mobile cloud computing framework for patients ... - World Scientific

3 downloads 7674 Views 449KB Size Report
Feb 21, 2014 - MOBILE CLOUD COMPUTING FRAMEWORK. FOR PATIENTS' HEALTH DATA ANALYSIS. Mohammed A. Al-Zoube*,‡ and Yazan A. Alqudah.
Biomedical Engineering: Applications, Basis and Communications, Vol. 26, No. 2 (2014) 1450020 (9 pages) DOI: 10.4015/S1016237214500203

MOBILE CLOUD COMPUTING FRAMEWORK FOR PATIENTS' HEALTH DATA ANALYSIS Mohammed A. Al-Zoube*,‡ and Yazan A. Alqudah†,§ *Department

of Computer Graphics Princess Sumaya University for Technology, Amman, Jordan † Department of Communication Engineering Princess Sumaya University for Technology, Amman, Jordan ‡ [email protected] Accepted 2 April 2013 Published 21 February 2014

ABSTRACT The advent of cloud computing and the ubiquity of broadband wireless coverage and wide spread usage of smart phones around the world carries the potential for transforming health care services, reducing health care cost and ensuring faster care for urgent cases. To these objectives, we present a cloud-based mobile health monitoring solution that takes advantage of cloud infrastructure and mobile processing capability to address the rising cost of health care monitoring. The solution enables health care providers to remotely analyze, monitor and diagnose patient's data. The solution integrates a powerful data analysis tool, cloud computing and mobile services. This paper presents a proof of concept that has been developed to monitor, record and analyze heart rate. The design enables a physician to develop custom analysis and monitoring to collect key indicator or set alerts without a need for infrastructure implementations to store or transfer the data. Keywords: Component; Telemedicine; Vital signs; Cloud computing; Mobile health.

INTRODUCTION Governments and health care organizations are challenged by the rising cost of health care. Both in developed and undeveloped countries, health care expenditure constitute a large portion of the country's gross domestic product. As a result of high health care, many citizens are not able to a®ord medical treatment in a timely manner. To address this challenge, technological advancements are sought to provide means of reducing the cost of health care system. One of the areas that received large interest by industry and health care provides is remote and distant health care. With the ubiquity of

internet access and growth in mobile devices capabilities connectivity, it is possible to develop tools that can assist physicians and care providers that enable them to remotely monitor patients and provide alerts when intervention is needed. Many medical conditions require constant or frequency monitoring of vital signs such as heart rate, body temperature and blood pressure to asses body conditions. Through advancement of mobile devices, internet connectivity and cloud storage, we propose a mobile cloudbased monitoring of health indicators. In this work, we adopt heart sounds as an indicator that can be monitored and analyzed remotely to assess patients conditions.

§

Corresponding author: Yazan A. Alqudah, Department of Communication Engineering, Princess Sumaya University for Technology, Amman, Jordan. E-mail: [email protected]

1450020-1

M. A. Al-Zoube & Y. A. Alqudah

A stethoscope is used for auscultation, listening to cardiac and respiratory sounds. Physicians can interpret the sound to determine the presence of problems. One of the main potentials of mobile health is remote monitoring and data collection.1 Several research works have considered vital signs monitoring using mobile applications. These applications utilize body sensors and the mobile device with Internet access to display and transmit vital signs such as heart rate,2 ECG signal,3 and temperature for patients who require continuous monitoring and have di±culty in attending health centers. However, to design such application, the following challenges should be considered:

run on a PC platform) and used by engineers and scientists to analyze data. The solution takes advantage of data connectivity infrastructure and does not require additional hardware or software for the exchange of data. The solution uses the cloud to setup patient data storage and transmission that can have a signi¯cant impact on healthcare institutions by reducing IT infrastructures.

.

In the last few years cloud computing emerged as a new technology for delivering computing resources such as development platforms, storage and services, that enables the user to access these resources over the Internet.4 For optimum resource utilization, these resources are available on-demand to external customers and dynamically con¯gured to adjust to variable load. Compared with conventional computing, this new computing paradigm has several advantages. For example, it moves the responsibility to install and maintain hardware and software from the customer to the cloud service provider. Cloud computing is ideal for projects that require periodic computational changes, rapid prototyping or fast shift time. This eliminates the responsibility by users and enables payment per usage. Furthermore, the cloud is an attractive alternative to the limitations imposed by a local computing environment such as long job queues, unsupported software or limited server resources. However, despite the advantages of cloud computing there are still concerns with respect to the security and privacy. Therefore, it is very important for the cloud service provider to address these concerns by preventing unauthorized users to access data and safeguard them, and to have a backup and disaster policies. Cloud computing can be classi¯ed based on the services delivery model. Typically three service types have been universally accepted which are software as service (SaaS), platform as a service (PaaS) and infrastructure as a service (IaaS).5 These diverse delivery models enable cloud to o®er di®erent services for satisfying different requirement of customers. For example, large Internet companies such as Google and Microsoft are heavily invested in clouds to support their own operations. Amazon provides virtual servers on its cloud (EC2), along with several other services which provides underlying physical resource as a service (IaaS), and users are allowed full control from the kernel to the entire software stack.6 Microsoft provides management

Mobile limitations: Usually health data analysis is a time-consuming process, requiring high processing capabilities and large storage. Furthermore, the diversity of mobile devices makes the deployment of complex mobile applications rather di±cult. Although the latest mobile devices use high-speed processors, power consumption is still a problem. Therefore, alternative approaches such as cloud computing should be used to o®load power consuming computations to external resources. . Communication: E±cient and secure communication between the mobile users (patients and physicians), and the computing servers (middleware and cloud) is an essential part of health applications. . Scalability and cost: In order to provide e±cient health monitoring application which should be used by a large number of patients and physicians, located in di®erent places, and performs analysis of the patients' data, it is essential to have an architecture that has the ability to easily expand its resources while maintaining performance and low cost. The traditional client/ server and Web services applications models are currently the most widespread application architecture. However, they are di±cult to implement and requires costly installation of software. This is also true for distributed applications. Other approaches, such as cloud-based architecture becomes more convenient. In this work, we present a solution that addresses di®erent aspects of patients' health data collection, analysis and monitoring. The solution delivers an integrated telemedicine service that automates the process of data collection to information distribution and remote access by medical sta®. The physician can customize and con¯gure the analysis based on the patient's age, conditions, etc. The system may be used by a variable number of users located in di®erent locations. The solution integrates Matlabr , a powerful analysis tool (traditionally

MATERIALS AND METHODS Background Cloud computing and healthcare

1450020-2

Mobile Cloud Computing Framework for Patients' Health Data Analysis

and .NET programming capabilities through Azure (PaaS).7 Google o®ers several cloud services such as Google Doc and Spreadsheets (SaaS) and App Engine (PaaS).8 In order to remain e±cient, healthcare requires continuous cost e®ective and systematic innovation to provide high-quality services. Cloud computing, therefore, is an ideal choice to improve health care services and to reduce startup expenses, such as hardware, software, networking, personnel and licensing fees.9 Cloud computing can help healthcare organizations share information stored across disparate information systems in real time and can free up IT sta® to attend to more critical tasks in an e±cient and cost-e®ective manner. For instance, Harvard Medical School took the bene¯ts of cloud computing and built a platform to enable collaborative research between several departments and partners.10 The United States Department of Health & Human Services has selected Acumen Solutions for Health Information Technology to provide a cloud computing customer relationship management and project management solution.11

Dropbox Personal cloud storage is a kind of service which provides virtual online repository that can be used for storing information. It is an easy and convenient way of storing ¯les that will be accessible over the Internet, and performing distributed computation using them. One of the most popular cloud storage services is Dropbox.12 Dropbox can be seen as a ¯le management and transformation service, where ¯les shared among di®erent clients will have the same content. It allows users to access and update their ¯les from any location using any computer, as well as mobile devices such as iPhone and Android. One of the most interesting features of Dropbox is the simpli¯ed ability to share folders easily with other Dropbox users. To do so, a new folder is created and then using the \Share folder" option to select other users and send them an invitation e-mail. Once they accept, they will be able to access, edit and update the ¯les in the shared folder. Dropbox is scalable, paid only when needed and it is free from 2 GB up to 8 GB of storage. Dropbox architecture comprises of two main components, the server and the client. The server comprises data storage and control servers. Amazon Elastic Compute Cloud (EC2) and Simple Storage Service (S3) are used as storage servers. Amazon stores data over several large-scale data centers, where data automatically replicated.12 Files are divided into several pieces of

data with size of up to 4 MB, which is treated independently. To reduce the amount of transmitted data, each piece is identi¯ed by a SHA256 hash value and is compressed using delta encoding, before transmitting them. Dropbox also keeps a database of meta-data information in each device. HTTPS is used to access all services, except the noti¯cation service which runs over HTTP. When the Dropbox client is installed on a PC or mobile, it creates its own special folder on the device. Any ¯le that is copied or moved into this folder is then automatically replicated into the folders in other computers owned by the same person running the Dropbox client. Dropbox client exchanges control information with control servers such as noti¯cation, meta-data administration and system-log servers which are managed by Dropbox Inc. System-log servers collect run-time information about the clients, including exception backtraces and other event logs possibly useful for system optimization. The Dropbox client keeps a TCP connection open to a noti¯cation server (notifyX.dropbox. com), used for receiving information about changes performed elsewhere. Contrarily to other tra±c, noti¯cation connections are not encrypted. A noti¯cation request is sent by the client asking for eventual changes. In case of no change, the client immediately sends a new request. Changes on the central storage are instead advertised as soon as they are performed.13 Dropbox also provides developers with Application Programming Interface (API) which allows them to build applications with robust and easy way to read and write ¯les securely across di®erent platforms (Windows, Mac, Linux, iPhone, iPad, BlackBerry and Android). Developers will also have access to powerful features such as simple sharing, search and restoring ¯les to past revisions. For example it was shown in Ref. 14, by performing some experiments using laptop computers and the free service level of the Dropbox ¯le-storage and sharing system, that these kind of cloud storage systems can be used pro¯tably for distributed evolutionary algorithms, without the need to acquire or set up complicated cloud or grid infrastructure.

Heart sound The heart sounds are a result of blood °ow of the beating heart. Two sounds are generated by the movement of the heart. The ¯rst sound (S1) is a result of ventricular contraction. The second sound (S2) is produced by relaxation of ventricles. A third sound (S3) can occur after S2. This sound is lower in pitch than S1 and S2. The sound is caused by oscillation of blood back and forth between the walls of ventricles.

1450020-3

M. A. Al-Zoube & Y. A. Alqudah

Fig. 1 (Color online) ECG and acoustic cardiac signals.

The activities of the heart can be measured using ECG. The ECG trace is composed of Q, R and S waves to form the QRS complex. These waves are associated with contraction of ventricles due ventricular depolarization. The ECG and acoustical signals are illustrated in Fig. 1.

analyze patients' health data without having to go through the details of the application. Recently, a framework for cloud-based ECG monitoring and analysis was presented in Ref. 18. In this framework, a web-based application which includes ECG analysis services is developed and contained in a Tomcat server. The ¯nal application (Tomcat server and the web application) was deployed on Amazon cloud using the Amazon web service and the Amazonrelated database service. There are also other mobile cloud-based frameworks with similar objectives,19,20 however, the distinction between them lies in two things: the ¯rst is the algorithms developed to analyze the ECG signal, and the second is the type of cloud computing model used in the framework. For example, in Refs. 15 and 17 the cloud computing models used are PaaS (Azure and Aneka) and IaaS (Amazon EC2), while in Refs. 18–20 only the IaaS cloud service model was used.

Related Work There are several research works to improve healthcare services through the collaboration of cloud computing and mobile computing. For example, in Ref. 15 the authors proposed a cloud-based framework to analyze ECG signals. The framework uses the mobile device to acquire the patient's ECG signal, and then transmit it to a cloud service for processing and analysis. If the cloud service ¯nished the job of massive computing of signal processing, the results were then automatically received by the mobile devices. The framework was built using Microsoft Window Azure platform (PaaS). Azure allows users to develop .NET applications and to create the software as a service (SaaS) in the cloud. A SaaS layer was developed to process the ECG signals which can be used by both computers and mobile devices using web browser. Security mechanism was also designed to protect the ECG data in the cloud, and ECG data were transmitted through secure sockets layer (SSL), where ECG ¯les were protected by certi¯cate-based encryption and veri¯cation instead of plain text based HTTP. Another real-time health monitoring and analysis framework integrating cloud and mobile technologies is proposed in Ref. 16. The framework is based on Aneka cloud computing platform.17 Similar to Azure, Aneka is PaaS which is a workload distribution and management platform that hosts applications in Microsoft.NET. A SaaS layer that contains the tools for conducting custom designed analysis of ECG data was hosted as a web service inside a Tomcat container. Using this layer any client-side implementation can simply process and

Proposed Solution Architecture The main objectives of the proposed solution are collection, transmission and processing of patients' health data. Solutions based on manual health data acquisition are slow and labor intensive. The solution is built based on cloud services which provide convenient, on-demand network access to a shared pool of con¯gurable computing resources. These resources can easily be deployed with minimal management e®ort or service provider interaction. This centralized architecture of data, applications and computational power may be accessed and used by any mobile device with an Internet connection. Figure 2 shows the architecture of the proposed system. The solution is based on mobile agents and cloud

Fig. 2 Architecture of the proposed mobile cloud healthcare system.

1450020-4

Mobile Cloud Computing Framework for Patients' Health Data Analysis

replacing the air piece with a tiny microphone that ¯ts inside.21 The other end of the microphone is an audio jack that can be plugged into any PC or laptop with audio interface. The conversion of the audio signal to a digital form is achieved by using the functionalities on the PC that enables recoding of an audio signal.

The cloud computing component

Fig. 3 Components of the proposed system.

computing that provides adaptive, °exible and simple access to computing resources. Below we present the architecture and components to deliver an integrated telemedicine service that automates the process of data collection to information distribution and remote access by medical sta®. The overall proposed new infrastructure consists of four main components as shown in Fig. 3. Below we describe these components.

Data aquisition and patient mobile cloud agents This component is responsible for data collection (e.g. ECG, body temperature, pulse, etc.) and then transmitting this data from the sensor to the mobile devices and then to the cloud for further analysis, storage and processing. When the data becomes available in the cloud, it can be distributed to medical sta® for processing and analysis. Any medical measurement device capable of interfacing to a mobile or PC audio or can provide a digital output can be used. In this work, we adopt a digital stethoscope that is built by transforming the ear piece of an ordinary stethoscope to a microphone capable of recording the audio signal to PC. The stethoscope is then used to record the sound of the heart and the breathing of a patient. In order to integrate the stethoscope in our system, the readings are converted to a digital form that can be sampled, stored and processed. This is achieved by

This component is based on Dropbox's personal cloud ¯le-storage and sharing services. Personal cloud storage allows for remote access to a subscriber ¯les from anywhere with an Internet connection. Dropbox services consist of two main components: the servers and client. The servers are used as storage service (IaaS). On the other hand if a Dropbox client application is installed on a client's computer or mobile phone, it will allow user to store ¯les on Dropbox's servers by copying those ¯les to a designated folder on the client computer. The ¯les are then uploaded to the servers automatically in the background, and those ¯les can be accessed from any other computer or mobile device with Internet access. Dropbox also o®ers sharing capabilities that allows choosing who can access those remotely stored ¯les. Dropbox client application is used to transfer the patient's health data to a middleware data analysis server located at the hospital. Once the data is collected and stored in the designated patient's mobile folder, Dropbox automatically runs the service in the background which transmit the data to the Dropbox servers and then to the hospital server for further analysis. As mentioned in Sec. 2, before Dropbox transmit data, it encrypts and compresses data and transmits it in secure layer. We believe that this considered an e±cient utilization for the new Communication as a Service (CaaS) cloud computing model.

Intelligent middleware data analysis server This component consists of a server running Matlab and possibly other application for data indexing and management. Matlab is considered as one of the most used platforms for information processing and analysis because of the huge number of implemented algorithms in di®erent scienti¯c ¯elds. If these algorithms are accessed over the Internet, then from the cloud computing perspective these are considered as SaaS. Matlab also allows developers to implement their own algorithms, and hence, PaaS if it is accessed over the Internet. To be able to record and analyze the audio recording of patients and other patients' health data, a number of

1450020-5

M. A. Al-Zoube & Y. A. Alqudah

functions and libraries that implement various signal processing algorithms are available.

Physician mobile agent A Java-based application or Matlab Mobile is installed on the physician's mobile device that communicates with the middleware server to initiate data analysis. Matlabr introduced Matlabr Mobile in 2010. The tool enables command line access to Matlabr , as well as an ability to view graphs and ¯gures on iPhones. The tool is run by installing connector software that runs on the iPhone. If the PC running Matlabr and the iPhone are operating on the same network they can connect by simply entering the internet protocol (IP) address of the PC. In case the iPhone is on a di®erent network, then a virtual private network (VPN) is used to securely access the PC workspace.

Data Analysis Once the patient completes recording his/her heart sound audio and the ¯le is uploaded to the Dropbox shared folders, the physician is noti¯ed of the presence of the recording through email. The physician can now connect to the Matlab server in order to analyze the data. Since Matlab is intended for use behind a ¯rewall on a private network, if the mobile is used over a public or cellular data network, a VPN is needed. Once connected, Matlab analysis and visualization capabilities can be used to analyze the patients' data. The physicians can choose to run their analysis or run existing scripts. It should be emphasized that the algorithms discussed here are intended to be developed by physicians and clinical experts. The results below attempt to illustrate the ease by which algorithms can be developed and results are visualized. In this prototype, the physician speci¯es the audio ¯le to be analyzed by entering their patient's ID and running a script as shown in Fig. 4.

RESULTS In this section, we describe a prototype implementation that demonstrates system capabilities.

System Setup A patient that chooses to subscribe to remote monitoring is set up for a cloud storage account that can be accessed by the health provider. In this prototype, a Dropbox folder is created for the patient and is given a name that corresponds to the patient's ID. The health provider is responsible for giving physicians access to the patient's folder by granting them read-access. Once a physician accesses the Dropbox folder, he or she is able to view the patient's folder. The system is proposed for its ability to provide physicians access to a patient's data irrespective of the patient and physician locations. In this prototype, we assume the patient is at home. To demonstrate the sharing of the patient's data, we adopt heart sounds recording, since it lends itself to demonstrating the system capabilities. The patient is trained to record his/her heart sounds during a hospital visit. On a schedule set by the physician, the patient records his/her heart sounds and stores the audio signal in his/her Dropbox patient folder. In this prototype, a regular stethoscope is converted to enable digital recording of the audio by attaching a microphone piece to the tubing of the stethoscope. The patient can then record heart sounds using a PC or mobile phone.

Fig. 4 Analysis is run by user specifying patient ID and executing GenerateAnalysis.m.

1450020-6

Mobile Cloud Computing Framework for Patients' Health Data Analysis

The script fetches the audio ¯le based on the patient idea and uses the audio ¯le to identify systole and diastole timings. The script follows the following procedure: (a) Iterate over signal length to ¯nd peaks separated by at least Ws1 apart, where Ws1 is selected to exceed diastol time and to be less than expected normal systol and diastol time combined. (b) After the peaks are identi¯ed, they are designated as S1 times. The algorithm search for S2 by seaching for peaks in a window starts at S1 ðiÞ þ Ws2 and ends at S1 ði þ 1Þ  Ws2 , where i is the index of ith S1 time. Ws2 is selected to be smaller than the expected systol time. When more than one peak is found, the algorithm selects the time with maximum peak. Upon executing the script, Matlab generates ¯gures of the analysis as shown in Fig. 5. The ¯gures include a display of raw data and locations of the S1 and S2 times. The user can click on any image to interact with the graph. A magni¯ed display of S1 and S2 is shown in Fig. 6.

Fig. 6 Peak locations, S1 , S2 are identi¯ed by triangles facing upward and downward, respectively.

The time di®erence between adjacent peaks is used to calculate the timing of the heart signal is found. These times refer to s1  s2 (Systole) and s2  s1 (Diastole) timing. The display in Fig. 7 shows the histogram of these parameters. It should be emphasized that users can choose from available analysis or by generating their own custom

Fig. 5 Figures generated upon executing Matlab script.

Fig. 7

1450020-7

Histogram of S1 , S2 and S1  S2 times.

M. A. Al-Zoube & Y. A. Alqudah

analysis. Matlab provides a wealth of pre-de¯ned functions and tools that help analyze and visualize the data.

DISCUSSION We have presented a new mobile cloud-based framework for patients' health data analysis. Compared to those solutions mentioned above, the main advantage of the proposed solution is that we have utilized personal cloud storage, namely Dropbox, infrastructure for data communication between the mobile devices and the cloud, and for patients' data storage. To the best of our knowledge this is the ¯rst framework which adopts the CaaS cloud service model (beside other cloud models). This utilization has the following bene¯ts: .

Security and e±ciency of data transmission is handled by Dropbox, alleviating the need to design security methods to safeguard the medical data. Dropbox uses SSL for data transfer and ¯les stored on Dropbox are encrypted using the AES-256 standard. . Lower cost, because Dropbox provides every subscriber up to 8 GB of storage for free. . Ability of distributed analysis of the patients' data using the ¯le sharing property provided by Dropbox. Furthermore, the proposed solution has the following practical advantages: .

Provides a remote and real-time collecting and transmission of patients' data without the need for experts' assistance. . Supports °exible and scalable deployment process in large healthcare environments and the integration of di®erent institutions. . Minimizes the cost of computing resources usage, i.e. the application runs on low-cost computing devices. . Facilitates the development of analysis and visualization software. The proposed architecture can be visualized as a cloud where sensors connected to medical devices are plugged in to collect and transmit data; computer resources available in this cloud are con¯gured to receive, store, analyze and distribute the information. The use of Matlab as a PaaS simpli¯es building the SaaS layer based on the available powerful analysis and visualization tools. The proposed system also provides functionalities that include vital sign measurements, transmission and storage of data, analysis and visualization and indicator

generations. These functionalities are provided to the physicians to enable monitoring vital signs. The physician can generate a customized alert based on a patient's conditions. The analysis of acoustical signals has been studied with the objective of developing a frame work for automating heart sound analysis.5–12 In this work, we adopt the algorithm in Ref. 22 to display and analyze signals.

CONCLUSION In this paper, we presented a new solution for the remote monitoring of a patient's vital signs. The main feature of the proposed solution is the utilization of a free personal cloud storage service (Dropbox) for reliable and secure transmission and storage of the patient's vital signs. Dropbox infrastructure comprising of Dropbox client, server, storage, with communication and sharing capabilities provides a simple and convenient way of building distributed system for collecting and analyzing patients' medical data and storing them securely in the cloud. An added feature of Dropbox is the initial free storage provided to every subscriber, and hence, scalability of the system marginally a®ects the cost of the cloud services. The proposed solution utilizes available signal processing libraries provided by Matlabr . This enables access to patient's data from a smart phone with no new infrastructure or set up. Therefore, the user can alter parameters of the analysis, view graphs and interact with plots. The solution was demonstrated by building a prototype that records heart signal. This enables a PC to record and store a stethoscope's audio signals. A Matlabr program is written to demonstrate the analysis of audio recording through access to cloud. Finally, access by the mobile to Matlabr workspace is presented. The physician can analyze data based on initial parameters and can tune parameters based on patient conditions. A typical presentation of patients' data analysis is presented.

REFERENCES 1. Vital Wave Consulting, mHealth for Development: The opportunity of mobile technology for healthcare in the developing world, United Nations Foundation, Vodafone Foundation, Available at http://www.vitalwaveconsulting.com/pdf/mHealth.pdf, p. 9, 2009. 2. Mattila J, Ding H, Mattila E, Sarela A, Mobile tools for home-based cardiac rehabilitation based on heart rate and

1450020-8

Mobile Cloud Computing Framework for Patients' Health Data Analysis

3.

4. 5.

6. 7. 8. 9.

10.

11. 12. 13.

movement activity analysis, Engineering in Medicine and Biology Society, 2009. EMBC 2009. Annual Int Conf IEEE, pp. 6448–6452, IEEE, 2009. Wang F, Syeda-Mahmood T, Beymer D, Finding disease similarity by combining ECG with heart auscultation sound, Comput Cardio 34:261–264, 2007. Al-zoube M, E-learning on the cloud, Int Arab J of e-Technol 1(2):58–64, 2009. Rimal BP, Choi E, Lumb I, A taxonomy and survey of cloud computing systems, INC, IMS and IDC, 2009. NCM'09. Fifth Int Joint Conf IEEE, pp. 44–51, 2009. Amazon Web Services (AWS), Available at http://aws. amazon.com/. Windows Azure Platform, Available at www.windowsazure.com/en-us/. Google App Engine (GAE), Available at http://code.google.com/appengine/. Kuo AMH, Opportunities and challenges of cloud computing to improve health care services, J Med Internet Res 13(3):e67, 2011. Amazon Web Services, AWS case study: Harvard Medical School, Available at http://aws.amazon.com/solutions/ case-studies/harvard/, 2011. Acumen Solutions, Available at http://www.acumensolutions.com/. Dropbox, Available at http://www.dropbox.com/. Drago I, Mellia M, Munafo M, Sperotto A, Sadre R, Pras A, Inside dropbox: Understanding personal cloud storage services, IMC'12, ser. IMC'12, pp. 481–494, ACM, New York, NY, USA, 2012.

14. Arenas MG, Guervos JM, Castillo PA, Laredo JLJ, Romero G, Mora AM, Using free cloud storage services for distributed evolutionary algorithms, in Krasnogor N, Lanzi PL (eds.), GECCO'11, pp. 1603–1610, ACM, 2011. 15. Hsieh J, Hsu M, A cloud computing based 12-lead ECG telemedicine service, BMC Med Inform Decis Mak 12:77, 2012. 16. Pandey S, Voorsluys W, Niu S, Khandoker A, Buyya R, An autonomic cloud environment for hosting ECG data analysis services, Future Gener Comput Syst 28(1):147– 154, 2012. 17. Vecchiola C, Chu X, Buyya R, Aneka: A Software Platform for .NET-Based Cloud Computing in High Speed and Large Scale Scienti¯c Computing, IOS Press, pp. 267–295, 2009. 18. Henian X, Asif I, Zhao X, Cloud-ECG for real time ECG monitoring and analysis, Comput Methods Programs Biomed 110(3):253–259, 2013. 19. Miao F, Miao X, Shangguan W, Li Y, MobiHealthcare system: Body sensor network based M-Health system for healthcare application, E-Health Telecommunication Syst Netw 1(1):12–18, 2012, doi:10.4236/etsn.2012.11003. 20. Bourouis A, Feham M, Bouchachia A, Ubiquitous mobile health monitoring system for elderly (UMHMSE), Int J Comput Sci Inform Technol (IJCSIT) 3(3):74–82, 2011. 21. Myint W, Dillard B, An electronic stethoscope with diagnosis capability, Proc. 33rd IEEE Southeastern Symp System Theory, pp. 133–137, 2001. 22. Yazan A, Vital signs remote monitoring and analysis: Seamless integration with a smart phone, Biomed Eng: Appl Basis Commun 25(04):1350003, 2013.

1450020-9