Skip to main content
PMC home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2014 Feb 19.
Published in final edited form as: J Med Entomol. 2013 Jul;50(4):879–889. doi: 10.1603/me13008

Cell Phone-Based System (Chaak) for Surveillance of Immatures of Dengue Virus Mosquito Vectors

SAUL LOZANO–FUENTES 1,2, FADI WEDYAN 3, EDGAR HERNANDEZ–GARCIA 4, DEVADATTA SADHU 1, SUDIPTO GHOSH 1, JAMES M BIEMAN 1, DIANA TEP-CHEL 5, JULIÁN E GARCÍA–REJÓN 4, LARS EISEN 1
PMCID: PMC3929104  NIHMSID: NIHMS551680  PMID: 23926788

Abstract

Capture of surveillance data on mobile devices and rapid transfer of such data from these devices into an electronic database or data management and decision support systems promote timely data analyses and public health response during disease outbreaks. Mobile data capture is used increasingly for malaria surveillance and holds great promise for surveillance of other neglected tropical diseases. We focused on mosquito-borne dengue, with the primary aims of: 1) developing and field-testing a cell phone-based system (called Chaak) for capture of data relating to the surveillance of the mosquito immature stages, and 2) assessing, in the dengue endemic setting of Mérida, México, the cost-effectiveness of this new technology versus paper-based data collection. Chaak includes a desktop component, where a manager selects premises to be surveyed for mosquito immatures, and a cell phone component, where the surveyor receives the assigned tasks and captures the data. Data collected on the cell phone can be transferred to a central database through different modes of transmission, including near-real time where data are transferred immediately (e.g., over the Internet) or by first storing data on the cell phone for future transmission. Spatial data are handled in a novel, semantically driven, geographic information system. Compared with a pen-and-paper-based method, use of Chaak improved the accuracy and increased the speed of data transcription into an electronic database. The cost-effectiveness of using the Chaak system will depend largely on the up-front cost of purchasing cell phones and the recurring cost of data transfer over a cellular network.

Keywords: Aedes aegypti, dengue, surveillance, informatics, cell phone

Effective control of neglected tropical diseases with explosive outbreak dynamics, such as dengue (Kan et al. 2008, Vazquez–Prokopec et al. 2010, Jeefoo et al. 2011), is dependent upon timely analysis of collected surveillance data and rapid response. Dengue virus is transmitted by Aedes mosquitoes, principally Aedes aegypti L., and maintained in a human–mosquito cycle (Halstead 2008). Because a commercially available vaccine against dengue virus is still lacking, control of the mosquito vector is the primary means to prevent and control dengue outbreaks (WHO 2009). Surveillance for Ae. aegypti, which exploits artificial containers as larval development sites (Focks and Alexander 2006), commonly focuses on the immature larval and pupal life stages, for which a series of surveillance indices have been developed to estimate abundance (Focks 2004, WHO 2009). These indices can be used to guide the need for control efforts and to evaluate the impact of a vector control intervention. Delays in the transfer of mosquito surveillance data from paper-based field collection records to an electronic database can prevent the timely completion of data analyses. This, in turn, can result in slow and ineffective mosquito control response.

Another critical issue, noted recently in a systematic review of vector control services for dengue, is the lack of standards and information technology tools for monitoring and evaluation (Horstick et al. 2010). We previously addressed the need for improved tools for data management and analysis through the development of a novel disease data management system/decision support system for surveillance and control programs, with current capacity for dengue and malaria (Eisen et al. 2011, Lozano–Fuentes et al. 2011). However, the value added by using such a system still depends, to a large extent, on timely transfer of data into the system. Most dengue-endemic settings rely heavily on paper-based field data collection, including data related to the surveillance and control of dengue virus mosquito vectors. Transcribing large amounts of data from paper-based field records into a computer-based system is a time-consuming, tedious, and potentially error-prone undertaking, thus potentially preventing timely data analysis and vector control response. In the city of Mérida, México, vector control currently employs field workers to visit premises and collect data on mosquitoes on paper forms. The forms are collected, summarized, and then transcribed into computers at a later time with the risk of lost forms, calculation errors, or transcription errors. Using the Chaak system described herein, data instead are captured in digital form immediately, and transferred into a database using the fastest available mechanism. Uploading data can be done by direct connection or by transmission of data over wireless networks (Wi-Fi or cell phone networks). In an ideal scenario, the data-capturing device also has the capacity to act as a Global Positioning System receiver and thus define the spatial locations where the data were entered (Vanden Eng et al. 2007).

Mobile capture of electronic data is a rapidly evolving information technology with great potential for incorporation into a data management system and decision support system to achieve timely transfer of field-collected data into the system. With the aid of mobile computers (e.g., laptops or netbooks), remote sensors, or even cell phones, electronic data capture can now be achieved during the initial data-capturing session in the field or the laboratory. Mobile computing has been improved with the advent of faster, cheaper low-power processors and more robust wireless data transmission technologies. The technical capabilities of mobile devices are constantly improving and mobile Internet access is now achieved in most large urban areas of the world.

There is strong interest in using mobile devices for data capture, data transfer, or both, in public health, including the surveillance and control of neglected tropical diseases (Kaplan 2006, Safaie et al. 2006, Tatem et al. 2009, Tomlinson et al. 2009, Blaya et al. 2010, Kahn et al. 2010, Adedeji et al. 2011, Lang 2011, The malERA Consultative Group on Diagnosis and Diagnostics 2011, Matthews et al. 2012). Recent studies have evaluated the use of personal data assistants (PDAs) for data collection during household surveys for malaria or bed net use (Vanden Eng et al. 2007, Ahmed and Zerihun 2010), and collection of data for suspected dengue patients in clinical studies (Aviles et al. 2008). The potential for use of cell phones as mobile data-capturing devices was evaluated for infectious disease surveillance in Africa and South America (Johnson and Blazes 2007, Randrianasolo et al. 2010), and for malaria surveillance and monitoring in Africa and Southeast Asia (Mueller et al. 2009, Meankaew et al. 2010). Moreover, a recent study (Matthews et al. 2012) suggests that cell phones potentially can be used for rapid diagnosis of dengue in conjunction with a medical test patch for which the presence of dengue virus biomarkers result in color changes. An image of the test patch captured by the cell phone’s optical sensor is analyzed via a colorometric-based algorithm that can be run on the cell phone, and the result can then be transmitted over a wireless network.

The primary aims of this study were to 1) develop and field-test a novel system, Chaak, for cell phone-based capture of data relating to the surveillance of the immature stages of dengue virus mosquito vectors, and 2) assess, in the dengue endemic setting of Mérida, México, the cost of using this new technology versus paper-based data collection followed by manual entry into an electronic database. We reported previously on the evaluation of the system performance, with regards to time needed to complete data entry per premises and the rate of data entry error (Lozano-Fuentes et al. 2012). In this article, we provide an extended description of the system and present additional information for the cost of using the system, based on the time needed per premises to complete a survey, the projected salaries for surveyors, and the local cost for data transfer over the cellular network.

Materials and Methods

System Development Strategy

Projected users of the Chaak system include a system administrator, management personnel for vector surveillance and control, and surveyors (field workers). System administrators and management personnel use a desktop thin client and surveyors use cell phones to enter data. The system was developed in an iterative process over a 2-yr period, with close contact between the software developers, at Colorado State University, Fort Collins, CO, and Universidad Autónoma de Yucatán, Yucatán, México, and subject matter experts from Servicios de Salud de Yucatán, Yucatán, México. Furthermore, the system was developed based on the workflow and requirements of Servicios de Salud de Yucatán, and it complies with national Mexican guidelines for conducting entomological surveys (Hernandez Avila 2011).

System Architecture and Software Components

The software has a four tier architecture (Fig. 1). The top tier is the application tier consisting of thin clients on desktops running Microsoft Windows .NET (Windows.Net 2012) and cell phones running Android (Android 2012). We used a variety of Android phones running unmodified Android versions 2.1 and higher: Google Nexus 1, S, and Galaxy Nexus (Galaxy Nexus 2012); HTC Desire and HTC Legend (HTC 2012); and Motorola Backflip (Motorola 2012). The Android and thin client applications communicate with the business tier using web services in the business access tier. The Web services are hosted on an Apache Webserver (Apache–Tomcat 2012) with module mono installed (Mono 2012). The business tier supports data validation, decisions, evaluations, calculations, and data CRUD (create, read, update, and delete) operations. In the bottom tier (the data tier) storage is achieved with a PostgreSQL server (PostgreSQL 2012), and a PostGIS plugin (PostGIS 2012) to manage spatial data, such as city blocks or individual premises. A basic geographic information system, based on Lozano–Fuentes et al. (2008), was extended to allow the system to manage spatial data and create maps, and to capture spatial data in the field using an Android device. Reports are generated in HTML and optionally converted to PDF with wkhtmltopdf program (Truelsen 2012).

Fig. 1.

Fig. 1

Open in a new tab

System architecture.

System Requirements and Licensing

As previously stated, the system is made of several pieces of software. Some of these pieces run on a desktop thin client, whereas others run on a database-web services server or a cell phone. The server application requires the Ubuntu 10.04 Server Edition (32 bit) operating system. The desktop thin client was made to be run on Microsoft Windows XP, but it has been tested and used in Windows 7 (32 bit) as well. The server application hardware requirements are a minimum of 1 GB RAM, 100 GB hard drive, and Intel Atom 1.6 GHz (model N270) to operate on a stand-alone machine. The thin client application can run on the basic requirements for Windows XP plus a network interface card. The cell phone-based application of the system was developed for Android 2.1 (unmodified) but was also tested on Google Nexus 1, S, and Galaxy; HTC Desire and Legend; and Motorola Backflip cell phones. Each application has a dedicated installer for fast and easy deployment of the system.

The source code, installable packages, and documentation can be found at: http://www.sourceforge.net/projects/chaak/. The system and the accompanying user manual are available in English and Spanish language versions. The system has been released under a General Public License version 3 (Free Software Foundation 2012), which permits commercialization, enhancements, or customized versions of the software, as well as the incorporation of the system or pieces of it into other software packages as long as the source code is kept open. There is no cost associated with the license. The license, and the way the software is designed, will allow researchers to modify the source code to share modifications with colleagues or submit changes to the central repository. The original and future updated versions, including contributions such as extensions, compatible modules or plugins of the software, will be made freely available for download from the previously mentioned URL.

Field-Testing of the System

The performance and use cost of the cell phone-based Chaak system was compared, in the dengue endemic setting of Mérida, México, with pen-and-paper-based field data capture followed by manual data entry into an electronic database. The evaluation of system performance, with regards to time needed for data entry and error rate on data entry, was reported previously (Lozano–Fuentes et al. 2012) but is summarized briefly in this article for completeness of the description of the system. In this article, we also present new information for the cost of using the system, based on the time needed to complete a survey, the projected salaries for surveyors, and the local cost for data transfer over the mobile network.

The time needed for data entry was assessed under field conditions in 13 randomly selected Mérida neighborhoods. We randomly selected two blocks from each neighborhood, one of which was assigned for Chaak system surveys (three surveyors) and the other one for pen-and-paper-based surveys (three surveyors). Entomological surveys were only conducted on premises for which the residents granted permission for the surveyors to examine containers for presence of mosquito immatures. In summary, the pen-and-paper-based method had a performance time, including subsequent entry of data into an electronic database, of 9.6 min per survey, while the Chaak system survey had a performance time of 7.8 min per survey (representing a 19% time reduction with the cell phone-based system) (Lozano–Fuentes et al. 2012).

The error rate on data entry was assessed in a controlled environment with mock entomological surveys, where premises were represented by sections of a large room and containers by paper printouts. The printouts were divided into four panels that depicted, respectively, container type, absence or presence of water, whether immatures were present in the container, and whether it contained pupae. The mock entomological surveys were conducted in 10 rounds of 5 min each with the data-gathering method assigned randomly to six surveyors. The number of containers in the premises was also randomly selected at the beginning of each round. In total, 30 surveys were completed for each method. The surveyors using pen-and-paper also reentered the collected data into an electronic database. Errors were scored as deviations from the known exact value (perfect score = 0, all entries incorrect = 1). Each breeding site type has four records in the survey as previously described. Because there were 11 different container types in the survey that the surveyors had to record 44 values, that is, they had 44 chances of success. Thus, the error rate is the total number of errors divided by the total number of entries. The averaged error rate was 0.23 for the pen-and-paper-based method versus 0.17 for cell phone-based data collection (Lozano–Fuentes et al. 2012).

Results

Key Features of the Chaak System

The Chaak system was developed as a proof-of-concept project for cell phone-based field data capture and rapid transfer of collected data into a central database, including over a wireless network. The system deals specifically with the collection of surveillance data for mosquito immatures (larvae and pupae) in artificial containers located on individual premises, which is a common surveillance methodology for mosquito vectors of dengue virus.

There are three different roles defined in the Chaak system: system administrator, manager, and surveyor (field worker). A system administrator can create an individual user and assign the role of that user, and also can register individual authorized cell phones for use to collect data. Managers have access to domain-specific functionalities, such as: 1) defining options in selectable lists, 2) defining geographic entities to be surveyed, or 3) managing task assignment to surveyors. The surveyor role is limited to data entry through the cell phone interface and transfer of collected data to a central database. The system allows a manager to assign specific tasks to a surveyor through the thin client desktop/laptop interface (Fig. 2), which results in the surveyor receiving the assigned tasks when activating the application on the cell phone (Fig. 3) and then being able to capture complex data on the cell phone interface to complete the tasks (Fig. 4).

Fig. 2.

Fig. 2

Open in a new tab

Screenshot of desktop client for task assignment. The task definition panel is used to describe the type of task. The personnel panel is used to select the people responsible to complete the task and which cell phone is assigned to each person. The geographic areas panel is used to select the geographic areas where the work will be done and assign the person responsible to cover each area. The radio buttons at the top of the panel group the work by geographic area (as shown in the figure), by person, and as a list. (Online figure in color.)

Fig. 3.

Fig. 3

Open in a new tab

Cell phone client screenshots. Left panel, main screen where premises are grouped by premises to be visited, premises to revisit, and premises completed by the surveyor; Middle panel, assigned premises shown in a list; Right panel, assigned premises grouped and shown in a tree view. (Online figure in color.)

Fig. 4.

Fig. 4

Open in a new tab

Cell phone client screenshots. Left panel, entry form for immature surveillance data; Middle panel, continuation of the immature surveillance data entry form after scrolling down to show more container types; Right panel, screen for batch data transmission. (Online figure in color.)

Another important feature is the capability to transfer data collected on the cell phone to a central database through different modes of transmission, including: 1) continuous Internet access where the surveyor enters data into the cell phone and sends the data immediately to the central database over the Internet; 2) periodic Internet access where the surveyor enters data into the cell phone, saves the data in the local system database, and then sends the data to the central database over the Internet or locally setup wireless network when a connection is available; and 3) desktop or laptop data transfer where the surveyor enters data into the cell phone, saves the data in the local system database on the cell phone, and then transfers the data to a desktop or laptop computer containing or connected to the central database via a file (no Internet access needed).

Managing the Surveillance Effort Through the Chaak System Desktop Interface

One of the most important features of the Chaak system is that it allows managers to create surveillance tasks, that is, assign specific premises to surveyors for completion in a given time period. These tasks then are relayed to the surveyors when they activate the Chaak system cell phone application (Figs. 2 and 3). To be broadly useful, systems such as Chaak must be adaptable to local settings. In the specific case of surveillance of mosquito immatures in artificial containers, this entails being able to: 1) define locally appropriate container type classifications (e.g., tire, flower pot, 20-liter bucket, etc.) and 2) handle different geographic hierarchies. Using the Chaak system desktop application, a person with the role of manager can customize options for container type classifications to match the local standard.

To ensure a locally acceptable geographic depiction, the desktop application allows the manager to define a customized geographic hierarchy. This is achieved by allowing the manager to: 1) recreate the local hierarchy of geographic classes (e.g., country, state, municipality, city, neighborhood, premises) without providing a fixed one; and 2) creating cases, which we call geographic entities, for the geographic classes (e.g., México, Yucatán, Municipality of Mérida, City of Merida, Santa Rosa, Calle 35 norte Numero 2). The geographic entities are related by the space they occupy. Using a verbal description and the City of Mérida example, we can say that the City of Mérida is located in the State of Yucatán, and, because the neighborhood of Santa Rosa is located in the City of Mérida, we can further say that Santa Rosa is located in the State of Yucatán. As a result of using this approach, the system can generate tree structures to present and aggregate data. The development of a robust geographic backbone for the system is critical because surveillance activities commonly are carried out and recorded at the spatial level of premises and data are expected to be aggregated at other levels. For example, premises are identified by their postal address and are assigned to surveyors to visit. The resulting data are later aggregated and displayed in reports at a higher level to decision makers. Finally, the system does not enforce a particular format of postal addresses.

Data Capture on the Chaak System Cell Phone Interface

When activating the Chaak system on the cell phone, the surveyor is presented with a log-in screen. After a successful log-in, the system displays a series of tasks organized in three groups: premises to visit, premises to revisit, and premises completed (Fig. 3). The premises to visit can be viewed as a list or organized in a tree. Figure 3 illustrates a scenario where premises assigned to a particular surveyor are grouped by block (073-7003 and 073-7006), and as both blocks are located in the Santa Rosa neighborhood, they are shown to belong to that neighborhood in the tree view. This system feature was created to help surveyors in resource-constrained environments to locate premises without the aid of global position system-enabled mobile devices.

The surveyor proceeds to select premises and enter the data for that specific premises through the cell phone interface. An initial selection refers to the basic outcome of the premises visit: 1) entry granted, 2) no one home, 3) entry denied, or 4) to revisit (Fig. 4). If entry is granted, the surveyor proceeds to complete the survey for mosquito immatures, which includes capturing data by container class (e.g., small trash containers, tires, barrels or drums, and water tanks) and location (indoors or outdoors) for total number of containers present, number of containers with water, number of containers positive for immatures (i.e., with at least one larva or pupa), and number of containers with pupae (Fig. 4). Upon completion of data entry, the surveyor can choose to immediately transmit the data to the server database or to save the data in the local system database on the cell phone (1 MB of storage capacity is sufficient to store data for thousands of premises). After saving the data locally, the surveyor can transmit the data from the surveyed premises at a later time. Figure 4 shows the status of premises and the option for selecting premises for data transmission.

To minimize data entry error for the cell phone interface, data entry fields make use of terms from predefined select lists, radio buttons, and buttons (Fig. 4) as opposed to free text input. Entries of numbers are typically achieved through buttons used to increase (+) or decrease (−) a number from a starting number of 0. Use of text fields to record the number of containers is optional.

We also made some design changes based on our field testing experience. We favor a white background and black lettering because is it easier to read under sun light. Furthermore, the (+) and (−) buttons were moved below the text boxes where the number of containers are recorded because otherwise the finger used to operate the buttons obstructs the view of the text box. Finally, we note that one-handed operation becomes difficult when the screen size, measured diagonally, is larger than 10 cm.

Reporting

The thin client application also has built-in reporting capabilities, including the presentation of data for commonly used larval surveillance indices (i.e., house index, container index, and Breteau index) in a table format or as a Google Earth map (Fig. 5). In addition, the thin client can automatically “recreate” survey paper forms with the data entered by each surveyor, and it allows for entering immature surveillance data by providing a data capture screen.

Fig. 5.

Fig. 5

Open in a new tab

Example of map report. The map report was generated by the system in the form of a *.kml file displayed in Google Earth. The map represents the data collected inside the Graciano Ricalde neighborhood (polygon) and automatically agregated at that spatial level. The pop-up text displays the values for three larval indices (house index, container index, and Breteau index). A color version of the figure is available at: http://sourceforge.net/p/chaak/screenshot/map_report.png. (Online figure in color.)

Comparison of the Chaak System with Pen-and-Paper-Based Method

As reported previously (Lozano–Fuentes et al. 2012), use of the Chaak system, compared with pen-and-paper data capture followed by manual entry of data into an electronic database, resulted in reduced time of data entry (averages of 7.8 vs. 9.6 min per premises, or 0.13 vs. 0.16 person-hours per premises, respectively). In the Table 1, we present new data for cost estimates for use of the Chaak system (surveyor salary and cost of data transfer over the wireless network) versus the pen-and-paper-based data capture method (salary for surveyors, data entry personnel, or both). The U.S.$5.0 daily minimum wage estimate represents the high end of the minimum wage range for Mexico (based on data from Mexico’s “Comision Nacional de los Salarios Minimos” for minimum wage by geographical area from 1992 to 2012; www.conasami.gob.mx/pdf/salario_minimo/sal_min_gral_area_geo.pdf and a U.S.$0.08 for one Mexican peso exchange rate). The amount (KB) of data used per survey of a single premises and transmitted over the wireless network was estimated from our field records (102.4 KB), and the cost of data transmission was calculated based on two different rates representing the worst and best case scenarios for Mérida (U.S.$0.02/KB vs. U.S.$0.0002/KB). The resulting estimates for total cost per premises survey are U.S.$0.10 for the traditional method of pen-and-paper data capture followed by manual entry of data into an electronic database (0.13 man-hour per premise × U.S.$0.63 hourly wage), compared with a range from U.S.$0.10 to U.S.$2.13 for the Chaak system depending on the data transfer cost rate used (Table 1).

Table 1.

Cost estimate (U.S. dollars) for daily use of the Chaak system vs the traditional method of pen-and-paper data capture followed by manual entry of data into an electronic database in Mérida, México

Estimates Traditional method Chaak system
Most expensive local mobile data plan Least expensive local mobile data plan
Man-hour used per premises 0.16 0.13 0.13
Minimum wage per houra $0.63 $0.63 $0.63
Data use per premises 0 kb 102.4 kb 102.4 kb
Data transfer cost rate $0 $0.02/kb $0.0002/kb
Data cost per premises $0 $2.05 $0.02
Total cost per premises $0.10 $2.13 $0.10
Open in a new tab
a

Based on an 8-hr labor day.

Discussion

We describe an example of a novel public health computer system with capacity not only for field data capture through a cell phone interface but also for using cell phones to receive specific surveillance tasks defined on the desktop component of the system. This system deals with the collection of surveillance data for mosquito immatures, but the underlying conceptual and technical solutions can be applied to a wide range of surveillance data for neglected tropical diseases. The freely available system, which was successfully field-tested in the dengue endemic setting of Mérida, México, addresses several problems inherent in dengue virus mosquito vector surveillance by: 1) providing capacity for data aggregation at different granularities by storing data at the finest spatial scale, 2) facilitating division of labor by helping program coordinators to manage and assign personnel to specific areas, and 3) enabling personnel oversight by tracking. The system can be adapted for local conditions through a configurable geographic information system and a configurable list of container types. Modifications to include additional functionalities (such as pupal demographic surveys) will require more development work with the participation of software developers. We went to great lengths to design and engineer the software to allow for further development, and we also made the full code available through an established code repository.

We previously reported that use of the system, compared with pen-and-paper data capture followed by manual entry of data into an electronic database, result in reduced time of data entry per premises as well as a reduced rate of data entry error (Lozano–Fuentes et al. 2012), and here we additionally show that using the system does not substantially increase cost, compared with pen-and-paper data capture followed by manual entry of data into an electronic database, as long as a control program can secure a low data transfer cost rate from a local provider. We did not yet perform a cost estimate for a large scale implementation of the system but this should be addressed in future work.

There has been an explosive development in mobile phone technology, including the emergence of cell phones that can store substantial amounts of data, over the last decade. Mobile phones are now commonplace throughout most of the world, and it has been estimated that there were over a billion mobile phones in developing countries by 2008 (Donner 2008). The proliferation of mobile phones in resource-constrained environments removes some previously noted obstacles for their use as surveillance tools: the risk of robberies and potential for arousing suspicions of residents through the use of mobile phones during field surveys (Curioso 2006). However, intermittent connectivity for mobile phones, especially with regards to Internet access, remains a problem in developing countries (Brewer et al. 2005). This can be counteracted, as done in our project, by the use of cell phones with capacity for local data storage, thus allowing for data to be continuously collected and then transferred when the Internet or locally setup network becomes accessible.

Data capture on mobile phones has several potential benefits, including reduction in error of transcribed data, rapid data transfer to a central database, and cost savings through reduced time to complete a survey. Our previous finding that the use of the Chaak system, compared with pen-and-paper data capture followed by manual entry of data into an electronic database, resulted in ≈25% reduction in data entry error (Lozano–Fuentes et al. 2012) is not surprising considering the tedious and error-prone nature of the task of manually transcribing data from paper records into an electronic database. A similar result was reported from a study where handheld computers (PDAs) were compared with paper-based collection of survey data in South Africa (Seebregts et al. 2009). Our testers reported that errors most commonly resulted from not being able to see the screen in direct sunlight. Common errors included selecting an incorrect container type, entering data for the container type located next to the intended one on the screen, and removing containers instead of adding them or vice versa. The Chaak system does not have built-in algorithms to pull out unusual or inconsistent data entries; however, it records basic data for auditing including the date and time of data entry, the user name making the entry, and data updates and deletes. Another study from South Africa where community health workers used mobile phones to conduct household surveys showed that data capture on mobile phones also can facilitate data transfer to a central database in real-time or near real-time in operational settings in developing countries (Tomlinson et al. 2009).

Our field trial revealed a 19% reduction in the overall time spent per survey while using the Chaak system, compared with pen-and-paper data capture, along with less errors in the data (Lozano–Fuentes et al. 2012). Because this essentially represents a salary cost saving, it has direct implications for the feasibility of implementing this type of data capture system in resource-constrained environments. Previous studies indicate that initial set-up costs for mobile data capture (primarily hardware including desktops and mobile devices) are offset in the longer term through increased data collection efficiency (Mueller et al. 2009, Seebregts et al. 2009). In our study, the best case scenario for the Chaak system provides no cost savings per premises survey when considering both salary and data transfer costs, but the time saved translate to 19 more surveys per 100 completed surveys. The practicality of implementing a system like Chaak is facilitated by the rapidly falling prices of cell phones with capacity to support the current state of the Chaak system (e.g., <U.S.$90 for Huawei 8100, Motorola XT316, Dell AERO).

One potential factor to consider is the salary for field surveyors and IT personnel. We used a high-end minimum wage estimate for México in our cost calculations. Lower salaries will reduce the cost of the pen-and-paper-based method (salary cost only) to a greater extent than for the Chaak system (salary and data transmission cost), unless agreements can be reached with local network providers to heavily subsidize this type of societally beneficial public health use of cell phones for data transfer.

Potential logistical problems for the use of mobile devices as data capture tools include battery life, difficulties reading the phone screen in direct sunlight, and cell phone durability. The cell phone can be protected by use of a hard case but this will add to the cost of the device. Data security is another potential problem, as the cell phones can be stolen or misplaced. The Android operating system provides tools to manage user access and prevent data theft from a cell phone, but there is no mechanism to render the cell phone unserviceable without the help of the cell phone service providers. Data theft prevention functionality does not currently exist in the Chaak system. Owing to limited readability of cell phone screens in direct sunlight, entry forms for field data collection cannot be very elaborate. An easier-to-read screen could allow more (or smaller) data entry elements in the screen. Commercially available potential solutions include devices with an e-ink screen, devices such as tablets with a larger screen, or devices that use voice commands for data entry. Another important logistical issue is that the Chaak system is highly dependent on address data. We designed the system around “premises” as used, for example, by a taxing authority. Urban environments are very dynamic and in poorly planned cities, there might not even be proper addresses. To address this problem, the system can use local administrative geographic hierarchies and does not restrict the format of the address.

Another issue that can only be resolved in larger scale field trials of the Chaak system is that we tested the system with a small workforce (<10 individuals), whereas a real-life scenario may involve a much greater number of field surveyors. Although the system performed robustly in our limited field trial, and the selected operating system (Linux) and the relational data base management system (PostgreSQL) are well suited for large scale implementation, unexpected problems may occur with larger scale use. A final possible problem area is the server and IT infrastructure. For the testing, we used a low power net-book (Dell Latitude 2100; 1 GB of RAM) with a price of approximately U.S.$250, but in an operational setting, we expect that a server class computer would be required (approximately U.S.$1,000). Managing the server and database also requires a control program to include a person with IT skills, adding to the cost of the computer system.

We focus herein on description of the Chaak system and preliminary small-scale testing for system performance and use cost relative to a pen-and-paper method. The system allows for the connection of multiple surveyors and supervisors at the same time, but more work is needed before we can adequately address how the system will perform on a large scale, for example, when used by hundreds of surveyors and multiple supervisors, and requiring the use of a large number of cell phones. One particular challenge is how to distribute and pair the surveyors to the cell phones because the system is dependent on this information for work distribution and auditing. This can be solved in different ways, such as by using radio-frequency identification tags or other more mundane means to identify individual cell phones. A possible path to a large-scale implementation of the system is to start with data entry in the field using paper and later recapture the data into the system. As local work-flows and issues are resolved a gradually growing number of mobile devices can then be incorporated into the surveillance activities until the system is fully implemented.

Operational use of the Chaak system could provide several logistical advantages. For example, a surveyor with a network connected device could bypass traveling to the central office and instead travel directly from his place of residence to the areas that need to be surveyed. If surveyors need to travel to central locations to “clock in” and “clock out” or to pick up supplies every day, data transfers could be completed twice a day. We speculate that use of the system may lead to more rapid entry and analysis of surveillance data and reduced reaction time for vector control response measures. Another notable advantage is the recording of data at the premises level that can later be retrieved and summarized for use at other spatial scales.

Our study underscores the great potential for use of cell phones in the surveillance of neglected tropical diseases. The presented system is adaptable in that the geo-political hierarchy is customizable and that options for data entry can be readily changed to suit local data collection practices. In addition, the system makes use of a mass-produced consumer product— the cell phone—and allows for use of various cell phone models to ensure that there are available options in local markets. One important future direction is to extend the functionalities offered in the system so that it can be used more broadly as a surveillance tool to support efforts to reduce the burden of neglected tropical diseases.

Acknowledgments

We thank the development and testing team members at Universidad Autónoma de Yucatán (José Fernando Chan–Canul, Mildred P. López–Uribe, and Wilbert A. Chi–Chim) for their hard and dedicated work, Maria del Rosario Najera– Vazquez of Servicios de Salud de Yucatán for her invaluable advice and support regarding field surveys, and the homeowners who kindly gave us permission to conduct mosquito surveys on their premises. We also wish to convey a special acknowledgment to Roger Arana-Guardia (UADY) for his continuous and dedicated fight against dengue. This work was supported by the United States National Institutes of Health and National Institutes of Allergy and Infectious Diseases (one R21 AI080567-01A1), and by a generous gift from Qualcomm (http://www.qualcomm.com/) to purchase cell phones. The United States National Institutes of Health and National Institutes of Allergy and Infectious Diseases or Qualcomm had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Footnotes

References Cited

  1. Adedeji AA, Sanusi B, Tella A, Akinsanya M, Ojo O, Akinwunmi MO, Tikare OA, Ogunwande IA, Ogundahunsi OA, Ayilara OO, et al. Exposure to anti-malarial drugs and monitoring of adverse drug reactions using toll-free mobile phone calls in private retail sector in Sagamu, Nigeria: implications for pharmacovigilance. Malar J. 2011;10:230. doi: 10.1186/1475-2875-10-230. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Ahmed SM, Zerihun A. Possession and usage of insecticidal bed nets among the people of Uganda: is BRAC Uganda health programme pursuing a pro-poor path? PLoS One. 2010;5:e12660. doi: 10.1371/journal.pone.0012660. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Android. Android operating system. 2012 ( http://www.android.com/)
  4. Apache-Tomcat. The Tomcat project. 2012 ( http://tomcat.apache.org/)
  5. Aviles W, Ortega O, Kuan G, Coloma J, Harris E. Quantitative assessment of the benefits of specific information technologies applied to clinical studies in developing countries. Am J Trop Med Hyg. 2008;78:311–315. [PubMed] [Google Scholar]
  6. Blaya JA, Fraser HSF, Holt B. E-health technologies show promise in developing countries. Health Affairs. 2010;29:244–251. doi: 10.1377/hlthaff.2009.0894. [DOI] [PubMed] [Google Scholar]
  7. Brewer E, Demmer M, Du BW, Ho M, Kam M, Nedevschi S, Pal J, Patra R, Surana S, Fall K. The case for technology in developing regions. IEEE Comput. 2005;38:25–38. [Google Scholar]
  8. Curioso WH. New technologies and public health in developing countries: The Cell-PREVEN project. In: Murero M, Rice R, editors. The Internet and Health Care: Theory, Research and Practice. Lawrence Erlbaum Associates; Mahwah, NJ: 2006. pp. 375–393. [Google Scholar]
  9. Donner J. Research approaches to mobile use in the developing world: a review of the literature. Inform Soc. 2008;24:140–159. [Google Scholar]
  10. Eisen L, Coleman M, Lozano–Fuentes S, McEachen N, Orlans M, Coleman M. Multi-disease data management system platform for vector-borne diseases. PLoS Negl Trop Dis. 2011;5:e1016. doi: 10.1371/journal.pntd.0001016. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Focks DA. A review of entomological sampling methods and indicators for dengue vectors. World Health Organization; Geneva, Switzerland: 2004. [Google Scholar]
  12. Focks DA, Alexander N. A multi-country study on the methodology for surveys of Aedes aegypti pupal productivity: findings and recommendations. World Health Organization; Geneva, Switzerland: 2006. [Google Scholar]
  13. Free Software Foundation. GNU General Public License, version 3. GNU General Public License. 2012 ( http://www.gnu.org/licenses/gpl.html)
  14. Galaxy Nexus. 2012 ( http://www.google.com/nexus/#)
  15. Halstead SB. Dengue virus–mosquito interactions. Annu Rev Entomol. 2008;53:273–291. doi: 10.1146/annurev.ento.53.103106.093326. [DOI] [PubMed] [Google Scholar]
  16. Hernandez Avila M. Norma Oficial Mexicana NOM-032-SSA2–2010. 2011 NOM-032-SSA2-2010. [Google Scholar]
  17. Horstick O, Runge–Ranzinger S, Nathan MB, Kroeger A. Dengue vector-control services: how do they work? A systematic literature review and country case studies. Trans R Soc Trop Med Hyg. 2010;104:379–386. doi: 10.1016/j.trstmh.2009.07.027. [DOI] [PubMed] [Google Scholar]
  18. HTC. HTC smartphones. 2012 ( http://www.htc.com/www/smartphones/)
  19. Jeefoo P, Tripathi NK, Souris M. Spatio-temporal diffusion pattern and hotspot detection of dengue in Chachoengsao Province, Thailand. Int J Environ Res Public Health. 2011;8:51–74. doi: 10.3390/ijerph8010051. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Johnson PR, Blazes DL. Using cell phone technology for infectious disease surveillance in low-resource environments: s case study from Peru. In: Lemon SM, Hamburg MA, Sparling PF, Choffnes ER, Mack A, editors. Global Infectious Disease Surveillance and Detection: Assessing the Challenges–Finding Solutions. Institute of Medicine Workshop Summary. The National Academies Press; Washington, DC: 2007. pp. 136–157. [Google Scholar]
  21. Kahn JG, Yang JS, Kahn JS. Mobile health needs and opportunities in developing countries. Health Affairs. 2010;29:254–261. doi: 10.1377/hlthaff.2009.0965. [DOI] [PubMed] [Google Scholar]
  22. Kan C-C, Lee P-F, Wen T-H, Chao D-Y, Wu M-H, Lin NH, Huang SYJ, Shang C-S, Fan IC, Shu P-Y, et al. Two clustering diffusion patterns identified from the 2001–2003 dengue epidemic, Kaohsiung, Tai-wan. Am J Trop Med Hyg. 2008;79:344–352. [PubMed] [Google Scholar]
  23. Kaplan WA. Can the ubiquitous power of mobile phones be used to improve health outcomes in developing countries? Global. Health. 2006;2:9. doi: 10.1186/1744-8603-2-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Lang TD. Advancing global health research through digital technology and sharing data. Science. 2011;331:714–717. doi: 10.1126/science.1199349. [DOI] [PubMed] [Google Scholar]
  25. Lozano-Fuentes S, Barker CM, Coleman M, Coleman M, Park BB, Reisen WK, Eisen L. Emerging information technologies to provide improved decision support for surveillance, prevention, and control of vector-borne diseases. In: Jao C, editor. Efficient Decision Support Systems: Practice and Challenges in Biomedical Related Domain. InTech-Open Access Publisher; Rijeka, Croatia: 2011. pp. 89–114. [Google Scholar]
  26. Lozano–Fuentes S, Elizondo–Quiroga D, Farfan–Ale JA, Lorono–Pino MA, Garcia–Rejon J, Gomez–Carro S, Lira–Zumbardo V, Najera–Vazquez R, Fernandez–Salas I, Calderon–Martinez J, et al. Use of Google Earth (TM) to strengthen public health capacity and facitlitate management of vector-borne diseases in resource-poor environments. Bull WHO. 2008;86:718–725. doi: 10.2471/BLT.07.045880. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Lozano–Fuentes S, Ghosh S, Bieman J, Sadhu D, Hernandez–Garcia E, Garcia–Rejon J, Wedyan F, Tep-Chel D, Eisen L. Proceedings of the Twenty-Fourth International Conference on Software Engineering & Knowledge Engineering. Presented at the The Twenty-Fourth International Conference on Software Engineering and Knowledge Engineering (SEKE 2012) Knowledge Systems Institute Graduate School; Redwood City, CA: 2012. Using cell phones for mosquito vector surveillance and control; pp. 763–767. [Google Scholar]
  28. Matthews J, Kulkarni R, Gerla M, Massey T. Rapid dengue and outbreak detection with mobile systems and social networks. Mobile Netw Appl. 2012;17:178–191. [Google Scholar]
  29. Meankaew P, Kaewkungwal J, Khamsiriwatchara A, Khunthong P, Singhasivanon P, Satimai W. Application of mobile-technology for disease and treatment monitoring of malaria in the “Better Border Health-care Programme”. Malar J. 2010;9:237. doi: 10.1186/1475-2875-9-237. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Mono. The mono development framework. 2012 ( http://www.mono-project.com/Main_Page)
  31. Motorola. Motorola mobility. The mono development framework. 2012 ( http://www.motorola.com/us)
  32. Mueller DH, Abeku TA, Okia M, Rapuoda B, Cox J. Costs of early detection systems for epidemic malaria in highland areas of Kenya and Uganda. Malar J. 2009;8:17. doi: 10.1186/1475-2875-8-17. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. PostGIS. PostGIS: spatial database extender. 2012 ( http://postgis.refractions.net/)
  34. PostgreSQL. PostgreSQL: Relational database manager system. 2012 ( http://www.postgresql.org/)
  35. Randrianasolo L, Raoelina Y, Ratsitorahina M, Ravolo-manana L, Andriamandimby S, Heraud JM, Rakoto-manana F, Ramanjato R, Randrianarivo–Solofoniaina AE, Richard V. Sentinel surveillance system for early outbreak detection in Madagascar. BMC Public Health. 2010;10:31. doi: 10.1186/1471-2458-10-31. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Safaie A, Mousavi SM, LaPorte RE, Goya MM, Zahraie M. Introducing a model for communicable diseases surveillance: Cell phone surveillance (CPS) Eur J Epidemiol. 2006;21:627–632. doi: 10.1007/s10654-006-9033-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Seebregts CJ, Zwarenstein M, Mathews C, Fairall L, Flisher AJ, Seebregts C, Mukoma W, Klepp KI. Handheld computers for survey and trial data collection in resource-poor settings: development and evaluation of PDACT, a Palm (TM) Pilot interviewing system. Int J Med Inform. 2009;78:721–731. doi: 10.1016/j.ijmedinf.2008.10.006. [DOI] [PubMed] [Google Scholar]
  38. Tatem AJ, Qiu YL, Smith DL, Sabot O, Ali AS, Moonen B. The use of mobile phone data for the estimation of the travel patterns and imported Plasmodium falciparum rates among Zanzibar residents. Malar J. 2009;8:287. doi: 10.1186/1475-2875-8-287. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. The malERA Consultative Group on Diagnosis and Diagnostics. A research agenda for malaria eradication: diagnoses and diagnostics. PLoS Med. 2011;8:e1000396. doi: 10.1371/journal.pmed.1000396. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Tomlinson M, Solomon W, Singh Y, Doherty T, Chopra M, Ijumba P, Tsai AC, Jackson D. The use of mobile phones as a data collection tool: a report from a household survey in South Africa. BMC Med Inform Decis Mak. 2009;9:51. doi: 10.1186/1472-6947-9-51. [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Truelsen J. wkhtmltopdf. wkhtmltopdf. 2012 ( http://code.google.com/p/wkhtmltopdf/)
  42. Vanden Eng JL, Wolkon A, Frolov AS, Terlouw DJ, Eliades MJ, Morgah K, Takpa V, Dare A, Sodahlon YK, Doumanou Y, et al. Use of handheld computers with global positioning systems for probability sampling and data entry in household surveys. Am J Trop Med Hyg. 2007;77:393–399. [PubMed] [Google Scholar]
  43. Vazquez–Prokopec GM, Kitron U, Montgomery B, Horne P, Ritchie SA. Quantifying the spatial dimension of dengue virus epidemic spread within a topical urban environment. PLoS Negl Trop Dis. 2010;4:e920. doi: 10.1371/journal.pntd.0000920. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. WHO. Dengue: guidelines for diagnosis, treatment, prevention and control. World Health Organization; Geneva, Switzerland: 2009. [PubMed] [Google Scholar]
  45. Windows. Net. Net framework. 2012 ( http://www.microsoft.com/net/)

RESOURCES