Resources

Potential Solutions

There are many ways you can influence health equity using analytics and technology. Analytics can be applied: at the point of healthcare delivery, while training future medical professionals, in policy decisions, or during healthcare research and development.

You might: build training application for future doctors to better treat disadvantaged communities, develop an interactive map that displays the nearest healthcare provider to uninsured individuals, re-train the AI in a medical device, build a chatbot that answers questions about health insurance, make a mobile health screening app to serve people who don’t have access to transportation.

In any of these domains, you may consider several approaches to applying AI, such as:

• Remove bias in an existing model
• Build or use unbiased training set (e.g., with new data, employing new processing techniques)
• Develop methodology for identifying bias in training datasets
• Apply an explainable model to a problem that currently uses unexplainable AI
• Build an explainable and interactive visual interface on top of model outputs
• Address a new problem with an explainable model
• Build a toolkit/common functions library with unbiased methods
• Develop or apply a method to monitor and detect bias in existing models or assess how bias may change over time in deployment
• Develop new model evaluation metrics for assessing fairness, model explainability, etc. 

This is not an exhaustive list, but should give you an idea of the breadth of solutions that are relevant to this challenge.