The importance of representative Data in Artificial Intelligence (AI) solutions in Health

• Frequently, populations, patients and conditions are underrepresented or outright missing in the data. As you know algorithms suffer from a “black box” effect which makes solutions difficult to explain so having source data which can lead to a logical interpretability in solutions using AI is of paramount importance.

AI not only processes the data, but it can also magnify biases and the data itself.

• Even balanced representative datasets may not be the answer when there are deeper inherent social constructs at play.
• Patients without access to care are those for whom AI may hold the greatest potential. However they are unlikely to be part of a dataset, thus widening digital divide. We can imagine this happening within borders but also in wider geographical ares such as Sub-Saharan Africa.
• Only by being thoughtful about this, can effective deployment and integration of the AI-based technologies be done, that bridge rather than worsen gaps in health equity and inclusivity.

Note that the presenter has used the concept of Augmented Intelligence instead of Artificial Intelligence. Please refer to the introductory article of this term.

Justin Ko, MD. AAD Position Statement on Augmented Intelligence. Fusing technology with human Expertise to enhance Dermatological Care. 8th World Congress of Teledermatology, Skin Imaging and AI in Skin diseases – November 2020