AI: Can artificial intelligence save a life?
Big Data and Mental Health
First-of-its-kind study of TikTok and its unique algorithm
With the emergence of publicly available AI tools like ChatGPT, the world is consumed with the possibilities and risks that AI brings to the table.
Researchers in the University of Minnesota GroupLens research lab have combined computer science and humancentered AI methods to advance the theory and practice of social computing. Their projects include recommender systems, online recovery communities for addiction and alcoholism, and tools for mental health.
Assistant Professor Stevie Chancellor’s research has public health implications.
Chancellor and her team did a first-of- its-kind study of TikTok, AI, and its unique algorithm. It found that the social media platform can serve as both a haven and a hindrance for users struggling with their mental state.
“While developing an AI solution to figure out when someone is in crisis and when to intervene, all of the decisions you make with people’s data must consider the individuals who will be impacted by the system,” Chancellor said. “I believe you can do this by involving people throughout the whole process through labeling or evaluating AI systems.”
Refining the outputs of these large data sets is the key to success in Chancellor’s area of interest. It is very easy for the researchers to gather a large amount of data, but it doesn’t provide any context for someone’s mental health status.
“When it comes to large language models, I’m curious how much these models help or hurt when the model guesses someone’s mental health. We have to get that right no matter what,” she said.
To learn more about this work, watch the "UMN CSE Webinar: AI for Mental Health Prediction in Social Media" — or share the video with your friends: z.umn.edu/CSEwebinarAI
AI vs ML
Artificial intelligence is an area of computer science that tries to get computers to be intelligent like humans. Machine learning is a branch of AI that does this by building models that use past data to make predictions on data it hasn’t seen before.
Pioneering AI therapeutics
Fasikl, a startup co-founded by achieved U.S. FDA “breakthrough Computer scientists around the Associate Professor Zhi Yang, device” status. Fasikl has licensed world are tapping AI for all sorts is unlocking new possibilities five IPs through the University of of solutions. Assistant Professor Zhi Yang is unlocking new possibilities for individuals with physical disabilities.
Two devices—one easing tremors for people with Parkinson’s and another improving mind-controlled robotic prosthetics—have achieved U.S. FDA “breakthrough device” status. Fasikl has licensed five IPs through the University of Minnesota. Based on more than a decade of research, these AI therapies are undergoing pivotal clinical trials with 12 hospitals including eight in the U.S. for global regulatory clearance.
AI to avoid burnout
Computer scientists around the world are tapping AI for all sorts of solutions. Assistant Professor Yoga Varatharajah is one of them. He is looking to alleviate burnout in medicine.
“Clinicians already have enough to deal with, and they often perform tasks they’re overqualified for, like reading scans and writing reports, leading to burnout and potential errors,” Varatharajah explains.
His research focuses on AI methods to interpret tests, optimize treatment decisions, and improve patient outcomes. He developed a new graph-based machine learning method to model brain activity that can detect conditions like Alzheimer’s and epilepsy very early using inexpensive EEG tests. He is also training students and professionals to ensure that AI applications in medicine are robust and trustworthy.
Making nanomedicine safer
More than a decade ago, industrial and systems engineering Professor Zhaosong Lu recognized the growing need for machine learning models and saw how his optimization expertise could enhance and improve the field.
Recently, Lu’s team explored how AI could be used to predict properties of tiny materials. Applications include nanomedicine or nanotech that administers drugs to fight diseases such as cancer and AIDS.
The researchers showed that they could predict the makeup and behavior of nanostructures, while minimizing side effects. Lu’s research also impacts future nanoscientists. He uses it as case studies to complement engineering and science courses in a handful of colleges in the Midwest and East Coast.