In the past four decades, the field of artificial intelligence (AI) has made major advances in many aspects of healthcare. But there are still many challenges ahead, according to Ram D. Sriram, Ph.D., chief, software and systems division, National Institute of Standards and Technology (NIST).

“NIST is going to play an increasingly important role in AI going forward,” said Sriram, adding, “Standards introduced at the right time will lead to innovation.”

Dr. Sriram’s Powerpoint Presentation

Dr. Sriram spoke at an April 19, 2022 talk at the University of Miami Richter Library on “Transforming Health Care Through Artificial Intelligence Revolutions.” His talk was sponsored by the University of Miami Institute for Data Science and Computing (IDSC) through the Data Citizens Distinguished Lecture Series, which is co-sponsored by the Miami Clinical and Translational Science Institute. He was introduced by Yelena Yesha, Ph.D., professor of computer science, innovation director at IDSC and the first IDSC Knight Chair.

Sriram said NIST promotes innovation in industry to enhance economic security. That includes studying how AI can be used to make better measurements, as well as promoting the use of AI systems in fields like healthcare. “We use AI to measure things, but also need to measure AI,” he added. “For instance, how do you trust an AI system to make a medical diagnosis? We do the foundational research to advance trust in AI, such as establishing benchmarks and metrics for authenticating data sets and testing algorithms.”

A Brief History of AI

Noting that it was his first in-person talk in almost two years, Sriram reflected on the long evolution of AI and recent revolutions. While AI has its roots in ancient Indian and Green thinking about analogies and inferences, the term was coined in 1956 by faculty members at Dartmouth College, he said.

“By the 1980s, there was widespread recognition that expert systems could solve a wide range of human problems,” Sriram said, citing the example of automated airport transit systems. Other advances in AI followed in the 1990s, such as the commercial development of speech recognition software, as well as the use of robots in space exploration.

A second revolution in AI occurred in the early 21^st century with the development of neural networks optimized to solve certain problems. However, it was expensive to build and train these systems using then-current computers.

Now, the third wave of AI is underway with the symbiosis of neural networks and open knowledge networks. “Neural networks can think fast, using unstructured data, but can be fooled,” he said. “Open knowledge networks are very good at applying background knowledge and complex logic, so their strengths balance each other’s weaknesses.”

Applying AI in Health Care

In healthcare, NIST looks at AI applications on multiple levels from public health ecologies, to individuals, to organs, cells and DNA. Sriram cited NIST’s contributions to research on protein-protein interactions in Alzheimer’s disease as well as cellular mutations that made some individuals more resistant to COVID infections.

Now, the convergence of AI and mobile devices can help drive the “P9 Concept” for personalized medicine of the future. “Smart health care will involve electronic health records (EHRs), genomic information and data from wearable sensors and mobile apps,” he said. “It’s important for computer and medical scientists to work together and share their expertise in developing these personal health records.”

Sriram said that AI has applications at every stage of an individual’s life cycle, from womb to tomb. That includes being able to identify health issues that require interpretation by a medical professional. One example is the use of wireless capsule endoscopy for assessing the small intestine. The capsule produces about 8 hours of video, but an AI system using machine learning can identify the particular region of interest that might indicate Crohn’s disease or inflammatory bowel disease.

On the cellular level, NIST experts are working with bioscientists growing stem cells in petri dishes for eventual transplant into patients. “Not all colonies are good, so machine learning tools can identify whether the colonies are good or bad at an early stage,” Sriram said. “That can be particularly helpful to patients with age-related macular degeneration (AMD), where good cells can be implanted in the eye.”

Finally, Sriram cited “The National Artificial Intelligence Research and Development Strategic Plan: 2019 Update,” noting that a further update would be coming in 2022 or 2023. “Anyone interested in research should read these documents,” he added. “They indicate the ways we are making long-term investments in AI research, expanding beyond the U.S. to the world.”

About Data Citizens

Data Citizens: A Distinguished Lecture Series is an ongoing course of in-depth talks by experts in the field of data science on a wide variety of topics including data visualization, big data, AI, and predictive analytics. The Data Citizens lecture series is co-sponsored by the Miami Clinical and Translational Science Institute (CTSI) and is free and open to the public.