Developing a Responsible AI Framework for Healthcare in Low Resource Countries: A Case Study in Nepal and Ghana

Ethical oversight is a critical priority: 85% of survey respondents in Nepal and Ghana identified ethical oversight as a key concern, indicating that healthcare stakeholders in low-resource settings view responsible AI governance as essential, not optional.

Local governance structures are needed: 72% of respondents emphasized the need for localized governance structures, suggesting that one-size-fits-all AI regulations from high-income countries may not be appropriate and that communities should have a voice in shaping AI policies affecting their healthcare.

Data privacy and trust are major barriers: The paper identifies data privacy, reliability, and trust issues as major obstacles to AI adoption in these settings, highlighting that building public confidence in AI systems is as important as the technology itself in low-resource healthcare environments.

Bias mitigation requires contextual validation: The proposed framework includes "contextual validation approaches to mitigate bias," recognizing that AI systems trained on data from wealthy countries may perform poorly or inequitably for patients in Nepal and Ghana without localized testing and adjustment.

Responsible AI frameworks must be tailored to resource constraints: Rather than importing existing frameworks, the paper advocates for AI governance approaches specifically designed for low-resource settings, acknowledging that healthcare systems in Nepal and Ghana face unique financial, infrastructural, and regulatory challenges.