Inconsistent power supply and unstable internet connectivity have been identified as major challenges impeding the effective deployment of Artificial Intelligence (AI) in rural healthcare facilities.

This was disclosed by Professor Jerry John Kponyo, Principal Investigator of the Responsible Artificial Intelligence Lab (RAIL) at the Kwame Nkrumah University of Science and Technology (KNUST).

Prof. Kponyo made the disclosure while speaking on the topic, “Responsible AI and Health in Ghana: Challenges and Opportunities,” at an International Conference on AI in Healthcare and Pharmacy.

The conference was held at the University of Health and Allied Sciences (UHAS) in Ho, in the Volta Region, on the theme: “Accelerating Adoption of Healthcare AI in Ghana: From Vision and Policy to Practice and Impact.”

Prof. Kponyo also identified algorithmic bias as another barrier to the effective integration of AI into the healthcare ecosystem.

He said several diagnostic tools had been trained on foreign datasets and, consequently, failed to account for the unique genetic markers and lifestyle determinants of local populations.

He also highlighted trust and adoption challenges, noting that cultural resistance, fears of workforce displacement, and the use of unexplainable AI tools could hinder widespread acceptance.

Prof. Kponyo further identified data fragmentation and security as significant impediments, citing non-standardised records, systemic interoperability challenges, and critical gaps in patient data privacy.

The Principal Investigator said responsible AI meant smart tools that supported medical decision-making safely, fairly and transparently, while respecting human dignity.

He stated that AI tools must support doctors rather than operate as unquestionable “black boxes,” noting that biased or unclear systems could result in dangerous medical errors and ultimately put patients’ lives at risk.

Prof. Kponyo emphasised that effective governance of AI integration into healthcare was vital to ensure that every citizen received the same high standard of care, regardless of their location.

He stressed that healthcare AI also required clear clinical rules, explainable models to promote transparency, rigorous real-world testing, and strong security standards to protect patient data.