acceptable accuracy but the education, business, and others. There are development of the model is complex in nature. many problems that people are facing in this To eliminate the hassle included in the situation. The medical staff is facing problems development of the model, we tried to design a in providing medical assistance to the people simple mathematical algorithm, called Generic in need, providing awareness among the hypothesis algorithm to make the predictions people has become difficult, many people need without compromising on the accuracy of financial help and the list goes on. We need to predictions. This paper aims to design and develop a reliable and easy to use the web application ISIC’21: International Semantic Intelligence Conference, February 25-27, 2021, Delhi, India EMAIL: jyotirchatterjee@gmail.com (J.Chatterjee) 2021 Copyright for this paper by its authors. Use permitted under Creative Commons License Attribution 4.0 International (CC BY 4.0). CEUR Workshop Proceedings (CEUR-WS.org) 522 through which help can be offered to the people level. Data is first fed into the logistic model in need. The flow of development starts with the and then cap value is given to the prophet requirement analysis, finalizing the design of the model for forecasting. This paper concludes application followed by the Chatbot, COVID-19 that a hybrid logistic and prophet model has prediction model development, and then been good in predicting the epidemic trend and integrating all the developed components. it is also capable of predicting the number of infections that might occur across the globe or 2. Related Work in particular country. [5] proposes a system that can identify and In a pandemic like this, providing timely track infected people and implement information to the public is very important. So, quarantine. To develop this system, the authors thought of developing a COVID- technologies like artificial intelligence, digital 19 tracker. A stage like Corona Tracker will recorders, quick response codes, and mobile help the public authority & specialists to applications are used. It helps in stopping the spread checked articles, give updates to the communal spread of disease. It fails to track circumstance, & backer great individual infected people who don’t carry their devices cleanliness to the individuals. They used the and the system also violates civil liberties. data from the John Hopkins University (JHU) which is a trusted source [1]. They used [6] proposes a model that diagnoses Susceptible-Exposed-Infected Removed infected people, monitors clinical status and (SEIR) model to do the predictions of COVID- also predicts the required capacity to provide 19 outspread. telemedicine, virtual care services. This can be achieved by using artificial intelligence (AI) In [2], authors propose a system which and machine learning (ML) techniques can be screens individuals for disease. They used artificial intelligence (AI), digital artificial used for providing telemedicine, virtual care intelligence (AI), digital thermometers, mobile services. Sometimes the system may fail in phone applications, thermal cameras, web- diagnosis of disease and development of based toolkits for developing this system. This system involves high costs. system gives data on infection pervasiveness [7] proposes a telemedicine service which & pathology, recognizes people for testing, can be accessed and used by the people in all contact following, & confinement. It neglects locations. Using services like this can reduce the to identify asymptomatic people whenever number of people coming out of homes and that dependent on self-detailed side effects or directly impacts the outbreak of COVID- observing of fundamental signs, includes 19. For the disease diagnosis, virtual checkups significant expenses & requires validation of and care authors used AI. System helps to screening tools. transport the medicine to the particular patient [3] proposes a system that helps in tracking at immediate from online booking but the the people who might get infected with COVID- transportation time may be large for some 19. The developed can identify and track people remote areas, which makes the patients into who may have come into contact with the tainted danger. individual utilizing worldwide situating [8] proposes a model for anticipating frameworks, constant checking of cell phones, COVID-19 threatening movement with AI and wearable intelligent gadgets. As the system methods. The proposed model can be viably identifies the people who got in contact with the utilized for discovering the mellow patients infected person, we can contact them, and ask who are anything but difficult to weaken into them to take tests, isolate them to stop the viral extreme/basic cases, so such patients get spread to some extent. There are few convenient therapies while reducing the disadvantages to the system like it can’t track the restrictions of clinical assets. There’s a scope exposed people when the device is offline, there for wrong predictions and this leads to the is a scope for cloud breach. wastage of medical facilities. In [4], the authors proposed forecasting Chatbots may be highly useful in pandemic models with logistic and prophet model to situations like this because people want to know predict COVID-19. The data is collected from where, how and at what rate the infection is JHU, which released a dashboard at the country spreading. But information dissemination, symptom monitoring, providing mental health support are challenging tasks in the 523 development of these chatbots [9]. If the idea of developing mobile applications to track chatbots are designed and developed in an their health. To do this they proposed usage of efficient way they can solve the problem of GRU neural network, SEIR model and other misinformation, which is one of the major techniques [14]. The cost of development is problems in the pandemic situation. high and it’s a challenge to collect useful In [10], authors proposed a forecasting prospective data from social media. model which can predict number of [15] proposes a system that predicts the confirmations, recoveries and deaths registered patient’s health condition using XGBoost because of COVID-19. Prediction models such classifier, machine learning based CT as the PA, ARIMA, and LSTM algorithms radiomics models. The predictions are made were used to predict the number of COVID-19 based on the patient health records submitted. confirmations, recoveries, & deaths over the Having access to the health records helps in next 7 days and acquired prediction accuracies studying the case properly and treat them in of 99.94%, 90.29%, and 94.18%, respectively. the best way possible. Besides the advantages Under this paper they also propose a diagnosis this model also has its disadvantages as the model using VGG-16 to detect COVID-19 system requires large amount of private data. utilizing chest X-ray images. The model In [20], authors suggested a model which allows the rapid & reliable detection of will be very helpful to analyze the expansion COVID-19, enabling it to achieve an F- of COVID-19 utilizing Multilayer perceptron, measure of 99% using an augmented dataset. Linear regression & Vector autoregression [11] proposes a system that works for approaches on a publicly available COVID-19 limiting the COVID-19 transmission, increase Kaggle dataset for COVID-19 cases in India. health care providers capability and capacity; [21] introduced a modified Random Forest prevent/predict the future outbreaks. For this model hybridized via the AdaBoost method for system they used telemedicine, tele-critical care, COVID-19 patient fitness forecast. tiered tele-mentoring. This system makes sure [22] tried to find out possible Statistical that the patient gets convenient healthcare from Neural Network (SNN) models along with their the comfort of their own home. This might be advanced methods for COVID-19 mortality good for treating patients with small diseases like prediction in Indian context & predict COVID- flu or general fever but are not efficient to treat 19 death cases. people with some serious health issues. 3. Proposed Approach Lately, social media is considered as one platform to share information to have maximum 3.1Objectives reach. To make use of this fact the authors have The main objective of the web application come up with an idea of bringing awareness and is to help every user in this crisis situation. The social control in the public using social media objectives are like giving government helpline [12]. They used smart phone thermometers numbers to the citizens, then finding top most instead of the regular apparatus and they also affected places in India and Tamil Nadu, then used cough type detection using an extensive set to provide COVID-19 tracker to find state case of acoustic features applied to the recorded details and also connect contributor and audio. This might not require huge investments receiver in the crisis. but requires a lot of time to do all the campaigns The chatbot assistant helps in getting every and show visible results. objective by means of chat. Then also helping The authors [13] propose a system that can- users in providing every guideline provided by do disease diagnosis using the radiology world health organization (WHO). images. AI & deep learning are some of the techniques that they preferred to use in 3.2Architecture Diagram building this system. This system helps in The figure 1, presents the architectural decreasing the exposure of patient to radiation diagram of the proposed system. In this and it requires no preparation but it is more approach the user will have to go to the expensive compared to the radiography and CoronaGo website and there he will get a provides basic anatomic information for only a forum, which is linked various contributing & few tissue densities. receiving units. The website is also having a As mobile partnership has widely increased in the recent years, the authors came up with the 524 prediction & mask (3D) ordering system. A the hotspot places of India (specially Tamil healthcare chatbot is incorporated in the Nadu) due to COVID-19, a COVID-19 tracker website which is linked with the various is also linked with the website. helplines units for COVID-19, one can check Figure. 1 Architecture diagram of our application 3.3Methodology India’s states and union territories helpline We approached solution for the pandemic numbers are trained, then by invoking state or situation using web application development union territory name we will get their state’s with 3 major functionalities like NLP Chatbot helpline number as a response. Integration, automating our 3D printer and Remote education is machine learning sending live print stream using Raspberry Pi 3, based where the data are trained and used forecast predicting the cases using exponential according to the node.js program we coded. function and forum for more. The Hotspot locations we get from the developed chatbot was developed by node-red NLP Chatbot Integration console in that using world map node, we We built Chatbot using Dialog flow console marked the Top 10 affected locations using works with the help of google cloud. This NLP their latitude and longitude coordinates by chatbot is a fully automatic chatbot where input getting dynamic API which was developed by gets invoked and response trained are processed reusing the JHU’s API. and sent by google cloud. The invoking phrases The COVID-19 tracker is developed with are trained, then the trained inputs processing the help of JHU API. can be manipulated using fulfillment coding The Chatbot is integrated in web using node.js program and the response for the application as a widget by using Botcopy to input phrases are also trained accordingly. make the widget as a script which connects The Chatbot in this application helps in with the google cloud directly to invoke the getting, input phrases. ❖ Government Helpline numbers ❖ Remote Education 3D Printer automation using IoT o Learn A – Z (which for We used Raspberry Pi 3 to automate our 3D children under age of 6) printer using octopi application and configured o Learn Tables (for above our 3D printer with that application. Then age of 7) connecting Raspy Cam with Pi then enabling ❖ Hotspot Locations and COVID-19 camera features in terminal. Later when we Tracker receive order from the web application, we will be sending the live stream URL of 3D printing 525 of their own order through mail and also in Figure 2 presents the result of the COVID- SMS. The streaming is prepared by coding the 19 tracker we developed using JHU dynamic spaghetti detective plugin connection in API, with HTML, CSS & JS coding. The Raspberry Pi, so that the customers can watch result from this tracker will be more accurate their mask printing lively and give feedback to because of the certified JHU API. us. Figure 3 is the result of the Hotspot locations we developed; this result was developed using Forecast Prediction of cases Node-red console with the help dynamic API of The forecast prediction of COVID-19 cases JHU, but the data we get from that API was is prepared by using general exponential coded accordingly for our idea to get only top 10 function. We used this mathematical function locations of India affected by Corona which also because the cases in America is increased gives the coordinates and case count details exponentially, so rather for every country it within the location pin. applies. So, after getting 10 days of case details we will be dividing every 2 days total cases (2 points in a graph) the resultant will be its growth factor. For that 10 days we will be getting 5 growth factors and taking mean for that growth factors. Then using the exponential function Y = abx where a is the current total cases, b is the mean of growth factor and x is the number of predication days we want to predict. By using this general exponential function, we got around 89 to 92% of accuracy in prediction. Forum In this which we used Laravel php framework to develop a contributor tab for contributing and receiver tab for needy, where Figure 3. Hotspot Location contributor can contribute money to PM funds or non-monetary things like mask, dry ration 5. Conclusion or food by updating their region details. So, the receiver tab contains form asking for The advancement of the web technologies region and display the contributions present in and techniques are used in this website. The that region and the needy can request the planned requirements and functions are contribution the contributor will receive the achieved in the development of this project. request mail from our team and they will send This project helps the user in getting most of the location to collect the contribution. the information’s majorly needed during this pandemic situation. The proposed systems are 4. Result mostly a single major feature application, but we combined everything together and made it work it as a light weight application. This is also has been developed in android for converting into a app. In future we can make chatbot more accurate and efficient, the geofencing concept can also be added. Then in Android still more upgradations can be done. The deep learning predication can also be integrated in web application in future. References Figure 2. Our proposed system 526 [1] Hamzah, F. B., Lau, C., Nazri, H., Ligot, D. Industrial Management Applications, 12, V., Lee, G., Tan, C. L., ... & Chung, M. 168-181. H. (2020). CoronaTracker: worldwide [11] Awasthi, R., Pal, R., Singh, P., Nagori, COVID-19 outbreak data analysis and A., Reddy, S., Gulati, A., ... & Sethi, T. prediction. Bull World Health Organ, 1, (2020). CovidNLP: A Web Application 32. for Distilling Systemic Implications of [2] Miner, A. S., Laranjo, L., & Kocaballi, A. COVID-19 Pandemic with Natural B. (2020). Chatbots in the fight against Language Processing. medRxiv. the COVID-19 pandemic. npj Digital [12] Scott, B. K., Miller, G. T., Fonda, S. J., Medicine, 3(1), 1-4. Yeaw, R. E., Gaudaen, J. C., Pavliscsak, [3] Weemaes, M., Martens, S., Cuypers, L., H. H., ... & Pamplin, J. C. (2020). Van Elslande, J., Hoet, K., Welkenhuysen, Advanced Digital Health Technologies J., ... & Laenen, L. (2020). Laboratory for COVID-19 and Future Emergencies. information system requirements to manage Telemedicine and e-Health. the COVID-19 pandemic: a report from the [13] Kumar, A., Gupta, P. K., & Srivastava, A. Belgian national reference testing center. (2020). A review of modern technologies Journal of the American Medical for tackling COVID-19 pandemic. Informatics Association. Diabetes & Metabolic Syndrome: Clinical [4] Wang, P., Zheng, X., Li, J., & Zhu, B. Research & Reviews. (2020). Prediction of epidemic trends in [14] Xing, Y., Mo, P., Xiao, Y., Zhao, O., COVID-19 with logistic model and Zhang, Y., & Wang, F. (2020). Post- machine learning technics. Chaos, discharge surveillance and positive virus Solitons & Fractals, 139, 110058. detection in two medical staff recovered [5] Wang, C. J., Ng, C. Y., & Brook, R. H. from coronavirus disease 2019 (COVID- (3). March 2020, posting date. Response 19), China, January to February 2020. to COVID-19 in Taiwan: big data Eurosurveillance, 25(10), 2000191. analytics, new technology, and proactive [15] Theeke, L. A., Mallow, J. A., Moore, J., testing. JAMA doi, 10. McBurney, A., Rellick, S., & VanGilder, [6] Rothan, H. A., & Byrareddy, S. N. (2020). R. (2016). Effectiveness of LISTEN on The epidemiology and pathogenesis of loneliness, neuroimmunological stress coronavirus disease (COVID-19) response, psychosocial functioning, quality outbreak. Journal of autoimmunity, of life, and physical health measures of 102433. chronic illness. International journal of [7] McCall, B. (2020). Shut down and nursing sciences, 3(3), 242-251. reboot—preparing to minimise infection [16] Nuruzzaman, M., & Hussain, O. K. in a post-COVID-19 era. The Lancet (2020). IntelliBot: A Dialogue-based Digital Health, 2(6), e293-e294. chatbot for the insurance industry. [8] Wang, W., Wang, Y., Zhang, X., Jia, X., Knowledge-Based Systems, 105810. Li, Y., & Dang, S. (2020). Using WeChat, [17] Shahid, F., Zameer, A., & Muneeb, M. a Chinese Social Media App, for Early (2020). Predictions for COVID-19 with Detection of the COVID-19 Outbreak in deep learning models of LSTM, GRU and December 2019: Retrospective Study. Bi-LSTM. Chaos, Solitons & Fractals, JMIR mHealth and uHealth, 8(10), 140, 110212. e19589. [18] https://eaca.eu/covid-19-related- th [9] Bai, X., Fang, C., Zhou, Y., Bai, S., Liu, works/ (26 July 2020) Z., Xia, L., ... & Xie, X. (2020). [19] https://www.who.int/emergencies/disease Predicting COVID-19 malignant s/novel-coronavirus-2019/technical- th progression with AI techniques. guidance (10 August 2020). [10] Alazab, M., Awajan, A., Mesleh, A., [20] Sujath, R., Chatterjee, J. M., & Hassanien, Abraham, A., Jatana, V., & Alhyari, S. A. E. (2020). A machine learning (2020). COVID-19 prediction and forecasting model for COVID-19 detection using deep learning. pandemic in India. Stochastic International Journal of Computer Environmental Research and Risk Information Systems and Assessment, 1. 527 [21] Iwendi, C., Bashir, A. K., Peshkar, A., Sujatha, R., Chatterjee, J. M., Pasupuleti, S., ... & Jo, O. (2020). COVID-19 Patient health prediction using boosted random forest algorithm. Frontiers in public health, 8, 357. [22] Dhamodharavadhani, S., Rathipriya, R., & Chatterjee, J. M. (2020). COVID-19 Mortality Rate Prediction for India Using Statistical Neural Network Models. Frontiers in Public Health, 8.