96230 Connections Transport & ICT Advancing Development with Mobile Phone Locational Data Improving the Effectiveness of Assistance Ryan Haddad, Tim Kelly, Teemu Leinonen, and Vesa Saarinen Mobile phones, and especially smartphones, are opening new ways to assess and improve assistance and the delivery of basic services in the developing world. Each year, developing countries see an annual gain of about 500 million new smartphones, virtually all of which generate not only call data records but also, with their GPS and Wi-Fi capabilities, a rich set of more precise data on location and movement. The rapid diffusion of the phones and the locational data they generate are helping fuel the “science of delivery”—the evidence-based, experimental approach to project assessment and improvement. The technology is finding an expanding variety of uses. Recent examples involving transport and logistics include: 6 billion • Transit route mapping in Abidjan; •Supply chain management for community health workers in Malawi; • Transport planning in Côte d’Ivoire; and • Malaria tracking in Kenya The number of A notable and more impromptu use arose after a tsunami mobile phone hit Japan in March 2011. Health care authorities used call subscriptions data records (CDRs) generated by mobile phones to in developing track the evacuation from the vicinity of the damaged countries, out of Fukushima nuclear power plant. They then meshed the about 7 billion CDRs with health records to optimize the delivery of worldwide needed emergency health treatment. Phone Data and the Science of Delivery project learning and adaption. The technology em- bedded in smartphones, including global position- Commenting on the concept of “delivery science,” ing system (GPS) sensors, Wi-Fi capabilities, and a 2013 World Bank blog post noted that during the cameras, allows users to generate information in several years it takes to prepare, implement, and the field—such as when gathering survey data or evaluate a project “the world has moved on, prob- monitoring project activities—that is automatically lems mutate and practitioners need real-time data tagged with locational data. to learn as they do and respond to shifting client priorities. There is value … in real-time learning and The locational capabilities are speeding a trend to- adaptive iteration.” ward quantification and measurement that under- lies the emerging science of service delivery. They Mobile phones, and applications that run on them, also advance the ability to predict outcomes and have shown a remarkable capacity to provide JANUARY 2015 NOTE 04 thus support rapid adaption of programs, which is For example, to help deal with the overburdened road particularly important for responding to epidemics. and transit networks in Abidjan, Côte d’Ivoire, IBM’s AllAboard project analyzed 500,000 CDRs generated Mobile Survey Applications over a period of five months. Locational data indicat- Typically, data collected in the field are updated ed the origins and destinations of much of the travel and edited offline before they are uploaded into flow in the city, and an optimization model suggested central databases. But computerized data collec- how to reduce waiting and travel times on mass tion and editing was often prohibitively expensive transit routes. The result was four new bus routes and for developing countries before the advent of the extension of an existing bus route, collectively mobile phones, particularly smartphones with their expected to cut travel time by 10 percent. touchscreens and extensive user options. In Kenya, researchers mapped every call or text However, in Manila, where no mass transit routes message made from more than 14 million mobile had been previously mapped, a World Bank proj- phones and combined the information with knowl- ect in cooperation with the Philippines government edge of the regional incidence of malaria in the developed a mobile phone application to automati- country. The results, published in 2012, represented cally collect route data from the field while avoiding the largest study to date of the interaction of hu- the costs of offline editing. Employing an open-data, man travel patterns and the spread of malaria. open-source software system, the app allowed tran- The researchers found that they could estimate the sit staff members simply to ride the routes and allow probability that a particular person was carrying ma- the GPS capability of the phone to generate route laria parasites and could map the movements of car- coordinates that were simultaneously transmitted to riers to identify source areas. Thus, besides mapping the database (see Connections Note #2). and predicting malaria movement, the data identified Another software system, CommTrack, provides locations to be targeted for malaria control and elim- inventory and logistics management and is widely ination. The potential applications of the technique used to improve the distribution of medications to other diseases, notably Ebola, are promising. and medical equipment. In Malawi, Health Surveil- lance Assistants (HSAs) carry and prescribe a pre- defined list of medicines, which they receive from Outlook for Use of Phones and health centers. Using cStock, a CommTrack app, Locational Data the HSAs report their prescriptions and stock levels As smartphones and mobile broadband service to the health centers via their mobile phones, which become more affordable in lower-income countries, reduces wasted trips by HSAs to health centers as they already are in the advanced economies, that do not have the supplies they need. they will become increasingly useful in the drive to Call Data Records improve public service delivery. Locational data can also be collected passively in the Moreover, satellite-based positioning capabilities form of CDRs, which are very large, complex sets of are improving as other countries and regions—in- data. The records include the time and duration of cluding China, the European Union, and Russia— each call and the approximate location of the user, begin to build systems to complement the existing which can be derived by triangulation between the U.S. system. Access to multiple systems can im- cell towers with which the caller was communicating. prove connectivity, overcome bottlenecks in data traffic, and make the locational data generated by The potential value of passively collected CDRs is smartphones more precise. often much higher than actively collected survey data, if only because the sample sizes are so much larger and unit costs lower. For more information on this topic: https://openknowledge.worldbank.org/handle/10986/19316 Connections is a series of concise knowledge notes from the World Bank Group’s Transport & Information and Communication Technology (ICT) Global Practice. The series is available on the internal and external online platform of the World Bank Group. Connections discusses projects, experiences, and front-line developments in Transport and ICT. The series is produced by Nancy Vandycke, Shokraneh Minovi, and Adam Diehl, peer-reviewed by experts of the Practice on a bi-weekly basis, and edited by Gregg Forte. For more information on Connections, and to read other notes in the series, please visit: http://www.worldbank.org/transport/connections JANUARY 2015 NOTE 04