NLP Text Classification for COVID-19 Automatic Detection from Radiology Report in Indonesian Language

Radiology is used as an important assessment for patients with pulmonary disease. The radiology images are usually accompanied by a written report from a radiologist to be passed to the other referring physicians. These radiology reports are written in a natural language where they can have different systematic structures based on the language used. In our study, the radiology reports were collected from an Indonesian hospital and written in Bahasa Indonesia. We performed an automatic text classification to differentiate the information written in the radiology reports into two classes, COVID-19 and non COVID-19. To find the best model, we evaluated several embedding techniques available for Bahasa and five Machine Learning (ML) models, namely (1) XGBoost, (2) fastText, (3) LSTM, (4) Bi-LSTM and (5) IndoBERT. The result shows that IndoBERT outperformed the others with an accuracy of 98%. In terms of training speed, the shallow neural network architecture implemented with the fastText library can train the model in under one second and still result in a reasonably good accuracy of 86%. © 2022 IEEE.

NLP Analysis of COVID-19 Radiology Reports in Indonesian using IndoBERT

The presence of COVID-19, a respiratory disease, can be detected through medical imaging, such as Chest X-Ray (CXR) and Computed Tomography (CT) scans. These radiology images can also show how the patient's condition progresses. Radiologists need to provide a written report for each image, so that other clinicians can use it in their decision making. In this study, we applied one of the Natural Language Processing (NLP) models called IndoBERT to analyze radiology reports of COVID-19 patients written in Indonesian. We performed two tasks, clustering to group reports by meaning and understand their content, and text classification to predict one of the five possible outcomes for each patient. We show the most frequent topics in radiology reports, and word scores in each topic. The IndoBERT model was fine tuned on a medical text, 'Kamus Kedokteran Dorland' in an attempt to further improve it. This proved unnecessary: on one hand, there were no additional benefits, on the other, the standard model alone achieved a very satisfactory classification accuracy of over 90 %. © 2022 IEEE.

Ideas and Policy Response to the COVID-19 Crisis: Evidence from Jakarta, Indonesia

Coronaviruses have emerged as a potential disruptive force in policymaking. Using a comparative case study method, we examine two social policy responses in Jakarta, Indonesia: the Social Safety Nets (SSN) programme and the health policy. Such examples demonstrate an aggressive change in policy direction from means-tested systems and government-centred approaches to a total relaxation of conditions with the involvement of non-state actors in the provision of services. Our study analyses the ideational dimensions of the policy process that produces abrupt and radical change. From our analysis, the policy change may be explained by the emergence of a new policy paradigm created through the emulation-contextual process - an alternative model of policy learning. The theoretical implication of our research is that policy response in this study cannot be viewed in a completely path-dependent process. Instead, we propose a 'path-creation accelerator,' which represents an infrequent instance of policy change.

The Impact of Twitter Sentiment Analysis on Bitcoin Price during COVID-19 with XGBoost

Bitcoin have become safe-haven for people who want to invest during COVID-19 with its volatile price. Numerous factors can affect the price but recently the most popular one was due to an Elon Musk tweet. We decided to investigate our questions. Do tweets regarding Bitcoin affect its price? Can we predict Bitcoin price by analysing sentiments from twitter? For our research, we decided to analyze the impact of twitter sentiments on Bitcoin price during the COVID-19 pandemic. Using VADER sentiment analysis, we attempted to find out what is the current public sentiment regarding Bitcoin. Coupling tweet sentiment with the Bitcoin price, we pursue making a predictive model to forecast whether Bitcoin price will rise or fall. We also compare whether having twitter sentiment analysis in our model will have an advantage compared to not using. In the end, we found out that twitter sentiment analysis have an impact to Bitcoin price. We hope that our research can help people during this financial stress period. © 2022 IEEE.

Forecasting Vaccination Growth for COVID-19 using Machine Learning

In this digital era, machine learning (ML) is becoming more common in the healthcare industry. It plays many essential roles in the medical field including clinical forecasting, visualization, and even automated diagnostics. This paper focuses on the future prediction of COVID-19 vaccination rates in different countries. Considering how destructive the novel Coronavirus has been and its continuous mutation and spread, clinical interventions such as vaccines serve as a ray of hope for many individuals. As of 2021, an estimated total of 8,687,201,202 vaccine doses by numerous biopharmaceutical manufacturers have been administered worldwide [1]. This study intends to estimate the probable increase or decrease in global vaccination rates, as well as analyze the correlation between future trends of daily vaccinations and new COVID-19 cases, along with deaths and reproduction rates. Three models were utilized in forecasting and comparing the overall prediction toward the COVID19 vaccine rates;Auto-Regressive Integrated Moving Average (ARIMA), an ML approach, Long-Short Term Memory (LSTM), an artificial Recurrent Neural Networks (RNN), and Prophet which is based on an additive model. The Vector Autoregression (VAR) model will also be utilized to compare COVID-19 cases, deaths and reproduction rates to that of COVID-19 vaccine growth. ARIMA resulted to be the best model, while Prophet turned out to be the worst-performing model. In general, our comparison of employing the ARIMA model vs the other three results in the conclusion that adopting this method shows to be a more effective approach in projecting vaccination growth in the future. Furthermore, a visible increase in future daily vaccinations can be seen to be correlated with the increase in COVID-19 cases, deaths reproduction rates, and a fluctuating trend in COVID-19 deaths. © 2022 IEEE.

Forecasting COVID-19 Total Daily Cases in Indonesia Using LSTM Networks

The COVID-19 virus has taken over the course of the world for over two years;governments all over the world have been trying to mitigate its effects in several ways such as instilling most jobs to be done at home instead of working from the office. Thus, it is important to be able to see predictions of COVID-19 cases to better plan the intervention of the virus spreading. With the use of machine learning, our paper aims to propose and evaluate an LSTM (Long Short Term Memory) model that can forecast daily COVID-19 cases in Indonesia. Several tests show that 50 epochs and a batch size of eight are the best parameters to use for our model. Furthermore, after comparison with differing amounts of lookbacks, we have decided that 10 is best for our model as it consistently does better than other numbers of lookbacks. Based on our model, there will still be an increase of COVID-19 cases in the future. © 2022 IEEE.

An XGBoost Model for Age Prediction from COVID-19 Blood Test

COVID-19 was declared a pandemic by the World Health Organization (WHO) in January 2020. Many studies found that some specific age groups of people have a higher risk of contracting the disease. The gold standard test for the disease is a condition-specific test based on Reverse-Transcriptase Polymerase Chain Reaction (RT-PCR). We have previously shown that the results of a standard suite of non-specific blood tests can be used to indicate the presence of a COVID-19 infection with a high likelihood. We continue our research in this area with a study of the connection between the patients' routine blood test results and their age. Predicting a person's age from blood chemistry is not new in health science. Most often, such results are used to detect the signs of diseases associated with aging and develop new medications. The experiment described here shows that the XGBoost algorithm can be used to predict the patients' age from their routine blood tests. The performance evaluation is very satisfactory, with R

Combating Bias in COVID-19 Disease Detection Using Synthetic Annotations on Chest X-Ray Images

Detecting a COVID-19 case by using a deep learning model poses a challenge due to the use of public datasets, where people can contribute and submit images without quality screenings. One of the challenges is that we found many images contain burned-in annotations, such as tubes, letters, numbers, pads, arrows, etc. The annotations become more problematic if multiple datasets are combined due to the limited number of data for COVID-19 cases, and the other datasets do not contain as many burned-in annotations as in datasets containing samples for COVID-19 cases. An example of the issues is that by using a saliency map method, we can find the troubled areas coincide with areas where the annotations are located. In order to combat this annotation bias, we investigate the effect of deliberately adding synthetic annotations to images for all classification classes. Encouraging results are shown in this paper. That is, by using the proposed method, the F1-score can be improved, e.g., an improvement of F1-score of 0.88 can be increased up to 0.94. Therefore, we conclude that adding synthetic annotations in the pre-processing pipeline for datasets having annotation bias could improve a machine learning model. © 2021 IEEE.

Experiment on Deep Learning Models for COVID-19 Detection from Blood Testing

Due to the equipment and expert shortages in diagnosing COVID-19 disease, detecting an individual infected with Coronavirus using hematochemical data could provide a cheaper and faster alternative. The quicker and less expensive alternative could be realized by utilizing deep learning to classify Coronavirus infection using complete blood count test results. Two architectures are developed and implemented in this study, which is custom-built DNN (Deep Neural Network) and TabNet. Also, three datasets from the hospitals in Italy, Brazil, and Indonesia are used for training the models. The deep learning models trained with the datasets from San Raphael Hospital in Italy, Albert Einstein Hospital in Brazil, and Pasar Minggu Hospital in Indonesia obtained average AUC scores of 0.87, 0.90, and 0.88, respectively. Based on the results obtained, this method of diagnosis could serve as an alternative in developing countries to diagnose COVID-19 disease without costly RT-PCR equipment and the expert to operate it. © 2021 IEEE.

IoT-based COVID-19 Patient Vital Sign Monitoring

The number of COVID-19 cases is growing rapidly, while there is not enough healthcare workers which can help the patients. Even worse, the highly contagious nature of this disease, requires the medical staff to be more restrictive and wear the Personal Protective Equipment (PPE) all the time when handling the patients directly. In this situation, a remote system which can monitor patient progress from a distant is inevitable. The emerge of Internet of Thing (IoT) technology has been implemented in many domain. The availability of smart technology, where almost all devices around us has connectivity to the internet, allow people to automate process from distance. The implementation of IoT has also been shown very helpful in medical domain, especially during the pandemics. The IoT technology can be a suitable solution for monitoring patients with a highly contagious disease. The technology can also be very helpful for people who live far from healthcare facility. This can allow people to report immediately and even connect to the hospital system in real-time. In this paper, we propose the use of three different sensors, namely: heart-rate and pulse oximeter sensor (MAX30102), temperature sensor(DS18b20) and accelerometer sensor, which is integrated in a web-based early warning monitoring system for COVID-19 patients. © 2021 IEEE.

COVID-19 Self-Detection Magic Mirror with IoT-based Heart Rate and Temperature Sensors

COVID-19 has many symptoms and one of the serious symptoms are heart problems and the increase in body temperature. Checking the heart rate and body temperature can still be useful to be included in our daily live to prevent further spread of the virus. We proposed a solution to let the user do a COVID-19 self-detection by using Magic Mirror with IoT -based technology. This Magic Mirror uses two sensors (heart rate and temperature sensor) to measure user's heart rate and body temperature. If the user is suspected of having COVID-19, an alert will be displayed on the Magic Mirror or smartphone to let the user take further necessary action. © 2021 IEEE.

Evaluating Deep Learning for CT Scan COVID-19 Automatic Detection

Aside from Reverse Transcription Polymerase Chain Reaction (RT-PCR), another common method to check for the 2019 novel Coronavirus disease (COVID-19) is by using a chest CT scan. Imaging data is profoundly useful in the diagnosis, detection of complications, and prognostication of COVID-19, displaying various spots in the lungs affected by the viral infection. The complex results often require some time before radiologists can analyze them and are more prone to human errors. Inventions of medical assisting tools, through enhancement of artificial intelligence, are crucial in fighting the COVID-19 pandemic through automation of classifications and the future of medicine. To overcome the above challenges, this paper aims to propose and evaluate the performance between Convolution Neural Network (CNN) and Transfer Learning (TL) in the detection of COVID-19 infections from a Lung CT Scan. Gradient-Weighted Class Activation Mapping (Grad-CAM) will also be utilized to display the infected areas in the lungs for explorative experiments. Transfer-learning using our pre-trained model resulted in a detection accuracy result of 89% while our proposed CNN demonstrated the best result in terms of classification accuracy at 97%. Training time required for the two frameworks are 12 and 22 minutes respectively. By and large, our comparison of using the CNN model versus the pre-trained model gives rise to the conclusion that using the former method proves to be a more effective technique of COVID-19 detection by CT-scan. © 2021 IEEE.

A Tree-based Mortality Prediction Model of COVID-19 from Routine Blood Samples

COVID-19 has been declared by The World Health Organization (WHO) a global pandemic in January, 2020. Researchers have been working on formulating the best approach and solutions to cure the disease and help to prevent such pandemics in the future. A lot of efforts have been made to develop a fast and accurate early clinical assessment of the disease. Machine Learning (ML) has proven helpful for research and applications in the health domain as a way to understand real-world phenomena through data analysis. In our experiment, we collected the retrospective blood samples data set from 1,000 COVID-19 patients in Jakarta, Indonesia for the period of March to December 2020. We report our preliminary findings on the use of common blood test biomarkers in predicting COVID-19 patient mortality. This study took advantage of explainable machine learning to examine the data set. The contribution of this paper is to explain our findings on predicting COVID-19 mortality, including the role of the top 11 biomarkers found in our dataset. These findings can be generalized, especially in Indonesia, which is now at its highest peak of the epidemic. We show that tree-based AI models performed well on predicting COVID-19 mortality, while also making it easy to interpret the findings, as they lend themselves to human scrutiny and allow clinicians to interpret them and comment on their viability. © 2021 IEEE.

COBY: COVID-19 Telegram Chatbot by Employing Machine Learning Algorithms

COVID-19 pandemic has been one of the biggest concerns nowadays. People always curious and ask for immediate responses regarding the current situation. The chatbot can be very useful in this kind of situation which allows the system to understand text, which means it can respond appropriately. In order to be able to return the correct responses, the chatbot needs to learn how to classify the text data input from the users. In this paper, we study three different machine learning algorithms to work on text classification problems, namely Naive Bayes, Neural Network, and Support Vector Machine (SVM). An experiment was carried out to study which machine learning algorithms produce the most accurate responses when they are implemented in the Artificial Intelligence (AI) chatbot systems. In order to make sure the tests are consistent and fair, we conducted the experiment on the same dataset, and assessed the accuracy of their respective responses. In addition, we have also successfully implemented each of these algorithms as chatbots on a social media platform, Telegram. © 2021 IEEE.

Impact of Aligning Saliency Maps on COVID-19 Disease Detection Using Chest X-Ray Images

The Coronavirus disease 2019 (COVID-19) has been spread across the world in the year 2020. During the same period, many deep learning researchers have proposed different screening or diagnostic methods as an alternative to the commonly used method, e.g., reverse-transcriptase polymerase chain reaction (RT-PCR). One of the alternatives is the use of chest X-ray (CXR) images. In this paper, we first highlight the fact that by using public, pretrained deep learning model can yield a bias result. For example, by applying a saliency map, we show that a model point to features that are located outside of the lungs. In addition, by applying multiple saliency maps, differences in locations where a model focuses on can be observed. Therefore, We propose a new loss function where we constraint the saliency maps to converge to the same region. The results show that the proposed method is better compared to the model without alignment, where the F1-score of the proposed model is 91.3% versus 89.2%. © 2021 IEEE.

Sonia: An integrated indonesia online tourism system in new normal era

Indonesian travel and tourism have grown to become one of the largest economic sectors in the world. Tourists are the main source of income for gross domestic income in Indonesia. However, although Indonesia as a beautiful archipelago country has so much offer to foreign and local tourist, until now it fails to give their potential to the country. Furthermore, the COVID-19 situation made Indonesia tourism worse. Travelling is banned as a consequence of lockdown in Indonesia. Business and economic need to adjust into their new normal. Unfortunately, in the midst of the current era of Internet technology and new normal, Indonesia’s tourism promotion is considered to be uneven and unwell targeted. Many tourist destinations are very interesting and clean but not yet known by potential tourists both from domestic and abroad due to the lack of targeted promotions and re-branding of a safe place to visit. Seeing to this prob-lem, this study aims to propose an integrated tourism system called SONIA, by using a Service-Oriented Architecture (SOA) and an Artificial Intelligence (AI) model to build a personalized recommender system. Through this system integration model, a significant socio-economic benefit can be made for tourism sector in Indonesia. The results of the research will be exploited for the tourism industry and Indonesia’s participation in the global tourism competition market. © 2020, ICIC International. All rights reserved.