In the future is probably a phrase Dr. Nancy Green has heard a lot. Especially as a computer scientist who works in the area of artificial intelligence.
Over the last several years, she's taken her knowledge of computer reasoning and applied it to another once-futuristic world genetics.
As scientists have discovered more and more about the human genome, genetic counselors are helping people understand their genetic probability for inheriting or passing along certain mutations which can lead to birth defects or later-in-life diseases.
Usually, genetic counselors spend 30 minutes at a meeting with a client, Green said. But sometimes the client might be too upset to process the information given at the meeting.
Genetic counselors draft follow-up letters which detail what they discussed from symptoms to diagnosis. The letters attempt to break down what is complicated genetics, laced with principles of inheritance and probability.
That's where Green comes in.
For the past several years with the help of a National Science Foundation Career grant, Green has been researching ways to use artificial intelligence (AI) to help genetic counselors write these letters.
The genetic counselor would enter basic information such as symptoms and diagnosis into the system. And then the computer would take it from there.
This is an interesting problem from the AI side, she said. AI is about simulating human reasoning.
So with the information at hand, the computer tries to make sense of it and puts together a letter that explains the reasoning behind the diagnosis. Green is quick to point out that the computer is not making the diagnosis, only tracing the steps it takes to get there.
Physically writing such a detailed letter is time-consuming for people. There is not enough time and not enough genetic counselors to keep up with the workload, she said. And as the field grows, time constraints are going to get worse.
Green partially selected this area of AI research because UNCG offers a genetic counseling program and has a Center for Biotechnology, Genomics and Health Research. She has worked closely with faculty in both areas. Their help has been invaluable, Green said.
She began the project by talking with them and studying sample patient letters from across the state. Along the way, she has involved both undergraduate and graduate students in her work.
Now that the reconstructive reasoning is done, the next step will be the inclusion of graphics. For some people, seeing visual representations of probability makes the concepts easier to grasp.
Green's work is right on target in the AI world. Health care is a frontier for AI right now, she said.
In November, Green co-chaired a symposium on Virtual Healthcare Interaction. Some projects her peers are working on include computer characters that can talk and interact with people. Those characters can remind people to take their medicine or encourage them to exercise.
We'll never replace humans, Green said. But computers can be useful without being as smart as humans.