Application of Articial Intelligence for Rapid Prevention of Epidemic Diseases (COVID-19)

9 Background: Epidemic diseases are hazardous in terms of a rapid outbreak. Rapid control 10 of these diseases by ﬁnding patients and quarantine and treatment can be the only tool to 11 reduce the number of cases and mortality at the beginning of the outbreak, in the absence 12 of therapy and vaccines. COVID-19 (coronavirus) is a deadly viral disease that causes severe 13 respiratory illness and spreads through the air. Artiﬁcial intelligence (AI) technologies have 14 played an essential role in solving complex problems. The use of these technologies in response 15 to the challenges posed by the COVID-19 epidemic can reduce the eﬀects of epidemics in various 16 contexts. 17 Objective: The purpose of this article is to review the applications of artiﬁcial intelligence 18 in cases of contagious disease. In this work, COVID-19 disease has been used as an example of 19 dangerous infectious diseases (while the studied methods can be used for all contagious diseases), 20 and a systematic review of the literature on the role of artiﬁcial intelligence as COVID-19 has 21 become a comprehensive and critical technology for combating epidemiology, diagnosis, and 22 disease progression. 23 Methods: A complete search of the literature has been done using the databases of PubMed, 24 Scopus, Web of Science, and Google Scholar, and other sources. In this work, the aim is to review 25 articles that the authors believe can be helpful in the prevention of infectious diseases in the 26 event of an outbreak of artiﬁcial intelligence in the prevention of more casualties. The ﬁrst steps 27 needed in a ﬂurry of a disease (including coronavirus) include identifying the primary suﬀerers 28 and isolating them latter were scanned using CT or ultrasound scans, chest radiographs, or positron emission to- 35 mography/computed tomography (PET/CT) scans. Chest x-ray and CT scan are the imaging 36 modalities that are most widely utilized for the diagnosis and management of COVID-19 pa- 37 tients, with chest CT scan being more accurate and sensitive in diagnosing COVID-19 at an early 38 stage. Only a handful of studies have looked into the roles of ultrasonography and PET/CT 39 scans in diagnosing COVID-19 infection. 40 Conclusions: We gathered research from the existing COVID-19 literature that employed 41 artiﬁcial intelligence-based methodologies to give insights into various domains of COVID-19 in 42 this systematic review. Our ﬁndings indicate critical variables, data formats, and COVID-19 43 sources to help with clinical research and translation. Findings from this study may also assist 44 in reducing the harm caused by the pandemic in the case of such epidemic diseases in the future. 45


Introduction
parts, we will look into COVID-19 illness and ways for recognizing the virus:  Dynamic lung CT of COVID-19 described and summarized in 4 steps (21). In summary, the first 137 four days after early symptoms, the early stage is considered, and GGO can be seen subpleural in 138 the lower lobes unilaterally or bilaterally. Progressing stage is 5-8 days when it is possible to find 139 diffuse GGO spreading over bilateral polylobed. At the peak stage (9-13 days), dense consolidation 140 becomes more common. Finally, when the infection is controlled, absorption occurs (usually after 141 14 days), consolidation is gradually absorbed, and only GGO remains. These x-ray patterns are 142 essential evidence for CT-based classification and COVID-19 severity assessment. UNet++-based segmentation model, which as a result segments lesions associated with COVID-19.

147
The results of this method were: accuracy = 95.2%, sensitivity = 100%, and specificity = 93.6%. In 148 another dataset of 16 patients with viral pneumonia and 11 patients without pneumonia, this model 149 identified all patients with and nine patients without pneumonia. As a result, the time of reading 150 for radiologists was reduced by 65% using AI results.    One of the significant challenges that have garnered considerable attention is the difference between 172 the lung injuries caused by COVID-19, pulmonary edema, and other cases in CT images. It was ob-173 served from early descriptions of respiratory failure due to COVID-19 that some patients experienced 174 hypoxemia that was disproportionate to the reported dyspnea or level of radiological opacity, with 175 greater than typical respiratory system compliance and less work of breathing. One idea that has 176 attracted much attention, especially on social media and in medicine, is the notion that lung injury 177 due to COVID-19 is more like pulmonary edema. This conclusion, expanded on social media, has 178 led to further speculation that therapies commonly used to prevent and treat pulmonary edema and 179 other acute altitude sicknesses may benefit patients with lung injuries due to COVID-19. However,       There have been multiple ways for lung segmentation in the literature with various aims (63).

279
The U-Net (Fig. 3) is commonly used in COVID applications to segment lung areas and lung lesions.    If a person has been diagnosed and proven to have COVID-19, the next critical step is to prevent 321 the disease from spreading further. According to the WHO, the virus is spread from person to person 322 primarily by contact with saliva, drops, or nasal discharge (69). Contact tracing is an essential public 323 health method for breaking the virus transmission chain to restrict the spread of SARS-Cov-2. To 324 avoid new outbreaks, the call tracking procedure identifies and manages persons who have recently 325 been exposed to a patient with COVID-19. Generally, this method identifies the afflicted person 326 and provides 14 days of follow-up from the moment of exposure. If this method is appropriately 327 executed, it can break the present novel coronavirus transmission chain, inhibit its spread, and lessen 328 the severity of the recent outbreak. In this regard, many infected countries use various technologies 329 such as Bluetooth, Global Positioning System (GPS), social charts, contact details, network-based 330 API, mobile tracking data, card transaction, digital call tracking process with the mobile application.

331
The system's data and physical address. The digital call tracking procedure is quicker and more 332 in real-time than the non-digital system. These digital applications are meant to capture personal 333 data, which is then evaluated by machine learning and artificial intelligence algorithms to follow a 334 person exposed to a new infection due to their recent contact chain.

335
The Google Scholar publications identify the several countries that have such ML and AL-based 336 call tracking schemes. According to studies, more than 36 countries have effectively implemented 337 digital call tracking in a centralized, decentralized, or a combination of the two approaches to 338 minimize work and boost the efficacy of traditional health care detection systems. (70).

339
In the case of contact tracking, studies have proven the use of ML and AI in enhancing the 340 contact tracking process against infectious chronic wasting disease (71). After applying graph theory 341 to animal infectious disease epidemic data, mainly inter-farm transport data, the resulting graph 342 properties produced by the proposed model can be used to increase more efficient contact tracking.

343
In addition, the graphs generated have a potential predictive effect on the number of infections that 344 can occur. However, there are still limitations to scenario handling, privacy, data control, and even 345 data security breaches. Countries strive to overcome challenges. Some countries, such as Israel, 346 have enacted an "emergency law for the use of mobile data" to combat the current epidemic (72).

347
Among global call tracking programs, some countries have violated privacy laws and been reported 348 to be insecure (70). So far, they have done a good job by completing the manual tracking process.