The Need for AI-Driven Healthcare Policy in South Africa
South Africa’s healthcare sector is on the brink of a transformative era, with artificial intelligence (AI) at its helm. However, the current policy frameworks are ill-equipped to foster innovation in this burgeoning field. Five key issues demand immediate attention: outdated legislation, data and algorithmic bias, workforce impact, liability dilemmas, and a dearth of AI system innovation. The establishment of a comprehensive national policy framework addressing these concerns is crucial for the safe, responsible, and regulated adoption of AI in healthcare.
The Innovation Gap: AI Patenting in South Africa
A closer look at the patenting activity within AI in South Africa reveals a stark reality. Despite the country being listed in over 9,000 AI patent applications between 2012 and 2021, only ten patents were filed from within South Africa. This disparity underscores the need for an enabling environment that promotes the development of complex cognitive and creative AI systems.
Public Sector Data: The Key to AI-Driven Healthcare
One of the significant hurdles to AI development in Africa is data availability and the costs associated with its acquisition. The National Digital Health Strategy for South Africa 2019 – 2024 identifies the development of a patient electronic health record as a key priority. This record system could provide a wealth of high-quality data to train AI systems, alleviating the significant investment and effort required to curate non-optimised data.
The concept of using public sector data to develop, train, and improve AI-enabled systems is not new. The Declaration of Cooperation on Artificial Intelligence, ratified by 25 European countries in 2018, saw member states agree to ensure better access to public sector data to fuel AI development.
However, access to sensitive health data raises privacy and security concerns. A robust legal framework or governance system is needed to encourage innovation while preserving patient privacy and security. A federated data system, where data do not leave the participating organisation but can be accessed by authorised individuals to train algorithms, could be the solution.
Conclusion
The establishment of a public sector data institution, alongside the proposed patient electronic record, could incentivise the development and deployment of AI for use in healthcare in South Africa. By providing developers with secure and safe access to health data, we can respect the rights to intellectual property and digital authenticity.
