Abstract:
Community health workers (CHWs) in low- and middle-income countries play a vital role in public healthcare. CHWs particularly assist in improving maternal and child health conditions of the poor and vulnerable who often remain unaware of the available services and face socio cultural barriers in accessing the health services. India, which is still undergoing a burden of high child mortality, implements its CHW program as a flagship program with close to a million CHWs appointed across its states. However, under-training significantly limits the ability of CHWs to provide quality services. We address the training problem in India by designing and deploying low-cost mobile training tools that can complement the existing face-to-face training mechanisms. Our system adopts a hybrid architecture to use Interactive Voice Response for facilitating online audio training sessions. Thus, allowing CHWs to access training from anywhere through their feature phones, a key need that has been well recognized by HCI4D research. We contribute on the following aspects: (1) Testing the feasibility and efficacy of our training tool through a controlled field experiment (2) Unpacking the training needs of CHWs by analyzing a question and answer record of 1178 and mapping it back to the existing reference material through a large-scale deployment on 500 CHWs, (3) Investigating the potential for peer-to-peer learning models to address the challenge of experts availability through a controlled field experiment, and (4) Finally, exploring the potential for automated techniques in this domain by proposing a semi-automated NLP approach for curating generated learning content and exposing CHWs and women to Chabot-based education for the first time. By using a range of mixed methods and field experiments, this dissertation expands the focus of HCI4D and mHealth research on CHWs competence development in low-resource settings, an area that has long been neglected.
