Duke Researchers Unleash AI to Transform Modern Medicine

While many know AI for its ability to write an email or craft a grocery list, the powerful technology behind tools like ChatGPT is rapidly advancing into the critical field of healthcare.

Innovations in artificial intelligence and machine learning are paving the way for a future where we can detect health changes before symptoms appear, understand the complexities of our brains, and develop novel treatments for challenging diseases. These groundbreaking topics were the focus of the recent Duke Summit on AI for Health Innovation.

Decoding the Language of Biology

For Pranam Chatterjee, an assistant professor of biomedical engineering, the true potential of large language models extends beyond human language to the fundamental language of biology. His lab is training these models to generate sequences of molecules, effectively designing new proteins from scratch.

This isn't just a theoretical exercise. His team is creating proteins with specific functions, such as editing DNA sequences, modifying disease-causing proteins, or even growing human eggs from stem cells. This work could lead to new therapies for diseases like Huntington’s and cancer and offer new hope for those struggling with infertility. “We make proteins that can make new cells and tissues from scratch,” Chatterjee explained.

Unlocking Insights from Patient Medical Records

Another significant challenge in healthcare is the vast and often unstructured data locked within patient medical charts. New faculty member Monica Agrawal is developing algorithms to mine this information effectively.

Crucial details about a patient's history, symptoms, and responses to treatments are frequently buried in doctors' notes, full of jargon and abbreviations. This makes it time-consuming and difficult to get a complete picture of a patient's health journey. Agrawal is building AI tools to analyze these notes, making the records easier to understand for clinicians, researchers, and patients. “The goal of my new lab is to make clinical language work for everyone,” she said.

Predicting Illness with Wearable Technology

Jessilyn Dunn, an assistant professor at Duke, is exploring how data from smartwatches and other wearable devices can provide early warnings for illnesses. Her team at the Big Ideas Lab uses AI to analyze continuous data streams from these devices.

This research could help individuals at risk of diabetes to take preventative action or detect infections like RSV, COVID-19, and the flu before symptoms even appear, potentially limiting their spread. “The benefit of wearables is that we can gather information about a person’s health over time, continuously and at a very low cost,” Dunn noted. The aim is to empower patients and provide timely, precise interventions.

Mapping the Brain to Treat Psychiatric Disorders

In what might sound like science fiction, associate professor David Carlson is using AI to interpret brain wave data to better understand emotions and behaviors. By analyzing the electrical activity in the brains of mice, his team can identify signals associated with aggression and even block them, making the mice more sociable.

Carlson believes this work has profound implications for human health. It could help researchers understand the neurological basis of social challenges in conditions like autism and social anxiety disorder. Ultimately, these insights could lead to innovative treatments for psychiatric disorders such as anxiety and depression.