Four months ago, we drew attention to the Global Breast Cancer Initiative released by the WHO in which they laid out a roadmap to save 2.5 million women’s lives by 2040. They highlighted three actions: “(1) recommending countries to focus on breast cancer early-detection programmes so that at least 60% of the breast cancers are diagnosed and treated as early-stage disease; (2) diagnosing breast cancer within 60 days of initial presentation can improve breast cancer outcomes, with treatment starting within three months of first presentation; and (3) managing breast cancer so that at least 80% of patients complete their recommended treatment.”
A week ago, the WHO turned their attention to key regulatory considerations on artificial intelligence (AI) applied for health. Their publication (click here to download) focused on the importance of safety and effectiveness, the rapid deployment of AI systems to those in need, encouraging discussion among stakeholders, including regulators, developers, manufacturers, healthcare workers and, of course, patients. The WHO recognized the potential of AI tools to transform the health sector.
Enhanced health outcomes could be accomplished by: improving clinical trials; strengthening medical diagnosis and treatment; and supplementing the knowledge and skill of healthcare professionals. The WHO suggested that AI could be extremely useful in places where there was a lack of medical specialists, such as the remote interpretation of radiology images. Dr Tedros Ghebreyesus, Director General of the WHO (seen left,© Wikipedia), commented: “Artificial intelligence holds great promise for health, but also comes with serious challenges, including unethical data collection, cybersecurity threats and amplifying biases or misinformation.”
He continued, “This new guidance will support countries to regulate AI effectively, to harness its potential, whether in treating cancer or detecting tuberculosis, while minimising the risks.” The WHO publication focused on six areas in which AI for health should be regulated: (1) importance of transparency and documentation; (2) risk management; (3) validation of data and clarity of intended use; (4) commitment to data quality; (5) privacy and data protection; and (6) collaboration between all interested parties.
The WHO recognizes that AI systems are complex, depending not only on the underlying software, but also the data they are trained on, which typically is gathered in a clinical setting. AI models can find it difficult to represent a diversity of populations and so the regulations can guide developers to ensure that relevant attributes are recorded to ensure datasets are representative. Given the WHO’s Global Breast Cancer Initiative, and their recent focus on regulatory considerations on AI for health, I am optimistic it should be possible to combine the two areas for the benefit of women worldwide.