Putting the world back to work: An expert system using big data and artificial intelligence in combating the spread of COVID-19 and similar contagious diseases

1Introduction

COVID-19 is a dangerous and infectious virus, which has evolved from an epidemic to a pandemic, due to the fact that it is not limited in space and time affecting many countries around the world. The contagiousness rate is very high (from 2.5 to 5.7) and the mortality rate is between 2 and 6%. The most common symptoms of COVID-19 are fever, dry cough and fatigue, with some patients experiencing pain, nasal congestion, and/or sore throat [1]. COVID-19 is transmitted by people carrying the virus. The disease is transmitted mainly from person to person through respiratory droplets expelled through the nose or mouth when an ill individual coughs, sneezes or talks [1, 2]. Governments around the world have taken several measures to protect themselves from this virus and monitor the evolution of symptoms with the adoption of strict instructions to people on self-isolation (Stay home, isolate yourself, social distancing) [3].

The International Labor Organization reported that in the second quarter of 2020, the global working hours are estimated to be 10.5 percent lower than in the previous pre-crisis quarter in 2019. This led to the loss of an estimated 305 million full-time jobs [4].

Almost every region in the world experienced un-employment percentage increase because of COVID-19. However, the world’s biggest economies are more affected by this. According to the U.S. Labor Bureau of Statistics, COVID-19 brought the total of unemployed individuals to a staggering 40.8 million, as it increased unemployment to a record high of 14.7% in April 2020 [5]. In Japan, 1.76 million are without jobs [6]. China has reported an estimated 80 million individuals without work as of March 2020 [7]. The European Union’s workforce has also taken a toll due to COVID-19, as of March 31, 2020 it was reported that an estimated 14 million people were unemployed [8].

When deciding to return to work after testing positive for COVID-19 the Center for Disease Control and Prevention in the USA recommends that the individual wait at least 10 days from the day their symptoms first appeared [9]. After the ten days the CDC states the person should wait at least an additional 3 days that are marked without the use of fever-reducing medicine and improvement in any respiratory symptoms such as, cough or shortness of breath, before returning to work. Once the person returns to work the CDC recommends that the individual should always wear a face mask and they should self-monitor in case symptoms reoccur or worsen, including taking their temperature twice daily.

Safety at work is even more crucial specifically for people with pre-existing health condition and older adults who may be frail and having some pre-existing health condition such as cardiopulmonary conditions and diabetes.

Preventive measures for older workers are necessary in order for them to continue working and remain healthy [10]. Now this is even more vital due to COVID-19.

The Occupational Safety and Health Administration (OSHA) in the USA advised employers and workers to follow some general practices to help prevent exposure to COVID-19 [11] such as:

While many people lost their jobs, some other workers such as healthcare workers and other essential workers are overworked and need more protection because the entire society now is relying on them. Studies have shown that healthcare workers are predisposed to work related muscular disorders [12]. The highest group of healthcare workers with this risk is nurses [13, 14]. A study found that 63.8% of nurses in a hospital in Saudi Arabia experienced discomfort during the last 12 months in their lower back, followed by shoulders (50%), and upper back (48.9%) [15]. Perhaps the struggle of these workers may increase with the influx of more sick people and most hospitals reaching their full capacity especially with more people who are very sick and need more assistance. Governments are now required to do even more to protect these very important workers such as nursing by providing them whatever equipment they need. OSHA for example recommends that healthcare workers with exposure to suspected or confirmed COVID-19 patients need to wear gloves, gowns eye/face protection (e.g., goggles, face shield) and N95 filter face piece respirators or better that are certified by the National Institute for Occupational Safety and Health [16].

Several countries such as the USA, China, and Taiwan for example, have turned to Big Data (BD) and Artificial Intelligence (AI) to contain this scourge, to detect, to monitor cases of contamination, and to predict the spread of the epidemic [17]. In order to contain contamination, some countries share medical, transportation data (trains and planes in particular), and people’s mobility among other things by utilizing the BD and monitoring the travel history of people. Since the Severe Acute Respiratory Syndrome Coronavirus (SARS-CoV) (outbreak of 2003 [18], Taiwan has put in place a robust pandemic prevention plan based on new technologies, AI and BD, among others [19].

As such, researchers are currently working at pandemic sites collecting vital data on COVID-19, such as transmissibility, risk factors, incubation period, and mortality rate [20]. Data are used for visualizations, creation of mathematical models and neural network formation. In addition to the work that has been done, this research team designed an expert system that combines multi solutions such as predictive analysis by learning algorithms Machine Learning (ML) to provide preventative and information system based on scientific data. An expert system is a programmed tool capable of answering questions, making a reasoning based on known facts and rules. This expert system is one of the common applications used to solve complicated problems in different fields. It focuses on the construction of high- performance programs in specific professional domains [21]. Generally speaking, this expert system is a computer tool based on ML and AI designed to solve certain problems due to the specialist expertise and knowledge in a specific field [22].

The expert system also allows the modeling of an expert’s reasoning, the manipulation of knowledge in a declarative form, the facilitation of its acquisition, modification and updating of data, the obtaining of inferences and conclusions, and the production of explanations on how results are obtained [23]. Therefore, this expert system will address the questions of how can we combat and stop the spread of COVID-19, saving people’s lives, and put the world back to work.

To answer this question, we developed a new sys-tem to provide a General Guide (GG) to the users, containing a set of information concerning the COVID-19 pandemic. Once the users receive enough information about COVID-19, the recommendation is to complete the online check-up. This will provide us with data that will be analyzed using BD and AI algorithms. The results obtained from this analysis will provide indicators that will help decision-makers to make the best decision in terms of monitoring and intervening to limit the spread of COVID-19. With the same objective, other complementary solutions will be introduced such as the as smart automatic car and passenger sanitizing system, smart thermal cameras, and multi-use drones.

In method section of this article it will be discussed how BD should be used and what is being done around the globe to put people back to work in a safe manner. In addition, the proposed expert system to stop the spread of COVID-19 is discussed in multi solutions. Moreover, the discussion will addresses the main solution to COVID-19 concerns and the proposed analysis models that will be used, and the analysis methods that are used in BD and ML to analyze and predict the data of the identified problem.

2Methods

2.1Predictive analyses process with BD

Predictive analysis not only predicts what is happening, but it can also predict what will happen in the future, which is very valuable for better planning and preparation. Rather than simply explaining who, what, where, when, predictive analysis predicts the best course of action that will generate the most optimal return based on an algorithm, such as a regression equation [24]. The objective of predictive analysis is to analyze data and variables to extract predictive capabilities from existing data. Predictive analysis techniques rely on variables that can be measured, manipulating metrics to predict future behavior or outcomes based on a variety of quantifiable approaches. Predictive analysis models combine multiple predictors or quantifiable variables into a predictive model [25, 26]. This approach allows for the collection of data and subsequent formulation of a statistical model, to which additional data can be added as they become available. The prediction process involves the following steps depicted in Fig. 1.

Fig. 1

Predictive analysis process [27].

The predictive analysis of data plays a major role in the fight against the factors that cause epidemics. Predictive data analysis has become more than a powerful tool to help prevent epidemics. Several areas of research have used prediction models such the return of workers with injuries in Hong Kong [28].

The use of predictive data analysis models based on AI, ML, and BD analytics techniques are well suited for tracking and controlling the spread of COVID-19 around the world. Sophisticated predictive analysis platforms that process billions of data obtained from a variety of sources are also helping the World Health Organization (WHO) and other health care organizations around the globe in the fight against the pandemic.

2.2Predictive technical analytics

The predictive technical analytics is a sub-branch of the BD also called predictive logic or advanced analysis, it brings together advanced analytical and statistical techniques to establish predictive scenarios about future events. Predictive analysis uses specific methods from ML and game theory. Thanks to complex algorithms, the combination of these techniques makes it possible to generate models that accurately predict future events [29, 30].

The algorithms of the ML are classified according to supervised or unsupervised learning type depicted in Fig. 2.

Fig. 2

Machine learning algorithms type [31].

2.2.1Supervised learning

For this type of algorithm, the input data is labeled and each is indicated by the corresponding target. The objective of the algorithm is to provide every new input with a label, then to learn from the initial data, while calculating a function that will represent the best model found after this learning, and to make predictions on the unlabeled data. In this type of training, we need to distinguish two types of algorithms: classification algorithms and regression algorithms [32, 33].

Classification algorithms: When the output variables are discrete, i.e. numeric and countable. The problem is then to predict the category of data. The classification can be binary if we have two classes or multinomial if the number of classes is more than two.

Regression algorithms: various type of regression can be applied in the data prediction as the linear regression model and the polynomial regression model among the models which will be applied in this work to predict an example of COVID-19 data. In this application, the output values to be predicted are continuous numerical variables.

2.2.2Unsupervised learning

This type of learning consists of analyzing unlabeled data, then extracting a pattern from this data so we can generate predictions. Unsupervised learning refers to the clustering or partitioning of data into homogeneous groups, calculating a measure of similarity, when the system has only examples, and the number of classes and their nature have not been predetermined. This is called unsupervised learning or clustering [34, 35].

2.2.3Principal algorithms for machine learning

In this research project, the team chose to work with the type of supervised learning and more precisely with the algorithms most used in the analysis and prediction of health data.

Logistic regression: Logistic regression is one of the most widely used ML algorithms for binary classification. Like many other ML techniques, it is obtained from the field of statistics, not as what its name refer to, it is not an algorithm for regression problems, in which one wishes to predict a continuous output. Instead, logistic regression is the reference method for binary classification [36].

The support vector machine (SVM): The SVM is a predictive analysis data-classification algorithm that assigns new data elements to one of labeled categories. SVM is, in most cases, a binary classifier. It assumes that the data in question contains two possible target values [37].

Decision tree: Decision tree is a predictive model to move from observations about an item to conclusions of its target values. It is one of the predictive modeling approaches used in statistics, data mining, and ML. Tree models, in which the target variable can take a discrete set of values. They are called classification trees. In these trees, leaves represent class labels and branches represent conjunctions of entities leading to these class labels.

For example, in the health-work subject we found the work [38] that used logistic regression to process data on past sick leave and demographic variables to predict absences due to illness other work [39, 40] used also the logistic regression to trait measuring psychosocial factors and predicting work ability among cemetery workers.

Gradient Boosting: Gradient Boosting is a technique for automatically learning regression and classification problems [41], which produces a prediction model in the form of a set of weak prediction models, usually decision trees. It builds the model in a stepwise fashion.

Neural network: The neural network algorithm is widely used for data classification. Neural networks process past and current data to estimate future values uncovering all the complex correlations hidden in the data in a manner similar to that used by the human cerebrum. In the rest of the expert system being developed, the structure of a neural network algorithm consists of three layers:

The input layer feeds past data values into the next (hidden) layer. The hidden layer encapsulates several complex functions that create predictors; the output layer collects the predictions made in the hidden layer and produces the final result which is the prediction of the model [42] as depicted in Fig. 3.

Fig. 3

Neural network predictor [43].

The Deep Learning DL is an approach from the representation learning family of ML. Representation learning is also often referred to as feature learning [44]. This type of algorithm gets its name from the fact that it uses data representations rather than explicit data features to perform tasks.

Other predictive models: There are other efficient models that can be applied in the prediction of many parameters of data related principally to seasonality in time series. The first model is Autoregressive Integrated Moving Average Model (ARIMA) that is based on parameterized analytical equations which these parameters are related to degree of divergence, order of the non-seasonal moving average of the error. The second preferred model is the Holt-Winters’ seasonal model when the seasonal variations are roughly constant through the series, comprises the forecast equation and parameters [45]. Also, we can use the genetic algorithms for more performance and results [46, 47].

3The proposed expert system to stop the spread of COVID-19

3.1The general guide application

The main solution of this research team system expert is to use the online application called General Guide (GG) application to limit the spread of COVID-19 in each country worldwide using BD and AI. The use of the GG application is depicted in the examples shown in Fig. 4.

Fig. 4

Example of the home screen of the GG application.

Application home page: guide containing all the information and news about COVID-19, ensuring a diversified awareness that targets all citizens of a specific country.

Online consultation page: Entry of data by citizens will identify the health status of each citizen related to the risk factors of COVID-19. This will generate a massive database that will help in minimizing the spread of this virus and that data could be used in making a scientifically informed decision in how to address COVID-19.

3.1.1GG application figures

This expert system is a web application. It is in Arabic and English and can be available on tablet, smartphone, and computer. It can inform the user about the following points:

• • General definition of COVID-19.

• • How COVID-19 is transmitted.

• • How to protect yourself from COVID-19.

• • Symptoms of COVID-19.

• • Mental health resources manual.

• • Counseling centers.

• • Number of infected people and other statistics.

• • Remote learning links from educational institutions.

• • Working from home tutorial manual.

• • Most asked questions with answers.

• • Emergency phone numbers.

3.1.2Online consultation using the app

This application allows each person to input their personal information (Obligatory step with full respect of the privacy and confidentiality protection).

Personal Information Input (PII): The PPI includes: The first name, last name, sex, date of birth, identification number, nationality, email address, phone number, address function, employer, workplace, etc.

Symptom Identification Input (SII): The SII includes: Fever, cough, respiratory problems, sore throat, runny nose, head-ache, diarrhea, loss of smell or taste, etc, or no symptoms.

Risk factors (RF):

• • Did you have any contact with a person contaminated with COVID-19?

• • Do you have any close family member contaminated with COVID-19?

• • Did you take a COVID-19 test?

• • Do you continue to work or are you in confinement?

• • Have you been in a group of more than 10 people in the past month?

• • Have you used public transportation in the past month?

• • Are you using a vehicle or public transportation right now?

• • Do you know at least 5 people infected with COVID-19 in your area?

• • Health history in the last three months: diabetes, hypertension, respiratory, asthma, allergy, weak immune system, influenza, cold, etc.

• • Are you already infected with COVID-19?

• • Did you visit a hospital or a medical office recently?

Just to point out that this is not an inclusive list of risk factors. Other risk factors can be added to it.

At the end of entering all the information requested by the GG application, a response evaluated by an algorithm will give one of three types of results depicted below in three colored types of flags:

Green flag: Person with very low risk infection of COVID-19.

Yellow flag: Person may have COVID-19.

Red flag: Person is at a high risk of COVID-19 infection.

4Figures and explanation

This result means that the person has no COVID-19 symptoms nor risk factors as depicted in Fig. 7.

Fig. 5

Example of the home screen of the GG application.

Fig. 6

Example of the home screen of the GG application.

Fig. 7

Green flag: Person with a very low risk infection of COVID-19.

This result means that the person has at least one symptom of COVID-19 or at least one risk factor as depicted in Fig. 8.

Fig. 8

Yellow flag: Person is probably contaminated with COVID-19.

The person with a yellow flag should stay isolated at home and contact the appropriate health care services.

This result means that the person has at least three major symptoms of COVID-19 and one or risk factors as depicted in Fig. 9.

Fig. 9

Red flag: Person is at a high risk of infection with COVID-19.

A person with a red flag needs to stay isolated at home, concerned medical facility will receive an alert about his situation and will contact the person immediately to schedule an appointment to perform a COVID-19 test at home. This will reduce traffic and congestions in medical emergency areas and hospitals, and reduce the number of emergency calls.

This will help in reducing overburden on health care workers and minimize the demand of additional heath care workers. In addition, it will help health care systems optimize test planning and response to prevent waste of tests.

The GG application was developed by this research team to allow for these benefits:

3.2Automatic vehicle and passengers sanitizing system

This vehicle and passenger sanitizing system is recommended to be used in the red flag areas or areas with high infection risks. It is composed of two units. The first unit called “Safe Vehicle Gate” which will sanitize all exterior and interior surface areas of the vehicle. The second unit called “Safe Passenger Gate” which will sanitize the entire body surface of the passenger. Both units should use revolving sprayers of disinfectants, also it can use Ultraviolet Light for a complete surfaces sanitizing.

To maximize the benefits of this disinfecting system (Fig. 18) which will be equiped with:

Fig. 10

Example of online consultation from the GG application for a person’s identification.

Fig. 11

Example of the online consultation form in the GG application risk factors.

Fig. 12

Proposed conception of future disinfecting system.

Fig. 13

Examples of multi-purpose drones.

Fig. 14

Actual values and predicted values.

Fig. 15

ARIMA model for daily time series.

Fig. 16

Structure and basic information of COVID-19 data in Morocco.

Fig. 17

ARIMA model for daily time series code.

Fig. 18

Holt-Winters model for daily time series code.

Fig. 19

Evolution of number of performed tests of COVID-19 in Morocco.

Fig. 20

Evolution of total number of daily new cases of COVID-19 in Morocco.

• • Touch screen for the GG application where passengers are required to fill out the check up form before going inside the protected area.

• • HD camera to retrieve the plate numbers of vehicles passing through the system and implement it in the GG application database automatically.

• • Thermal camera to check the passengers’ temperature with an automatic alert to flag passengers with high fever.

The purpose of this system is to limit the spread of COVID-19 between areas, especially that this virus does not move by itself but only through infected people and contaminated surfaces.

4Predictive analysis methods and results

Predictive analysis by the proposed expert system is one of AI models and algorithms. It is used to analyze data and statistics in order to extract predictions or predictive models.

5Putting everything into perspective

Countries around the world are doing everything in their power to tackle COVID-19 since there is no vaccine nor a specific medication for it. However, to tackle COVID-19 effectively, countries must use science and scientific data. The WHO stated that COVID-19 would stay with us for at least 2 years [48]. Countries are bracing themselves for a second wave of COVID-19 in the fall of 2020. CDC Director Robert Redfield told the Washington Post last April in an interview, “There’s a possibility that the assault of the virus on our nation next winter will actually be even more difficult than the one we just went through” [50].

The world still remembers what happened during the 1918 influenza pandemic and no one wants to relive that experience. The world witnessed 3 waves of that deadly influenza within one year. The first wave was in the spring of 1918, the second wave was in the fall of 1918, and the third one was in the winter of 1918. This resulted in at least 50 million deaths worldwide, which was more than the World War 1 deaths of civilians and military deaths [51].

6Conclusion

This developed expert system using BD and AI will effectively in a preventative way assist in combating the spread of COVID-19 and similar contagious diseases. The GG application is an adapted simplified model, which allows easy access to various information about COVID-19. It also permits citizens to do an online checkup, which will display their infection risks. Checkup database information will be collected to allow governments to have a precise vision on the pandemic situation and the state of the spread of COVID-19 using predictive analysis adapted model. This allows concluding indicators to apply preventative confinement in high risk areas of infection and to keep open low risk areas. This can be applied with other solutions, such as people and vehicles sanitizing system, multi-purpose drones, and other solutions that can be added to this expert system. This integrated expert system will facilitate keeping the pandemic situation under control and optimizing the governments’ resources. It will also minimize the impact on the economy by allowing people to continue working safely or go back to work in a safe and faster manner utilizing the flag color indicators for people and areas.

Our expert system is a preventative measure applied on people who are suspected to be infected while the contact tracing applications are used on confirmed infected people. In addition, our expert system is more accurate even with a small percentage of applicants compared to contact tracing applications.

The concept of this expert system can also be applied to other contagious diseases in the future.

Conflict of interest

None to report.