An intelligent and secure system for predicting

and preventing Zika virus outbreak using Fog computing

Sanjay Sareen, Sunil Kumar Gupta & Sandeep K. Sood

To cite this article: Sanjay Sareen, Sunil Kumar Gupta & Sandeep K. Sood (2017): An intelligent and secure system for predicting and preventing Zika virus outbreak using Fog computing, Enterprise Information Systems, DOI: 10.1080/17517575.2016.1277558

To link to this article: http://dx.doi.org/10.1080/17517575.2016.1277558

Full Terms & Conditions of access and use can be found at http://www.tandfonline.com/action/journalInformation?journalCode=teis20

An intelligent and secure system for predicting and preventing

Zika virus outbreak using Fog computing

Sanjay Sareena,b, Sunil Kumar Guptac and Sandeep K. Soodd

aComputer Section, Guru Nanak Dev University, Amritsar, Punjab, India; bI. K. Gujral Punjab Technical University, Kapurthala, Punjab, India; cDepartment of Computer Science and Engineering, Beant College of Engineering and Technology, Gurdaspur, Punjab, India; dDepartment of Computer Science and Engineering, Guru Nanak Dev University Regional Campus, Gurdaspur, Punjab, India

ARTICLE HISTORY Received 19 July 2016 Accepted 26 December 2016

KEYWORDS Zika virus; mosquito sensor; Fog computing; fuzzy k-nearest neighbour; Google map

1. Introduction

In the twenty-first century, Zika is the most rapidly spreading infectious virus which is raising new threats to the public health globally. It is transmitted into the human body from an infected Aedes mosquito. It was first found in 1947 in the Zika forest of Uganda from the blood of a Rhesus monkey. In 1948, it was isolated from a lot of A. africanus mosquitoes in the same forest (Dick, Kitchen, and Haddow 1952). It was first found in human in 1952 in Uganda and the United Republic of Tanzania. According to the World Health Organization (WHO), the Zika virus (ZikaV) transmission has been detected in a total of 64 countries and territories from 1 January 2007 to 13 April 2016 (World Health Organization 2016a). In Colombia, 58,838 cases of ZikaV are identified from 22 September 2015 to 19 March 2016 (Sarmiento-Ospina et al. 2016). The incubation period of Zika virus disease (ZVD) varies from 2 to 7 days. The symptoms of ZVD are similar to dengue and chikungunya which include mild fever, rashes, conjunctivitis, arthralgia, arthritis, muscle and joint pain, malaise and headache. The World Health Organization (2016b) has declared the ZikaV an emergency worldwide as it has affected the newborn babies with microcephaly and neurological disorders. Apart from mosquitoes, the ZikaV can be transmitted by (a) sexual intercourse (Musso

ABSTRACT Zika virus is a mosquito-borne disease that spreads very quickly in different parts of the world. In this article, we proposed a system to prevent and control the spread of Zika virus disease using integration of Fog computing, cloud computing, mobile phones and the Internet of things (IoT)-based sensor devices. Fog computing is used as an intermediary layer between the cloud and end users to reduce the latency time and extra communication cost that is usually found high in cloudbased systems. A fuzzy k-nearest neighbour is used to diagnose the possibly infected users, and Google map web service is used to provide the geographic positioning system (GPS)-based risk assessment to prevent the outbreak. It is used to represent each Zika virus (ZikaV)-infected user, mosquito-dense sites and breeding sites on the Google map that help the government healthcare authorities to control such risk-prone areas effectively and efficiently. The proposed system is deployed on Amazon EC2 cloud to evaluate its performance and accuracy using data set for 2 million users. Our system provides high accuracy of 94.5% for initial diagnosis of different users according to their symptoms and appropriate GPS-based risk assessment.

2 S. SAREEN ET AL. CONTACT Sanjay Sareen sareen.gndu@gmail.com Computer Section, Guru Nanak Dev University, Amritsar, Punjab, India © 2017 Informa UK Limited, trading as Taylor & Francis Group et al. 2015), (b) blood transfusions (Marano et al. 2015), and (c) perinatal transmission from mother to foetus during any trimester of pregnancy (Oliveira Melo et al. 2016). The rapid spread of ZikaV results in the rise in the number of ZVD cases and poses challenges for the government healthcare services. No specific antiviral medicine or vaccine is available to cure ZVD. Hence, a public healthcare system based on information and communication technologies are strongly needed which will focus the attention towards prevention of ZVD from spreading.

Nowadays, the remote detection and monitoring of infectious disease outbreaks in real time are strongly needed to control the outbreak. Such diseases are unable to control by existing healthcare systems due to three limitations: (1) use of traditional methods which are reactive, as opposed to proactive disease monitoring and controlling, (2) lack of people involvement in providing vital information about their health and environment conditions to public health agencies, and (3) lack of interactive health education for people to initiate necessary actions. Patients with various infectious diseases need regular monitoring. Sometimes, it is not possible for the patient to visit the hospital for regular check-up or doctors cannot monitor each patient regularly by visiting them. Hence, a remote monitoring system is required to provide ubiquitous healthcare support services using IT infrastructure. To capture, store and process the health data of the patients is a big challenge for any nationwide healthcare systems as it requires huge computing and storage resources.

With the advent of pervasive computing, Internet of things (IoT), mobile computing and cloud computing, it has now become possible to improve the quality of healthcare services by providing users a wide variety of computing services (Sareen, Sood, and Gupta 2016a). Cloud computing provides massively scalable, virtualised IT resource on demand over the Internet with pay only for service-used model (Lounis et al. 2016a). The advent of high-speed 4G network technology and advanced mobile phones has brought the cloud computing paradigm to the mobile domain. Many healthcare applications using mobile communication technology and context-aware technology are migrated onto the cloud platform, which can provide services remotely and at any time (Sareen, Sood, and Gupta 2016b; He, Fan, and Li 2013). However, the IoT-based applications generate unprecedented amounts of data that are difficult for cloud system to process in real time due to high communication overhead. Cloud computing is unable to provide low latency, mobility support and location awareness. Fog computing is designed to address these limitations by providing elastic resources and services to end users at the edge of network (Dastjerdi and Buyya 2016). As Fog computing is implemented at the edge of the network, it provides low latency and location awareness and improves quality-of-services (QoS) for streaming and real-time applications.

A Fog-based infectious disease monitoring system accessed by mobile phones for monitoring and detection of ZikaV-infected patients in real time can be an effective solution. Given the lack of Fogbased infectious disease control system that addresses the limitations in ZikaV outbreak prevention, a Fog-based mobile health system is proposed which stores the personal ZikaV symptoms and information related to the risk-prone area over the Fog servers which are distributed at different locations in a risk-prone area. The confidentiality of sensitive information of users is protected using the combination of information granulation and secret sharing scheme. Initially, each user is registered through a mobile-based application by entering personal and contact details. The system automatically generates a reference number (RNO) and is allocated to each registered user. The RNO is used for all communications between users, health workers and the public healthcare authorities. Fuzzy k-nearest neighbour (FKNN) provides an initial assessment of users which classify them into an infected or uninfected category using their vital symptoms. Once the users are diagnosed, the system will provide interactive alert messages and health education to users so as to aware them about the disease and preventive measures. Geographic positioning system (GPS)-enabled mobile phones carrying by the possibly infected users are

ENTERPRISE INFORMATION SYSTEMS 3 used to track their respective locations. Identification of risk-prone areas and re-routing of users is done using Google map web service. The government will use this information to control the growth of the infection among the infected population. 1.1. Novel contributions

The following is a summary of our novel contributions:

● We designed and implemented a Fog-based model to provide remote monitoring of ZikaVinfected users anytime and anywhere in real time.

● We implemented FKNN-based initial classification to categorise the users as uninfected or infected depending upon their ZikaV symptoms.

● Google map assisted risk assessment, and re-routing is provided with the help of which the user has been re-routed to the safer path in a risk-prone area.

● We also implemented information granulation and secret sharing scheme to protect the confidentiality of sensitive information related to patients.

The remainder of the article is organised as follows. Section 2 reviews related work on ZikaV infection and use of Fog computing for the detection and monitoring of ZVD patients. A model to monitor and detect the ZikaV is proposed in Section 3. In Section 4, we present and analyse the experimental results of our proposed system. Section 5 offers conclusions coming out of this model.

2. Related work

Related work is divided into four sections, which are Zika virus infection, Fog computing and IoT in healthcare, big data streams processing and data privacy and security. First section relates to characteristics, causes and results of ZikaV outbreaks. Second section provides the use of IoT and Fog computing in the field of healthcare services. Third section relates to processing of big data streams generated from sensor devices. Fourth section relates to sharing of sensitive information on the cloud in a secure manner.

2.1. Zika virus infection

The first evidence of human infected with ZikaV was found in 1952. Afterwards, the ZVD was active in several countries in Africa and Asia before transmitted to the Pacific region and more recently to the America. Paixão et al. (2016) reviewed the epidemic spread of ZikaV outbreak across the globe. The authors examined the clinical symptoms and complications arise from this virus. From the study, it has been found that the ZVD outbreak occurred in Brazil, in which 14,835 cases were identified. Among these cases infected with ZikaV, 2.3 per 1000 cases had neurological complications and 1.3 per 1000 cases had Guillain–Barre syndrome. During the ZikaV outbreak, the cases of infant microcephaly born from the ZVD-infected pregnant women have also noticed. Nishiura et al. (2016b) conducted a study to estimate the level of risk of microcephaly among pregnant women infected with ZVD during the outbreak occurred in Northeastern Brazil in 2015. Smet De et al. (2016) reported the first confirmed case of Belgian traveller infected from the ZikaV who came back after spending a 3-week holiday in Guatemala in December 2015. Lazear et al. (2016) proposed a mouse model for understanding the dynamics of ZikaV spread and transmission. Wikan et al. (2016) conducted a study to locate the evidence of ZikaV in Thailand by taking serum samples from patients suffering from acute undifferentiated fever. This study identified the first evidence of ZikaV transmission in Thailand. Petersen et al. (2016) reviewed the clinical symptoms and epidemiology of the ZikaV outbreak in Brazil. The authors also reviewed the

4 S. SAREEN ET AL. impact of the ZikaV outbreak in America on the public health and highlight the urgent need for the preventing the spread of infectious diseases at mass gathering events. Nunes et al. (2016) made another review on the association between ZikaV outbreak in Brazil and microcephaly cases. Chang et al. (2016) conducted a review on clinical manifestation, diagnosis, pathogenesis and treatment of ZVD. Nishiura et al. (2016a) studied the transmission potential of ZikaV infection by estimating the reproduction number R0. Chen and Tang (2016) reviewed the clinical characteristics of ZikaV. Chan et al. (2016) reviewed the clinical features, virology and pathogenesis of ZikaV and congenital Zika syndrome.

2.2. Fog computing and IoT in healthcare

Zheng et al. (2014) reviewed the different sensing and wearable technologies that can be used to develop efficient pervasive healthcare systems. Chen (2017) proposed a novel intelligent value stream-based food traceability cyber physical system approach integrated with enterprise architectures, EPCglobal and value stream mapping method by Fog computing network for traceability collaborative efficiency. In this article, the author assessed how the IoT-enabled cyber physical system concept can enhance the efficiency of food traceability system by means of the highly value-added traceable process using value stream mapping method. Ahmad et al. (2016) proposed a framework of health Fog for sharing and processing health-related information based on data acquired from multiple resources. In this article, the Fog computing is used as an intermediary layer between the cloud and end users to reduce the latency time and extra communication cost. Jalali et al. (2016) proposed a flow-based and time-based energy consumption models for shared and unshared network equipment, respectively. This article has compared the energy consumption of applications using centralised data centres in cloud computing with applications using nano data centres used in Fog computing.

2.3. Big data streams processing

Many Internet-based applications such as IoT-based monitoring systems, decision-making systems, recommendation systems and security generate huge data streams, which are known as big data streams. Such systems require processing such big data streams efficiently and accurately in real time. Xhafa et al. (2015) studied that using a heterogeneous cluster for big data stream processing could indeed incur into streaming inconsistency. They demonstrated the approach using the Yahoo!S4 for processing the big data stream from FlightRadar24 global flight monitoring system. Besides the challenges of processing huge amount of data, the Big Data Stream processing adds further challenges of coping with scalability and high throughput to process the data in real time. Xhafa et al. (2015) proposed frameworks such as Yahoo!S4 and Twitter Storm for big data stream processing. In this article, they implemented and evaluated the Yahoo!S4 for big data stream processing and demonstrated through the big data stream from global flight monitoring system. Maio et al. (2017) proposed a framework that faces with a generalisation of the problem of topic detection and tracking to monitor any kind of scope of interest whose behaviour could be determined by the analysis of the tweet stream. They implemented temporal fuzzy concept analysis that enables us to analyse at the same stage both time relations among tweets and semantic meaning of their content. Maio et al. (2016) proposed a context-sensitive fuzzy-based decision support system capable to use customers’ opinions to automatically tailor well-known correlation among quality of service, customer satisfaction, commitment and loyalty.

2.4. Data privacy and security

The sharing of highly sensitive information on the cloud raises major security and privacy issues that need to address. Wang et al. (2016a) proposed attribute-based data sharing scheme which is more

ENTERPRISE INFORMATION SYSTEMS 5 friendly for the cloud-based systems. In this model, an improved two-party key issuing protocol is applied which can assure that neither key service authority nor CSP can compromise the whole secret key of a user. Wang et al. (2016b) proposed an efficient file hierarchy attribute-based encryption scheme which integrated layered access structures into a single access structure, and then, the hierarchical files are encrypted with the integrated access structure. Lounis et al. (2016b) proposed a cloud-based architecture for medical wireless sensor networks which provides confidentiality and integrity of medical data using efficient access control mechanism. Wu et al. (2016) proposed a novel anonymous authentication scheme for wireless body area networks in order to ensure the security and privacy of data stored in the cloud. Thilakanathan et al. (2014) proposed a model which provides secure and confidential sharing of health data of patients among doctors, healthcare agencies and other authorised users.

3. Proposed model

The proposed system as shown in Figure 1 consists of backend section (cloud storage and processing), middle section (Fog computing) and front-end section (mobile phone and mosquito sensor). The cloud storage and processing section consist of different modules: data collection; information protection; FKNN-based classification; GPS-based risk assessment; and health communication. The mobile phone and mosquito sensor section consists of data acquisition and transmission module. The cloud server is connected to the mobile phone via Fog layer. Figure 2 shows the information and system workflow for preventing ZikaV infections. The following subsections will discuss in details various components of the proposed architecture.

Figure 1. An architecture of the proposed model.

6 S. SAREEN ET AL.

Figure 2. System workflow for preventing ZikaV infections.

3.1. Acquisition and transmission

This module consists of two subcomponents: (1) acquisition and (2) transmission. Initially, each user is registered with the system by entering his personal details through mobile application installed on the user’s mobile phone. Each registered user is provided with an RNO automatically generated by the system. Once the user is registered with the system, the ZVD-related symptoms are collected periodically through user’s mobile phone. The presence or absence of the symptoms of a user is entered in ‘Y’ or ‘N’.

The acquisition module is responsible for collecting personal information and ZikaV symptoms from the users as well as environmental data from the mosquito sensors. The information about sites with high mosquito density and mosquito breeding are captured continuously through wireless mosquito sensors which are placed at different locations. These sensors identify and count the number of mosquitoes in a specific area. It also measures air temperature, humidity and carbon dioxide values around standing water which can be used to evaluate the conditions under which the mosquitoes may lay eggs. The breeding of mosquitoes is increased by higher temperatures, humidity and carbon dioxide values; so these environmental parameters are captured and stored in the database for predicting the breeding sites. As a result, both presences of mosquito-dense sites and breeding sites can be identified using these sensors, and their locations are further transmitted to the cloud server and stored in the database for further processing. In certain regions where sensors are not deployed, the users can also enter the details of mosquito-dense and breeding sites along with images taken by the camera of the mobile phone, if they come across these spots using mobile application. At the time of taking the photograph, the GPS (latitude and longitude) coordinates are also automatically captured by the mobile application and stored along with other details. The doctors, hospitals and government healthcare departments can access and view this information to plan the required preventive actions.

3.2. Fog computing

Fog computing is used to process live big data generated from sensors and mobile phones in real time. Fog computing can act as a bridge between end users and large-scale cloud computing and storage services. Through Fog computing, it is possible to extend cloud computing services to the edge devices of the network. Because of their proximity to the end users compared to the cloud data centres, Fog computing has the potential to offer data, compute, storage and application services with low latency and improves QoS for streaming and real-time applications. The mobile application is used to transmit the collected data to the Fog servers using wireless communication technologies such as WiFi. A Fog server is a generic virtualised equipment with the on-board storage, computing and communication

ENTERPRISE INFORMATION SYSTEMS 7 capability. This allows a Fog server to autonomously and independently serve local computation and data processing requests from mobile users. Bridging the mobile and cloud, a Fog server can be conveniently used to collect the environmental data from the sensors as well as ZikaV symptoms data from the mobile users and transmit the collected data to cloud for in-depth data processing and analysis. Fog computing provides services with low latency, location awareness, and mobility support to process data in real time. A set of Fog servers is distributed within predefined positions in each area under observation. The Fog servers can be expanded to cover a city, countries or even a continent. The Fog server positions are evaluated such that it covers the monitored risk-prone area efficiently. Figure 3 shows the use of Fog computing in ZikaV monitoring system.

Figure 3. Fog computing in ZikaV monitoring system.

3.3. Data collection component

Once the data is collected and processed by the Fog servers, it is transmitted to the cloud for indepth analysis. The data collection component collects and stores personal information and information about ZVD symptoms of different users. Cloud storage provides easy, flexible and secure way to share information among users, doctors, hospitals and the governmental agencies. Multiple health-related attributes are stored as shown in Table 1. These attributes are divided into personal- and ZikaV-related attributes. Personal attributes remain the same for most of the periods, whereas ZikaV attributes can change over time. The location and environmental attributes of mosquito-dense sites and breeding sites are stored as shown in Table 2. The values of these parameters are stored in ZVD database.

Table 1. Personal and symptom attributes of users suffering from ZikaV. S. No Personal attribute Description Symptom attribute Response 1 RNO Reference number of user Fever (Y/N) 2 Name Name of user Skin rashes (Y/N) 3 Age Age of user (in years) Conjunctivitis (Y/N) 4 Gender Male or female (M/F) Joint pain (Y/N) 5 Residential address Permanent address of user Muscle pain (Y/N) 6 Office address Office address of user (if any) Headache (Y/N) 7 Mobile number Mobile number of user Exposure to risk area (Y/N)

8 S. SAREEN ET AL. 8 Family Mobile number of family member

Table 2. Location and environmental attri butes of risk-prone areas.

S. No Attribute Description

1 Mosquito-dense site location GPS location of mosquito-dense site 2 Mosquito density Number of mosquitoes counted by sensor 3 Breeding-site location GPS location of breeding site 4 Temperature Temperature around standing water 5 Humidity Humidity 6 Carbon dioxide (CO2) Value of carbon dioxide 7 Site image Images of mosquito-dense or breeding

sites 3.4. Information protection

The data collection component captures personal, environmental and ZikaV symptom-related information from the user. This information may contain attributes whose associations or attribute itself are sensitive and not required to be shared with everyone. Disclosure of such information to the unauthorised user can cause mass paranoid panic among citizens of any country. The proposed system incorporates information granulation (Yao 2007) and secret sharing mechanism (Sareen, Sood, and Gupta 2016c) in a two-stage process to prevent unauthorised access to data. In the first stage, the initial data table containing all the information is first fragmented into three fragments of different security levels: Level 1 (personal information), level 2 (environmental information) and level 3 (ZikaV information), which is then stored on different secure servers. Level 1 is highly sensitive information containing personal attributes such as RNO, name, age, address and mobile number of the user. Level 2 is the mediocre level of information containing environmental attributes such as mosquito density, humidity and temperature. Level 3 is least sensitive information containing ZikaV attributes and symptoms. Even if anyone could retrieve level 3 information because it is maintained at the least secure system, he/she will not be able to find an exact identity of the user. For retrieving exact identity of the user, a person requires the knowledge of all three levels of fragments. To solve the problem of preserving the personal information of the user, the data should be first preprocessed from data tables to appropriate data fragments. Figure 4 shows the transition of the data table to three fragments.

In the second stage, secret sharing scheme is applied to the highly sensitive attributes of security level 1 that splits the value of these attribute into secret shares. The secret shares generated are stored on different secure servers. Even if anyone could able to retrieve the knowledge of all three levels of fragments, the exact identity of the user cannot be retrieved.

Definition 1: A data table D having attributes ða1; a2;:::; anÞ of N users who have been registered over

the ZikaV mobile application. A mapping function F is used to map the value of each attribute a 2 A to

its specific values va 2 VA such that, F : N ! VA where VA are the all possible values of attribute A.

Definition 2: Initial stored data table D is divided into three fragments of different security levels which

are L1, L2 and L3 such that D ¼ ðN; L1 [ L2 [ L3Þ. These fragments are joined using a unique RNO.

ENTERPRISE INFORMATION SYSTEMS 9

Figure 4. Information granulation-based data table conversion.

Definition 3: A data table D containing information of level 1 may contain sensitive attribute as that should be protected from other users. The threshold secret sharing scheme is applied to sensitive attributes that distribute the values of sensitive attributes into n shares s1; s2;:::; sn so that:

● Knowledge of any i shares, where i n along with secret information, Xðx1; x2;:::; xnÞ is needed to

generate the original value.

● Knowledge of any i 1 or fewer shares is not sufficient to generate the original value even if X is

known to anyone.

3.5. FKNN method

This component provides an initial diagnosis to users by classifying them depending upon their ZVD attributes values, respectively, using FKNN classifier (Keller, Gray, and Givens 1985) as infected and uninfected. The FKNN classifier incorporates the fuzzy set theory into KNN and is used for solving classification problems in a variety of domains with greater accuracy. The FKNN assigns class membership as a function of the vector’s distance from its k-nearest neighbours and those neighbours’ memberships in the possible classes.

In this method, the fuzzy memberships of samples are assigned to different categories using the following equation:

k

P uij1= k x xjk2=ðm1Þ

uiðxÞ ¼ j ¼1 ; (1)

k P1= k x xjk2=ðm1Þ

j¼1

where fx1; x2;:::; xng be the set of n labelled samples and uiðxÞ be the assigned membership of the vector x. uij be the membership degree of the pattern xj from the training set to the class i, among the k-

10 S. SAREEN ET AL. nearest neighbours of x. The variables i = 1,2,. . ., c, and j = 1,2,. . ., k with c number of classes representing two categories uninfected and infected and k number of nearest neighbours. The variable m is used to determine how heavily the distance is weighted when calculating each neighbour’s contribution to the membership value, and its value is usually chosen as m 2 ð1;1Þ. k x xj k is the distance between x and its jth nearest neighbour xj, usually Euclidean distance is chosen as the distance metric. There are two ways to define uij, one way is the crisp membership, i.e. each training pattern has complete membership in their known class and non-memberships in all other classes. The other way is the constrained fuzzy membership, i.e. the k-nearest neighbours of each training pattern (say xk) are found, and the membership of xk in each class is assigned as:

uijðxkÞ ¼ 0ðn:51j=Kþ ðÞ0n:49j=K;Þ0:49; iiff jj¼ii; (2)

The value nj is the number of neighbours found which belong to the jth class. Note that, the memberships calculated by Equation (2) should satisfy the following equations:

c

X uij ¼ 1; j ¼ 1; 2;...:; n;

i¼1

n

0 <X uij < n;

j¼1

uij 2 ½0; 1

where uij ¼ uiðxjÞ for i ¼ 1; 2;:::; c, and j ¼ 1; 2;:::; n is the degree of membership of xk in class i.

The details of the FKNN algorithm are presented in algorithm 1. An algorithm 2 is designed to evaluate the category of the user using FKNN classification algorithm.

Algorithm 1: The FKNN Input The training set X ¼ x1; x2;:::; xn of unknown classification.

Set k; 1 k n;

i = 1;

for i n do

Compute the distance from x to xi using the Euclidean distance;

if i k then

Include xi in the set of k-nearest neighbours; else if xi is

closer to x than any previous nearest neighbours then

Delete the farthest of the k-nearest neighbours;

ENTERPRISE INFORMATION SYSTEMS 11 Include xi in the set of k-nearest neighbours;

end if

i = i + 1; end for

for c = 1 to C do Compute uiðxÞ using Equation (1);

end for

Algorithm 2: Re-evaluate symptoms and category of the patient

Input ZVD symptoms parameters and RNO of a user.

Output Revised category of a user based on symptoms.

Read the primary symptoms data and RNO of the user;

if RNO is already present then Update the database with newly entered data;

else Create a new record with RNO of the patient and store the primary symptoms;

end if Execute FKNN algorithm to predict the revised category of the user;

if revised category = old category then Update the database record;

else Update the category of the patient in the database; Send an alert message to user, doctor and nearby hospital;

end if

3.6. GPS-based risk assessment

ZikaV is primarily spread by mosquitoes which transmit the virus into the human blood through a bite. The virus replicates in the blood and picked by other mosquitoes that bite and inject into other persons. The objective of this component is to provide the information related to the identification of mosquito-dense sites, breeding sites and infected population in real time in order to strengthen the efforts of the public healthcare agencies. Such information might be difficult to identify by the healthcare authorities.

The geographic location of infected users, breeding sites and sites with high mosquito density can be utilised to identify and separate the risk-prone areas. Identification and eradication of mosquito-dense and breeding sites are a vital step to contain the source of ZikaV. Once the risk-prone area is identified, the spread of infection can be controlled by sending alert messages and infection-control suggestions to the people residing in those areas. The infected users are continuously diagnosed and monitored until they are completely recovered from the infection. The Fog server is engaged in continuously capturing the data from users as well as mosquito sensors so that any newly infected user or risk-prone site is automatically identified. Figure 5 shows the use of GPS for identifying the location of infected users and risk-prone areas identified by the proposed system.

Google Maps (2016) web service is used to visualise the spread of infection, high mosquito-dense sites and breeding sites using their GPS locations as shown in Figure 6. The probability of ZikaV infection is very high at the residence and workplace, breeding sites and sites with high mosquito density, so these

12 S. SAREEN ET AL. locations are used to identify risk-prone areas and infected users. The representation of risk-prone areas over Google map helps the government healthcare departments and uninfected users to control the epidemic. The locations of new infected users and sites are automatically detected by the system, and the Google map is updated accordingly. Algorithm 3 is designed to show the exact location of the infected users, breeding sites and mosquito-dense sites which are dynamic and adaptive in nature.

Figure 5. GPS-based location tracking model.

Figure 6. Representation of infected and risk-prone areas using Google map web service.

Algorithm 3: Creation/updation of Google map

Input Location parameters of users, mosquito dense areas and breeding sites.

ENTERPRISE INFORMATION SYSTEMS 13 Output Creation or updation of Google map. Read

the locations address and RNO of the patient; if

RNO is already exists then Update the record with new location data;

else Create a new record with RNO of the user;

end if Read the residential and office address of the user; Read the address of breeding spots and location of high mosquito dense areas; Update the Google map pinpointing these locations;

3.7. Health communication component

This component is used for controlling the spread of ZikaV outbreak, which is one of the important steps in our proposed model. System-generated health education and alert messages related to (a) preventing the growth of mosquitoes and (b) preventing the bites of mosquitoes are send to the infected or uninfected users through SMS or e-mail inboxes in order to improve the user’s health preventive behaviour in Zika-endemic areas. It can be a warning, reminders or any suggestions sent repetitively to improve knowledge of preventive measures among users which may lead to reductions in Zika infestation risk. These informational campaigns motivate the users for the adoption of preventive measures against the disease. Alert messages are also sent to nearby hospitals or healthcare agencies depending upon the GPS location of the patient’s mobile phone. Tables 3 and 4 show the health information for uninfected and infected users, respectively. Table 3. Messages for uninfected users.

S. No Messages 1 Cover the water used for personal consumption which will protect yourself and your family from

the ZikaV infection. 2 Use window screen, bed nets and repellents to prevent yourself from the exposure to mosquitoes. 3 Discard broken water reservoirs and old water bottles from your house. 4 Rush to the nearest hospital if you or a family member has symptoms such as fever, headache, joint

pain, red eyes and a rash. It could be Zika! 5 Keep yourself up to date with the latest information obtained from TV, radio, newspapers and other

social medias such as Facebook and WhatsApp. Table 4. Messages for ZikaV-infected users.

S. No Messages 1 Due to non-availability of any specific vaccine to prevent or drug to treat ZikaV, the patient should

be treated for symptoms. 2 Drink water frequently to prevent dehydration. 3 Take plenty of rest. 4 Take medicine such as aspirin or paracetamol to reduce fever and pain. 5 Consult your doctor before taking such medicines, if you are already taking medicine for another

disease. 6 If you are already infected with ZikaV, prevent yourself from the mosquito bites for the first week of

illness because it can be transmitted to others through infected mosquito bites. 4. Experiment setup and performance analysis

In order to evaluate the performance and effectiveness of our proposed system, different experiments were conducted. The experiment is divided into following segments:

● Synthetic data generation.

14 S. SAREEN ET AL. ● Training and testing of FKNN. ● Testing of the proposed model on Amazon EC2 cloud. ● Risk assessment using Google map. ● Discussion.

4.1. Synthetic data generation

A thorough search has been made on the Internet for the symptom-based data for ZikaV-infected patients for the testing of the proposed system. Since symptom-based data for ZVD patients are not available, synthetic data are generated in such a way that all possible combination of symptoms is considered. In our proposed system, seven different ZikaV attributes are used to diagnose any user for infection. Table 5 shows the sample of eight such possible combinations of attributes.

The probability of occurrence of different combinations of symptoms is different from one another as these symptoms are related to each other. Hence, all combination of different symptoms is categorised into four groups depending on their probability of occurrence, as shown in Table 6. A real dataset of around 2 million people containing personal and demographic information is obtained from census data (AdultDataset 2016). A dataset containing the location of breeding sites as well as mosquito-dense sites are synthetically generated. All possible combinations of ZikaV symptoms are randomly mapped with 2 million user data. Algorithm 4 is used to produce data of 2 million ZikaV users. Table 5. Sample containing different combination of different ZikaV symptoms.

S. No Fever Skin rashes Conjunctivitis Joint pain Muscle pain Headache Exposure to risk-prone area

1 Y Y Y Y N N N 2 Y N Y N Y N N 3 Y N Y Y N N Y 4 Y Y Y Y N N Y 5 Y N N Y Y Y N 6 Y Y N N Y Y Y 7 Y Y Y N N Y Y 8 Y N N N N Y Y

Table 6. Distribution of all possible ZikaV combinations.

Algorithm 4: Generation of 2 million ZikaV users

Input: Data containing all combination of ZikaV symptoms; Output: ZikaV users database

Group Probability range Description Number of cases

A B C D

P 0:35 0:20<P

0:35

0:05<P 0:20 P 0:05

Most possible combination Possible combination Rarely possible combination Very rare possible combination

65 32 20 11

ENTERPRISE INFORMATION SYSTEMS 15 Let n be the number of ZikaV users initialised with 1;

for n required number of users do

Pick a record randomly from the database of ZikaV categories Create a new user by mapping the census record with symptoms data.

Assign a new RNO to the user if RNO is already present in the database then

Discard the record else

Save the new user end if

end for

4.2. Training and testing of FKNN

FKNN classifier is used for the categorisation of users into uninfected and infected categories using the generated synthetic data. The FKNN classifier (Keller, Gray, and Givens 1985) was implemented on an Intel i5 CPU at 2.40 GHz with 2 Gbytes memory using MATLAB running on Windows 7. The dataset is first normalised to transform symptoms values accepted in ‘Y’ and ‘N’ into the interval [0,1] according to strength of each symptom to avoid the numerical difficulties during the calculation. The classification accuracy of FKNN algorithm is tested the fuzzy strength parameter m which varies in the range of (Dick, Kitchen, and Haddow 1952; World Health Organization 2016a) with the step size of 0.01 using different numbers of k. The classification accuracy is then validated via the 10-fold cross validation (CV) analysis on several numbers of neighbours k. For each choice of m, we test the average accuracy obtained by FKNN via 10-fold CV analysis. Finally, the one with the highest average accuracy is selected as the optimal fuzzy strength parameter. An algorithm 5 is designed for the classification procedure.

Algorithm 5: The classification procedure using 10-fold CV

for i = 1 to Mmax do

for j = 1 to k do

Training set = k-1 subsets; Test set = remaining subsets;

Train the model on the training set to find the optimal value of m when the

value of k is set to 1, 3, 5 and 7, respectively;

Test it on the test set and assigns the accuracy to V(j);

end for

Compute the mean value of vector V, and store the mean CV accuracy to the vector M(i); end for

Get the optimal m value whose corresponding mean CV accuracy is the highest in M(i);

for t = 1 to k do

Training set = k-1 subsets; Test set = remaining subsets; Train the model on the training set using the obtained optimal parameter combination; Test it on the test set and save the mean CV accuracy;

end for Return the average classification accuracy rates of FKNN over t test set.

16 S. SAREEN ET AL.

Table 7. Results of classification performance of FKNN using different parameters. Parameters Mean (%) SD (%) Min (%) FKNN1 Accuracy 95.84 0.40 94.75

m = 1.12 Sensitivity 94.35 0.60 94.75 94.85 k = 1 Specificity 94.79 1.55 94.97 95.27 FKNN2 Accuracy 95.90 0.55 94.35 96.78 m = 1.06 Sensitivity 94.45 0.46 93.76 94.95 k = 3 Specificity 96.26 2.10 92.17 98.33 FKNN3 Accuracy 95.72 0.82 93.88 96.86 m = 1.08 Sensitivity 94.50 0.97 94.02 96.25 k = 5 Specificity 94.16 3.27 89.22 98.64 FKNN4 Accuracy 95.97 0.65 94.83 96.92 m = 1.04 Sensitivity 94.48 0.43 93.02 95.21 k = 7 Specificity 95.33 1.74 93.11 97.29

It can be observed that the classification accuracy fluctuates between 90% and 98% with different values of m. It reveals that the fuzzy strength parameter has a big impact on the performance of FKNN classifier. The best classification accuracy was achieved with the parameter pair of (1, 1.12), (3, 1.06), (5, 1.08) and (7, 1.04). These optimal different parameter pairs, namely (1, 1.12), (3, 1.06), (5, 1.08) and (7, 1.04), are used in the subsequent experiments, and for convenience, they are named FKNN1, FKNN2, FKNN3 and FKNN4, respectively. A 10-fold CV is applied to evaluate the performance of the FKNN. The main advantage of using this method is that all of the test sets are independent, and the reliability of the results could be increased. Table 7 summarised the detailed results of classification performance in terms of accuracy, sensitivity and specificity in the form of average accuracy (mean), standard deviation (STD), minimal accuracy (Min) and maximal accuracy (Max).

True positives also known as sensitivity is the percentage of categories of ZVD cases correctly classified by the classifier. False positives) also known specificity is the percentage of ZVD cases wrongly classified by the classifier (Baldi et al. 2000). From Table 7, we can see that the results of the classification performance of FKNN classifiers with different optimal parameter pair are very close. Among them, FKNN4 achieved the highest accuracy of 95.97%, FKNN3 achieved the highest sensitivity of 94.50% and FKNN2 achieved the highest specificity of 96.26%. Hence, the use of FKNN classifier in our proposed architecture is justified.

4.3. Testing of the proposed model on Amazon EC2 cloud

The proposed system was deployed over the cloud to study its performance and effectiveness. For this purpose, 2 million user data are stored on the Amazon EC2 cloud (Amazon, 2014). General purpose m4.xlarge instance having 2.4 GHz Intel Xeon E5-2676 v3 (Haswell) processors and 16 GiB memory are used to set up an application over the cloud. Different classification algorithms such as neural network (Hagan, Demuth, and Beale 1996), multiplayer perceptron (Yan et al. 2006), linear regression (Ko and Barkana 2014) and naive Bayesian network (John and Langley 1995) are also tested in Weka 3.6 (Hall et al. 2009) to compare their performance with FKNNs. Initially, the system was started with 20,000 requests, and then after each 5-min request, the system was increased by 20,000 and system performance was studied for a total experiment time of 100 min. Figure 7 represents the comparison of different classification algorithms for various statistical measures using a different number of users. The accuracy of classification algorithms is shown in Figure 7(a) which shows that the FKNN-based classification algorithm performs better than all other algorithms. Figure 7(b) shows the classification time of different algorithms using different dataset. Results show that the FKNN algorithm takes less

ENTERPRISE INFORMATION SYSTEMS 17 time to classify the user data. Figure 7(c) shows different classification measures: sensitivity, precision, recall and F-measure. The proposed algorithm produces better results in all statistical parameters.

4.4. Risk assessment using Google map

To evaluate the GPS-based risk assessment, data related to infected users, breeding sites and mosquito-dense sites are generated over the Amritsar city in India. Three files in the .csv format containing the details of 5000 users, breeding sites and mosquito-dense sites are fed to the Google map using the Google API in JavaScript. Figure 8(a) shows the location of possible infected users, breeding sites and mosquito-dense sites. The figure also shows the routing of a user from location A that is D-322, Defence Colony, D-Block, Ranjit Avenue, Amritsar, Punjab 143001, India, to location B that is D-584, Hotel HK Intercontinental, Defence Colony, D-Block, Ranjit Avenue, Amritsar, Punjab 143001, India, without implementing any re-routing algorithm. Here, the user passes through the mosquito breeding sites and mosquito-dense sites which increase the risk of catching the ZikaV infection. Alternatively, the user has diverted to the safer path by using an appropriate re-routing algorithm as shown in Figure 8(b). The blue line shows the normal and proposed route by Google services.

4.5. Discussion

In this section, we compare the results of different classification methods obtained in (Hagan, Demuth, and Beale 1996) – (John and Langley 1995) with the results of our proposed FKNN classifier on the current dataset of 200,000 users. Neural network classifier (Hagan, Demuth, and Beale 1996) is used that produced sensitivity of 84.0% and specificity of 85.0%. Multiplayer perceptron classifier (Yan et al. 2006) produced sensitivity of 74.9% and specificity of 79.3%. Another linear regression classifier (Ko and Barkana 2014) produced sensitivity of 67.7% and specificity of 68.7%. Naive Bayesian network (John and Langley 1995) produced sensitivity of

18 S. SAREEN ET AL.

(C) Classification Algorithms

Classification Parameters

Figure 7. Performance analysis of different algorithms over the Amazon EC2: (a) classification accuracy, (b) classification time and (c) classification measures.

89.2% and specificity of 89.5%. The proposed FKNNs classifier, when applied on dataset, had resulted in sensitivity of 94.50% and specificity of 96.26%. From these comparisons, we conclude that the proposed

ENTERPRISE INFORMATION SYSTEMS 19 FKNN classifier presented in this study yield more accurate results. Table 8 represents the comparison of different classification algorithms for various statistical measures.

Figure 8. GPS-based re-routing of user from location A that is D-322, Defence Colony, D-Block, Ranjit Avenue, Amritsar, Punjab 143001, India, to location B that is D-584, Hotel HK Intercontinental, Defence Colony, D-Block, Ranjit Avenue, Amritsar, Punjab 143001, India; (a) default routing of user using Google services and (b) safe routing of user based on infected areas.

Table 8. Evaluation of proposed methodology with other related methodologies. State-of-the-art Methods Accuracy (%) Sensitivity (%) Specificity

(%) Hagan et al. (Hagan, Demuth, and Beale 1996) NN 83.10 84.20 85.00 Yan et al. (Yan et al. 2006) MPP 77.23 74.92 79.31 Koc et al. (Ko and Barkana 2014) LR 65.35 67.74 68.72 John and Langley (John and Langley 1995) NBN 89.10 89.20 89.50 Proposed methodology FKNN 95.97 94.50 96.26

N: neural network; MPP: multiplayer perceptron; LR: linear regression; NBN: naive Bayesian network; FKNN: Fuzzy k-nearest neighbour.

5. Conclusion

Mosquito is one of a fatal insect that spreads various human pathogens such as malaria, dengue and ZikaV. It is prevalent in different parts of the world and is affecting more and more countries. In this article, we presented a cloud-based monitoring system for a crowd of patients by using a mobile device, wireless sensor technology and Fog computing. The aim of this system is not only to predict and prevent the ZikaV outbreak but also to provide a solution to improve the health of patients. The proposed system is designed to handle a huge number of patients simultaneously by collecting their vital signs and environmental data and transmitting the data to a dedicated Fog server. A set of Fog servers is distributed in a risk-prone area which will handle the patient’s big data in real time by reducing the latency time and communication cost. FKNN is used to diagnose the users as infected or uninfected based on their symptoms, and cloud computing is used for effective information analysis and sharing. Mosquito sensors are deployed in different parts of the risk-prone sites to obtain the information about mosquito-dense sites and environmental parameters to discover the breeding sites. Geographic positioning system (GPS) is used to display ZikaV-infected users, mosquito-dense sites and breeding sites on Google map. Using Google map, re-routing to uninfected users is provided that helps the users to protect themselves from the risk-prone areas. The sensitive information of the patients is protected using information granulation and secret sharing scheme. We have implemented the proposed system on Amazon EC2 cloud to evaluate its performance. Our proposed system provides 94.5 % accuracy in

20 S. SAREEN ET AL. classification and accurate identification of risk-prone areas. It will help the government healthcare departments to control the mosquito population more effectively.

We can summarise some of the benefits and advantages that can be reaped from our system as follows:

● Improve patients’ quality of life. ● Early detection of ZikaV infection associated with patients and hence minimising the spread of

infection. ● Improvements in the diagnosis process, treatment and future decisions concerning the patients’

health. ● Support a large number of patients irrespective of their geographic locations. ● Assists healthcare agencies and hospitals to promptly intervene in the case of an emergency

situation and control the spread of ZikaV infection. ● Help physicians and clinicians in making their work more efficient and their decisions more

reasonable and accurate.

Disclosure statement

No potential conflict of interest was reported by the authors.

References AdultDataset. 2016. https://archive.ics.uci.edu/ml/datasets/Adult. Last Accessed on 5th May (2016). Ahmad, M., M. B. Amin, S. Hussain, B. H. Kang, T. Cheong, and S. Lee. 2016. “Health Fog: A Novel Framework for Health and

Wellness Applications.” The Journal of Supercomputing. doi:10.1007/s11227-016-1634-x. Amazon, A. W. S. Amazon EC2 Instance Types, 2014 [Online]. Available http://aws.amazon.com/ec2/instance-types/

(accessed 19.03.16). Baldi, P., S. Brunak, Y. Chauvin, C. A. Andersen, and H. Nielsen. 2000. “Assessing the Accuracy of Prediction Algorithms for

Classification: An Overview.” Bioinformatics (Oxford, England) 16 (5): 412–424. doi:10.1093/bioinformatics/ 16.5.412. Chan, J. F. W., G. K. Y. Choi, C. C. Y. Yip, V. C. C. Cheng, and K.-Y. Yuen. 2016. “Zika Fever and Congenital Zika Syndrome: An

Unexpected Emerging Arboviral Disease.” Journal of Infection 72: 507–524. doi:10.1016/j.jinf.2016.02.011. Chang, C., K. Ortiz, A. Ansari, and M. E. Gershwin. 2016. “The Zika Outbreak of the 21st Century.” Journal of Autoimmunity

68: 1–13. doi:10.1016/j.jaut.2016.02.006. Chen, H.-L., and R.-B. Tang. 2016. “Why Zika Virus Infection Has Become a Public Health Concern?” Journal of the Chinese

Medical Association 79: 174–178. doi:10.1016/j.jcma.2016.03.001. Chen, R.-Y. 2017. “An Intelligent Value Stream-Based Approach to Collaboration of Food Traceability Cyber Physical System

by Fog Computing.” Food Control 71: 124–136. doi:10.1016/j.foodcont.2016.06.042. Dastjerdi, A. V., and R. Buyya. 2016. “Fog Computing: Helping the Internet of Things Realize Its Potential.” Computer 49 (8):

112–116. doi:10.1109/MC.2016.245. De Smet, B., D. Van den Bossche, C. van de Werve, J. Mairesse, J. Schmidt-Chanasit, J. Michiels, K. K. Arien, M. Van Esbroeck,

and L. Cnops. 2016. “Confirmed Zika Virus Infection in a Belgian Traveler Returning from Guatemala, and the Diagnostic Challenges of Imported Cases into Europe.” Journal of Clinical Virology. doi:10.1016/j.jcv.2016.04.009.

Dick, G. W. A., S. F. Kitchen, and A. J. Haddow. 1952. “Zika Virus (I). Isolations and Serological Specificity.” Transactions of the Royal Society of Tropical Medicine and Hygiene 46 (5): 509–520. doi:10.1016/0035-9203(52)90042-4.

Google Maps. 2016. https://www.google.co.in/maps. Last Accessed on 7th May 2016. Hagan, M. T., H. B. Demuth, and M. H. Beale. 1996. Neural Network Design. Boston: Pws Pub. Hall, M., E. Frank, G. Holmes, B. Pfahringer, P. Reutemann, and I. H. Witten. 2009. “The WEKA Data Mining Software: An

Update.” ACM SIGKDD Explorations Newsletter 11 (1): 10–18. doi:10.1145/1656274. He, C., X. Fan, and Y. Li. 2013. “Toward Ubiquitous Healthcare Services with a Novel Efficient Cloud Platform.” IEEE

Transactions On Biomedical Engineering 60 (1): 230–234. doi:10.1109/TBME.2012.2222404. Jalali, F., K. Hinton, R. Ayre, T. Alpcan, and R. S. Tucker. 2016. “Fog Computing May Help to Save Energy in Cloud

Computing.” IEEE Journal on Selected Areas in Communications 34 (5): 1728–1739. doi:10.1109/JSAC.2016.2545559.

ENTERPRISE INFORMATION SYSTEMS 21 John, G. H., and P. Langley. 1995. Estimating Continuous Distributions in Bayesian Classifiers. San Francisco, CA: Morgan

Kaufmann Publishers Inc. Keller, J. M., M. R. Gray, and J. A. Givens. 1985. “A Fuzzy K-Nearest Neighbor Algorithm.” IEEE Transactions on Systems, Man,

and Cybernetics 15: 580–585. doi:10.1109/TSMC.1985.6313426. Ko, M., and A. Barkana. 2014. “Application of Linear Regression Classification to Low-Dimensional Datasets.”

Neurocomputing 131: 331–335. doi:10.1016/j.neucom.2013.10.009. Lazear, H. M., J. Govero, A. M. Smith, D. J. Platt, E. Fernandez, J. J. Miner, and M. S. Diamond. 2016. “A Mouse Model of Zika

Virus Pathogenesis.” Cell Host & Microbe 19: 720–730. doi:10.1016/j.chom.2016.03.010. Lounis, A., A. Hadjidj, A. Bouabdallah, and Y. Challal. 2016a. “Healing on the Cloud: Secure Cloud Architecture for Medical

Wireless Sensor Networks.” Future Generation Computer Systems 55: 266–277. doi:10.1016/j. future.2015.01.009. Lounis, A., A. Hadjidj, A. Bouabdallah, and Y. Challal. 2016b. “Healing on the Cloud: Secure Cloud Architecture for Medical

Wireless Sensor Networks.” Future Generation Computer Systems 55: 266–277. doi:10.1016/j. future.2015.01.009. Maio, C. D., G. Fenza, V. Loia, and F. Orciuoli. 2017. “Unfolding Social Content Evolution along Time and Semantics.” Future

Generation Computer Systems 66: 146–159. doi:10.1016/j.future.2016.05.039. Maio, C. D., A. Tommasetti, O. Troisi, M. Vesci, G. Fenza, and V. Loia. 2016. “Contextual Fuzzy-Based Decision Support System

through Opinion Analysis: A Case Study at University of the Salerno.” International Journal of Information Technology & Decision Making 15 (5): 923–948. doi:10.1142/S0219622016500231.

Marano, G., S. Pupella, S. Vaglio, G. M. Liumbruno, and G. Grazzini. 2015. “Zika Virus and the Never-Ending Story of Emerging Pathogens and Transfusion Medicine.” Blood Transfusion = Trasfusione Del Sangue 14 (2): 95–100. doi:10.2450/2015.0066-15.

Musso, D., C. Roche, E. Robin, T. Nhan, A. Teissier, and V.-M. Cao-Lormeau. 2015. “Potential Sexual Transmission of Zika Virus.” Emerging Infectious Diseases 21: 359–361. doi:10.3201/eid2102.141363.

Nishiura, H., R. Kinoshita, K. Mizumoto, Y. Yasuda, and K. Nah. 2016a. “Transmission Potential of Zika Virus Infection in the South Pacific.” International Journal of Infectious Diseases 45: 95–97. doi:10.1016/j.ijid.2016.02.017.

Nishiura, H., K. Mizumoto, K. S. Rock, Y. Yasuda, R. Kinoshita, and Y. Miyamatsu. 2016b. “A Theoretical Estimate of the Risk of Microcephaly during Pregnancy with Zika Virus Infection.” Epidemics 15: 66–70. doi:10.1016/j. epidem.2016.03.001.

Nunes, M. L., C. R. Carlini, D. Marinowic, F. K. Neto, H. H. Fiori, M. C. Scotta, P. L. Á. Zanella, R. B. Soder, and J. C. D. Costa. 2016. “Microcephaly and Zika Virus: A Clinical and Epidemiological Analysis of the Current Outbreak in Brazil.” Jornal De Pediatria 92 (3): 230–240. doi:10.1016/j.jped.2016.02.009.

Oliveira Melo, A. S., G. Malinger, R. Ximenes, P. O. Szejnfeld, S. Alves Sampaio, and M. Bispo de Filippis. 2016. “Zika Virus Intrauterine Infection Causes Fetal Brain Abnormality and Microcephaly: Tip of the Iceberg ?” Ultrasound in Obstetrics & Gynecology 47: 6–7. doi:10.1002/uog.15831.

Paixão, E. S., F. Barreto, M. D. G. Teixeira, M. D. C. N. Costa, and L. C. Rodrigues. 2016. “History, Epidemiology, and Clinical Manifestations of Zika: A Systematic Review.” American Journal of Public Health 106 (4): 606–612. doi:10.2105/AJPH.2016.303112.

Petersen, E., M. E. Wilson, S. Touch, B. McCloskey, P. Mwabae, M. Bates, O. Dar, et al. 2016. “Rapid Spread of Zika Virus in the Americas – Implications for Public Health Preparedness for Mass Gatherings at the 2016 Brazil Olympic Games.” International Journal of Infectious Diseases 44: 11–15. doi:10.1016/j.ijid.2016.02.001.

Sareen, S., S. K. Sood, and S. K. Gupta. 2016a. “An Automatic Prediction of Epileptic Seizures Using Cloud Computing and Wireless Sensor Networks.” Journal of Medical Systems 40: 226. doi:10.1007/s10916-016-0579-1.

Sareen, S., S. K. Sood, and S. K. Gupta. 2016b. “A Cloud-Based Seizure Alert System for Epileptic Patients that Uses Higher-Order Statistics.” IEEE Computing in Science & Engineering 18 (5): 56–67. doi:10.1109/MCSE.2016.82.

Sareen, S., S. K. Sood, and S. K. Gupta. 2016c. “Towards the Design of a Secure Data Outsourcing Using Fragmentation and Secret Sharing Scheme.” Information Security Journal: A Global Perspective 25 (1–3): 39–53. doi:10.1080/ 19393555.2015.1134732.

Sarmiento-Ospina, A., H. Vásquez-Serna, C. E. Jimenez-Canizales, W. E. Villamil-Gómez, and A. J. Rodriguez-Morales. 2016. “Zika Virus Associated Deaths in Colombia.” The Lancet Infectious Diseases 16 (5): 523–524. doi:10.1016/S14733099(16)30006-8.

Thilakanathan, D., S. Chen, S. Nepal, R. Calvo, and L. Alem. 2014. “A Platform for Secure Monitoring and Sharing of Generic Health Data in the Cloud.” Future Generation Computer Systems 35: 102–113. doi:10.1016/j. future.2013.09.011.

Wang, S., K. Liang, J. K. Liu, J. Chen, J. Yu, and W. Xie. 2016a. “Attribute-Based Data Sharing Scheme Revisited in Cloud Computing.” IEEE Transactions on Information Forensics and Security 11 (8): 1661–1673. doi:10.1109/ TIFS.2016.2549004.

22 S. SAREEN ET AL. Wang, S., J. Zhou, J. K. Liu, J. Chen, J. Yu, and W. Xie. 2016b. “An Efficient File Hierarchy Attribute-Based Encryption Scheme

in Cloud Computing.” IEEE Transactions on Information Forensics and Security 11 (6): 1265–1277. doi:10.1109/TIFS.2016.2523941.

Wikan, N., Y. Suputtamongkol, S. Yoksan, D. R. Smith, and P. Auewarakul. 2016. “Immunological Evidence of Zika Virus Transmission in Thailand.” Asian Pacific Journal of Tropical Medicine 9 (2): 141–144. doi:10.1016/j.apjtm.2016.01.017.

World Health Organization, Zika Situation Report, 2016a[Online]. http://www.who.int/emergencies/zika-virus/situa tion-report/14-april-2016/en/[accessed 19.04.2016].

World Health Organization, Zika Outbreak: WHO’s Global Emergency Response Plan, 2016b [Online]. http://www.who. int/emergencies/zika-virus/response/en/[accessed 20.04.2016].

Wu, L., Y. Zhang, L. Li, and J. Shen. 2016. “Efficient and Anonymous Authentication Scheme for Wireless Body Area Networks.” Journal of Medical Systems 40: 134. doi:10.1007/s10916-016-0491-8.

Xhafa, F., V. Naranjo, L. Barolli, and M. Takizawa (2015). On Streaming Consistency of Big Data Stream Processing in Heterogenous Clutsers. In Proc. of 18th international conference on Network-Based Information Systems (NBiS), 476– 482. doi: 10.1109/NBiS.2015.122.

Xhafa, F., V. Naranjo, and S. Caball (2015). Processing and Analytics of Big Data Streams with Yahoo!S4. In Proc. of IEEE 29th International Conference on Advanced Information Networking, 263–270. doi: 10.1109/AINA.2015.194.

Yan, H., Y. Jiang, J. Zheng, C. Peng, and Q. Li. 2006. “A Multilayer Perceptron-Based Medical Decision Support System for Heart Disease Diagnosis.” Expert Systems with Applications 30 (2): 272–281. doi:10.1016/j.eswa.2005.07.022.

Yao, Y. 2007. “The Art of Granular Computing.” Lecture Notes in Computer Science (Including Subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) 4585 LNAI: 101–112.

Zheng, Y.-L., X.-R. Ding, C. C. Y. Poon, B. P. L. Lo, H. Zhang, X.-L. Zhou, G.-Z. Yang, N. Zhao, and Y.-T. Zhang. 2014. “Unobtrusive Sensing and Wearable Devices for Health Informatics.” IEEE Transactions on Biomedical Engineering 61 (5): 1538–1554. doi:10.1109/TBME.2014.2309951.