Keyword: SARS-CoV-2; COVID-19; Machine Learning; Pandemic
Introduction
Coronaviruses are the common pathogens responsible for infections like
cough and cold next to rhinovirus. Recently the family added its seventh
generation coronavirus SARS-CoV-2, framed to be pandemic by World Health
Organization (WHO) having a close to 97 % similarity to SARS-CoV
[1][2]. Unlike, Severe Acute Respiratory Syndrome (SARS) and
Middle East Respiratory Syndrome (MERS) epidemic outbreak in 2003 and
2012, SARC-CoV-2 also known as COVID-19 mutated to transmit from animal
to human. This virus is assumed to be transferred to humans from bats
sold at a meat market in Wuhan, China. [3] The virus is coined
deadly because of the ease of spreadability infecting around 552,598
people in 197 Countries with an exponential growth in two months leading
to global shutdowns. The number of deaths marked 33,553 as on
29th February 2020 (WHO, 2020). The mortality rate is
found to be very low, with 3% of the infected, when compared to earlier
generation coronaviruses. MERS and SARS had a 30% and 10% death rate
respectively. Howereve, the transmissibility of the COVID-19 is severe
considering the way it easily spreads and people being passive carriers
as well [5]. There are few medicines suggested to control the
effects, but no vaccine has been found till date for coronavirus family,
including COVID-19 [6] [7]. This is infact the largest
single-strand RNA virus known to the humankind as where all the other
viruses have a single protein spike that gets attached to the human
cell, this coronavirus family hold 10 to 12 spike proteins which
increases its ease of attachement to the ACE-2 protein [8]. The
virus follows an unusual double step replication mechanism thereby
growing at an uncontrollable rate [9]. Though the infection has been
contained in China, there are a lot of countries at the early stage of
the outbreak making them vunnerable to the pandemic. As the incubation
period is noted to be 2 to 14 days, the infected person will not have
serious symptoms, rather showing commen flu like symptoms including
fever, dry cough and breathlessness [10]. As there is no vaccine
found to this virus till date, it is very much necessary to control the
transmissible rate by alternative means [11]. Quantitative COVID-19
pandemic impact analysis is very scarce in literature and has to be
considered seriously given the seriousness of the infection. Epidemics
are assumed to have an exponential growth at an early stage and the
number of infections over time goes down due to the interventions taken
by the government of countries. Mathematical modelling incorporating
various precaution measure taken during the viral outbreak using
exponential growth analysis coupled with machine learning will give a
better prediction model to control COVID-19 transmission [12–14].
Policy changes in pandemic and epidemic situations involve social
distancing, lockdowns, travel restrictions, awareness campaigns etc. It
is also proven in past research that environmental conditions of
countries like temperature and humidity also sometimes play a
significant role in controlling pandemics [15].
The objective of the research is twofold an involves data collected from
42 countries. First, it seeks to understand the countries that show an
early sign of containment of the COVID-19 virus. Secondly, the research
aims at building supervised machine learning models with high accuracies
for predicting signs of early continment with infrastructure
availability, environmental factors, infection seviority factos, and
government policies of countries as independent variables. This report
will also involve a discussion on other activities taken by the
government of various nations as an attempt to flatten the infections
curve and their corresponding effectiveness.
Theory
The COVID-19 origin was linked to Wuhan, China’s animal meat market and
it is assumed to be transmitted from bats. This slowly spread all across
the world from humans to humans through fluids and aerosol particles. In
the initial stages, all diagnozed cases outside china had a travel
history to the Wuhan market. However, at later stage, in countries like
Itally, US, UK, Korea, Japan etc, community transfers of the disease
began infecting people exponentially over time. The nature of the
SARS-COV-2 virus bearing the ability to double replicate with the spike
protein challenges researchers in vaccines development [3]. Few
researchers claim Hydroxychloroquine and azithromycin can be used to
treat COVID-19 but not much clinical trials were made to confirm the
claim [7]. Hence, until a date arises when a complete cure for the
virus is announces, the fate of nations now lie in the hands of
epidemiologists to predict spreading patters so that policy makers can
take appropriate measures to contain the infection. Several viruses
including SARS have reported to be vunnerable to hot temperatures making
spreading patterns different based on geographical locations [16].
However, this has not been tested for the COVID-19. Other factors like
government policies and interventions, infrastructure availability, and
the serveiority of the infection itself can affect the ability of a
country to contain epidemics and pandemics. This research seeks to
explore all the above factors.
Social Distancing
Social distancing, though a new terminology for the
21st Century, is an older process used by United
Kingdom during 1912 to control the Influenza Virus outbreak that caused
about a 100 million casualties. Social distancing involves avoiding mass
gathering and isolation from an unaffected person with a minimum of 6
feet and is generally combined with enhanced personal hygiene through
regular hand wash, and wearing a protective mask for flu like outbreaks
[17][18]. This is done primarily because flu causing viruses are
generally spread through aerosol that can be transmitted through saliva
and nasal fluid at a distance of 3 feet. The average lifetime of
COVID-19 viruses in the outer environment is predicted to be 12 hours
which makes the transmissibility even higher as aerosol of infected
persons on doorknobs, lifts, transports, hotels, malls etc. can widen
the outbreak. As it is a communicable disease, it transfers more through
greeting the people by shaking hands. Reducing social contact is proven
to significantly reduce flu like diseases [19]. The closure of
schools and malls flattened the infection during the influenza pandemic
in 2009 [20][21]. Governments hence for the COVID-19 case,
stress on social distancing and quarantining measures for a period of
atleast 14 days to contain the spread of the virus as it is its
incubation period [22][23].
Lockdowns
Lockdown is a preventive strategy taken by local or central or global
administration during the spread of epidemic or pandemic diseases by
closing transportation between cities or provinces or counties. Pandemic
is when the spread of the disease crosses countries and international
borders rather than only a local region or nearby country. The world has
so far seen four pandemics namely plague in the 14thcentury, Infulenza in 1912, SARS in 2009, and the current COVID-19 in
2019 as reported by WHO. [24]The Guardian 2020)[26]. In all
these cases, lockdowns were employed by various countries to control the
outbreaks. China announce lockdown in the past few months to flatten the
exponential curve of the COVID-19 infections over time. Most of the
countries around the globe went on a lockdown of local transport,
office, industries, city and national borders following china to contain
the virus [27]. Though quarantine centers for the infected are
available in hospitals, when the infections are huge and uncontrollable
spreading happens, the only way to contain infections will be a lockdown
where people are self-quarantined at the comfort of their homes
[28].
Environment
The environmental conditions of countries like temperature and humidity
play an inevitable role in both airborne and aerosol virus
transmissions. The 30 year human relationship with the influenza virus
has proven that the mortality rate is directly related to the
temperature and humidity [29]. Hence, inorder to minimize
transmission of diseases, isolation wards in hospitals generally tend to
have optimized pressure, temperature, and humidity (WHO, 2014.). The
reproduction number R0 was found to be between 2.06 to
2.52 for COVID-19. However, research on the virus in the Diamond Cruise
Ship off the coast of Japan has proved that a one degree rise in
temperature and a one percent increase in pressure can bring the value
down to 0.0383 to 0.0224 though the validity of the study is
questionable as the ship was a contained environment [31]. However,
the effect of viruses once it enters the human body is not influenced by
climatic changes. Since the virus lives outside the human body for a
period of atleast 12 hours on normal situations (Richard, 2020), it
becomes necessary to study the environmental effects on the spreading
patterns itself.
Impact of Health care Infrastructure on transmission
During epidemic and pandemic viral outbreaks health care infrastructure
such as availability of hospitals, beds, doctors, clinical equipments,
first aid kits, ventilators, and protective equipments plays a vital
role in flattening the outburst curve. [33][34]. During the
massive influenza outbreak, even developed countries felt the inadequacy
in health care infrastructure, which further expands the outbreak
(George 2008). The ebola outbreak in South Africa also experienced
uncontrollable infections due to lack of infrastructure facilities
[36]. After the outbreak, WHO in South Africa had asked the
hospitals to report their available facilties to plan for future
infections optimally [37]. Certain researches focus on innovative
measures to create necessary healthcase infrastructure during pandemic
and epidemic situations by converting school, college, theatre, stadium
as hospitals [38][39]. Sometimes, health care workers supported
by NGOs, youth, and volunteers also play a significant role in
containing outbreaks (QH Plan, 2018)[41]. Hence studying health care
infrastructure availability across countries can predict COVID-19
containment at an early stage.
Predictive Modelling
Understanding spreading patterns and predictive modelling for policy
implementations and actions by the government especially when vaccines
to control outbreaks are non existant ( Brooks-pollock, & Thompson,
2019). During the onset of any epidemic, it is crucial to use
exponential growth models to understand the infection rates and with
proper policy implementations and behavioural changes in the susceptible
people, over time the slope reduces and the curve flattens [12]. For
various other outbreaks like small pox, ebola, SARS, and influenza,
mathematical modelling was used to understand the growth [46]
[47,48]. Infact, the Center for Disease Control has an exclusive
book for analysing disease outbreaks stressing more on the importance of
the forthmentioned [49]. In outbreaks, epidemiologists generally use
the exponential growth model on the onset of an outbreak and proceed to
prediction and classification techniques like regression, decision
trees, neural networks deepl learning, etc. to forecast outbreaks.
[13,50]. COVID-19 being a very new outbreak, has limited information
and researches on modelling and predicting containment [15,23] and
this research seeks to integrate crucial variables concerning
infrastructure, environment, policies, and sevearity of the disease to
predice initial signs of containment using machine learing and
exponential growth model. The variables doctors per 1000 population,
beds per 1000 population, average temperature , average humidity, days
since official lockdown, percentage of lockdown days, total cases per
million population, deaths per million population, days since first
contact, and percentage of serious cases of infected were used as a part
of the predictive model.