Leslye Dias

Dissertationsprojekt

"Machine learning in medical diagnosis: an evaluation of chances and risks"

As computer science develops more efficient algorithmic systems capable of performing tasks that outperform human technical capability, a significant problem has arisen between those eager to implement machine learning systems and those ringing the alarm bells. In the field of healthcare, these warnings carry the added weight of the complexity of potential risks to human life, public health, and human rights.
 

Medical practice is a notoriously complex setting for technologies to be deployed, not only at a technical level but also at an ethical one. The present project seeks to bridge this epistemological and normative gap by connecting an analytical assessment of the risks and possibilities of implementing machine learning in medical diagnosis and a rights-based approach that evaluates them to assist stakeholders in minimizing or averting the risks when possible and enhance the benefits.
 

The project is divided in three stages. The first one seeks to provide a coherent understanding of the possibilities of Machine Learning in medical diagnosis; the second stage will consider the implications for the rights of stakeholders in three areas of impact: Big Data, Algorithmic and Society, and their benefits at individual and collective levels arising from the opportunities found. Finally, the third will use a rights-based approach to assess the impact of the second stage findings and provide a framework that could help avert or minimize threats to rights and increase benefits. The project seeks to proceed with a normative ethics approach to provide comprehensive and meaningful explanations and solutions.