Results


Interpretable bias mitigation for textual data: Reducing gender bias in patient notes while maintaining classification performance

arXiv.org Machine Learning

Medical systems in general, and patient treatment decisions and outcomes in particular, are affected by bias based on gender and other demographic elements. As language models are increasingly applied to medicine, there is a growing interest in building algorithmic fairness into processes impacting patient care. Much of the work addressing this question has focused on biases encoded in language models -- statistical estimates of the relationships between concepts derived from distant reading of corpora. Building on this work, we investigate how word choices made by healthcare practitioners and language models interact with regards to bias. We identify and remove gendered language from two clinical-note datasets and describe a new debiasing procedure using BERT-based gender classifiers. We show minimal degradation in health condition classification tasks for low- to medium-levels of bias removal via data augmentation. Finally, we compare the bias semantically encoded in the language models with the bias empirically observed in health records. This work outlines an interpretable approach for using data augmentation to identify and reduce the potential for bias in natural language processing pipelines.


Artificial Intelligence in Africa's healthcare: Ethical considerations ORF

#artificialintelligence

Artificial intelligence (AI) can improve various aspects of healthcare. It can help reduce annual expenditure,[1] allow early detection of diseases, provide round-the-clock monitoring for chronic disorders, and help limit the exposure of healthcare professionals in contagious environments. The use of AI in healthcare systems in Africa, in particular, can eliminate inefficiencies such as misdiagnosis, shortage in healthcare workers, and wait and recovery time. However, it is important to safeguard against issues such as privacy breaches, or lack of personalised care and accessibility. The central tenet for an AI framework must be ethics. This brief discusses the benefits and challenges of introducing AI in Africa's healthcare sector and suggests how policymakers can strike a balance between allowing innovation and protecting data. This paper is for ORF's Centre for New Economic Diplomacy (CNED).