Design and Evaluation of an AI-Assisted E-Governance ICT Solution to Empower Non-Literate Rural Healthcare Consumers

Abstract

Rural healthcare consumers in low-literacy regions face persistent barriers to accessing e-governance health services because most digital health portals assume reading proficiency, stable internet access, and familiarity with graphical interfaces. This study designs and evaluates an AI-assisted e-governance ICT solution that replaces text-dependent interaction with a voice-first, multilingual conversational interface built on Automatic Speech Recognition (ASR), Natural Language Understanding (NLU), and Text-to-Speech (TTS) components layered over an existing government health database. A mixed-methods pilot was conducted across three rural districts (N = 240 households) over an eight-week deployment window, comparing the proposed system against a conventional text-based portal on task completion time, System Usability Scale (SUS) scores, ASR/intent accuracy, and adoption rate. Results indicate that the AI-assisted voice interface reduced mean task completion time by approximately 61%, achieved a mean SUS score of 81.1 (interpreted as 'good' usability) across literacy groups, and reached a cumulative adoption rate of 76% by week eight, compared with markedly slower uptake historically reported for text-based government portals. The findings suggest that voice-first, AI-mediated e-governance interfaces can materially close the digital health access gap for non-literate rural populations, while highlighting connectivity, dialectal variation, and trust-building as remaining design challenges. The paper concludes with policy implications for national digital health inclusion strategies and directions for future work on offline-capable, federated AI models.