Read our Microsoft Research blog post, summarizing our work showing how children make up less than 1% of public medical imaging datasets and its downstream consequences:
www.microsoft.com/en-us/resear...
What's the harm in using adult-trained AI algorithms on children?
We show that AI trained to detect cardiomegaly in adult chest radiographs fail spectacularly and increasingly in younger children: Around 50% of healthy infants are falsely diagnosed with cardiomegaly.
The release of public datasets is driving innovations in healthcare AI, but are children seeing the benefit?
Our systematic dataset review reveals that children account for less than 1% of patients across 181 public medical imaging datasets.
#HealthEquity #SafePediatricAI
What's next?
We are calling on hospitals and policy-makers to act now. To close this gap, we must prioritize the careful release of pediatric data.
Learn more about this in our full paper: www.medrxiv.org/content/10.1...
