The sensors & wearables revolution is upon us. Because of their effectiveness as a monitoring & evaluation tool, development professionals should be actively seeking opportunities to integrate sensor technology into their data collection efforts.
What are Sensors & Wearables?
Like the smart phone in the early 2000’s, sensor technology is now the next big thing of digital devices. Sensors are small digital data collecting devices that have the ability to ‘sense’ the world around them via a number of sensory input tools. These small devices are reshaping the way that we collect, share and use information about the world around us. You might be surprised to know that sensors are already collecting data all around you, being used by corporations and governments to collect data in order to inform decision making, create diagnosis and signal warnings. Continuous data streams accumulated over long periods of time and that touch all aspects of society can provide incredibly powerful indicators that can be used to inform and influence decision making and policy development.
How Do Wearables Work?
Sensors take the form of wearables when designed to be part of a users clothing, jewelry, medical device or anything else that is worn. Some of the types of data collected by wearables include temperature, heart rate, glucose levels and perspiration. As internet connectivity and sensor distribution becomes more pervasive and sensors gain the ability to speak with each other and other systems a fundamental shift in decision making will occur.
Problems With Traditional Forms Of Monitoring & Evaluation
Traditionally development professionals have collected data via person-to-person surveys, subjective observations, and/or expensive and time-consuming experimental studies. Data is most often recorded by hand, and processed for single projects only. These traditional methods have numerous limitations. Often times the data is skewed because of courtesy bias (when surveyor wants to please analyst) or recall bias (forgetfulness on analysts part). Frequency and general presence in areas of data collection by non-locals can also skew data. Because of time and resource constraints, even in cases where experiments are well designed as in random samples in controlled trials results are often not available until the project itself is considered finished which basically makes the information null as it cannot be applied. While these methods may work on a small scale they don’t provide anywhere close to the magnitude of data required to conduct big data analysis of the multitude of intersecting systems that are all contributing to large scale development outcomes.
In the recent article, A Proposed Integrated Data Collection, Analysis and Sharing Platform For Impact Evaluation, Andreas Kipf et. al. surveyed 19 development professionals involved in impact analysis projects about their field operational and research needs. Their survey focused on these categories:
They found that most data collected in the field is done using tablets and mobile phone devices with some sort of data collection application to support. Most surveys consist of 100 questions or more and often times 1,000’s people are interviewed. In some areas sensors are used to collect physical data such as energy, power and temperature. Depending on infrastructure some of these systems store data locally to be collected at a later date, whereas others have a direct satellite connection to upload data in real-time into the cloud. Open Refine is most commonly used to clean the data, and the programming languages most commonly used were R, Python and C/C++. Once data is collected it’s shared in numerous ways including ’email, cloud storage services, scientific data platforms, and web sites or databases of research groups’. Only about half of the respondents from the survey said thy shared their data publicly via Github, email or through publications. This is partially because of security concerns, but also data transparency hasn’t been a historical tradition in development. Some of the security protocols used for sensitive data sets included ‘access control checks using databases or protected files, encryption of data that is in transit, and de-identification of data prior to its publication’. The results of this survey reveal much room for improvement in terms of instituting best practices for the industry so that we can not only save on costs of managing these types of systems, but to better share data with each other in a standardized format leading to better a faster development outcomes.
Why Use Sensors & Wearables For International Development?
If we are to make a real impact on the sustainable development goals, the integration of sensors & wearables technology at every level is absolutely necessary in order to attain the magnitude of data required to accurately inform on primary indicators pertaining to each goal. As we have learned from the millennium development goal epoch, eradicating something like global poverty is an incredibly complex problem that requires vast quantities of real-time data to inform accurately on the impact of development activities. Gaining access to information that we know has a high level of correlation to these issues and in real time is one way to improve our chances of having a long-term sustainable impact on these difficult social and environmental issues.