One of the most pervasive and frustrating public health challenges in low and middle-income countries (LMICs) is identifying individuals over time and space. The success of non-communicable disease (NCD) management, antiretroviral therapy, tuberculosis treatment, antenatal care, infant vaccinations, and even longitudinal research studies, hinges upon accurate identification at initial contact and re-identification thereafter. Currently, systems for individual identification in LMICs are either lacking or woefully inadequate.2 This perpetuates an inefficient status quo that burdens patients and families with coordinating their own longitudinal care. Surely we can do better. Recent years have seen an explosion of interest in developing electronic medical records (EMRs) and information technology systems for hospitals and health care centers in LMICs, seeking to facilitate longitudinal care. Yet it is self-evident that electronic records offer no benefit over paper records unless the identification problem has already been solved, which is hardly the case. With the proliferation of smart phone ownership and expanding cellular network coverage, mHealth Apps may be ideally suited to solving the identification problem.7 In Project SEARCH (Scanning EARs for Child Health), our multidisciplinary team of public health, engineering, and computer science faculty and students at Boston University (BU), partnering with private-sector social enterprise software companies, have focused on solving the identification problem through pattern recognition analysis of biometric data, using ear morphology as the identifier. Biometric data offer distinct advantages over external identifiers. By definition, biometric features are intrinsic and cannot be lost, left at home, sold, or traded. And the choice of ears is logical, offering clear advantages over other biometric identifiers. Over the past four years, with funds from a BU pilot grant and a crowd-funding campaign, we have developed a beta-test prototype of the SEARCH App. Our App runs a powerful, open-source, pattern recognition algorithm (Scale Invariant Feature Transformation (SIFT)), operates on Apple iOS or Android platforms, and has been programmed to be integrated within a popular open-source EMR that itself runs on smartphones (CommCare, Dimagi, Inc.). In a 2017 study among 194 BU undergraduates, SEARCH achieved 96% first rank matching, and 99.6% matching within the top 10. This establishes strong proof of concept and justifies taking the next logical step: evaluation and field testing in Zambia. Our overarching goal in Project SEARCH is to create a powerful but simple, non-invasive, user-friendly system for individual patient identification, optimized for use in LMIC settings, that is acceptable to patients, caregivers, and providers alike.
One of the most pervasive and frustrating public health challenges in low and middle-income countries (LMICs) is identifying individuals over time and space. In Project SEARCH (Scanning EARs for Child Health), our team has focused on solving the identification problem through pattern recognition analysis of biometric data, using ear morphology as the identifier. Our overarching goal in Project SEARCH is to create a powerful but simple, non-invasive, user-friendly system for individual patient identification, optimized for use in LMIC settings, that is acceptable to patients, caregivers, and providers alike.
