At the 2023 Pediatric Academic Societies meeting, a presentation exploring how pediatric clinicians can best use AI in their research.
Artificial intelligence (AI), which is defined as any technique which enables computers to mimic human behavior, once a relatively novel concept in health care, is now being integrated into medical practice and research at warp speed. At a session entitled, “Artificial intelligence for researchers,” Anthony C. Chang, MD, MBA, MPS, MS, chief intelligence and innovation officer at Children’s Health of Orange County in California, along with Alfonso Limon, PhD, principal consultant at Oneirix Labs in Carlsbad, California, offered up a colorful history of AI and then shared the myriad ways that practitioners/researchers can utilize its various platforms in their study of pediatric medicine.
The exploration of AI in health care, noted Chang, goes back to 1949, when Claude Shannon, an American mathematician, first worked on a computer program for playing chess. Seven years later, Dartmouth University hosted a summer research project in Hanover, New Hampshire, where the term “artificial intelligence” was used by John McCarthy, an American computer scientist from the Massachusetts Institute of Technology. The next few decades saw tools and systems for AI develop, and now, in the 21st century, AI progression includes the 2017 developments of CheXNet, an algorithm created to diagnose pneumonia and Cardio DL, the first FDA-approved, AI-assisted cardiac imaging in the cloud. In 2019, the FDA proposed a regulatory framework for the use of AI, suggesting in a paper an approach to premarket review of AI and machine learning-driven software modifications.
“In pediatrics,” noted Chang, “AI can be used by researchers to deal with complex data sets; AI algorithms can analyze data sets that may be difficult for humans to discern.” Additionally, AI can be used to develop predictive health outcomes for children; analyze x-rays and MRIs to help diagnose various health conditions; improve clinical decision making (again, by analyzing data and offering insights and recommendations); and more.
In fact, the uses of AI in medicine are practically endless, according to Chang’s presentations. ChatGPT, one of the most popular AI tools currently being utilized in health care, a language-processing tool with unsupervised machine learning for unstructured and unlabeled data, is being used to analyze and summarize large amounts of data, speeding up various clinical research projects. AI is now being used for everything from clinical outcomes (demographics, lab test results, disease diagnoses, electronic health records, and more); to medication (orders, concomitant therapies, point of sale data, prescription refill); medical and prescription drug claims; molecular profiling; family history; mobile health (fitness trackers, wearable devices); environmental issues (analyze climate factors, pollutants, infections and more); organization and assessment of patient reported outcomes, surveys and diaries; social media and literature.
Looking ahead, Chang predicts that AI will be more readily utilized in precision medicine and population health, advance wearable and sensor technology, genomic information, and social determinants of health. “Essentially,” explains Chang, there will be a health care nervous system, bringing forward continuous learning from a worldwide community,” and through AI, offer continuous medical education and training, such as the ability to enable a source hospital to share any imaging with a destination expert anywhere in the world.
Chang ended his presentation with a number of resources for the AI-curious researcher, including books, educational certifications, and review courses.
Chang ACC, Limon A. Artificial intelligence for researchers: primer and update. Presented at: Pediatric Academic Societies Meeting, April 27-May 1, Washington, DC.