AI Helps Accelerate Clinical Trials, Medical Research, and Decision-Making

Healthcare professionals at the University of Florida are developing medical systems using artificial intelligence and machine learning to analyze medical data, prepare surgeons for procedures, and improve overall patient care.

Image credit: University of Florida

Artificial intelligence is providing the fuel for Industry 4.0 and transforming nearly every sector of the economy. Using AI to analyze massive amounts of data has dramatically affected retail, accounting, finance, transportation, medicine, agriculture, aerospace, and more.

Now, the University of Florida (UF) is making artificial intelligence the centerpiece of a major. UF is preparing students for computer and information science positions with the university’s new Artificial Intelligence Initiative and Artificial Intelligence Academic Initiative Center.

The AI Initiative, which encourages AI-based research and technology development, is supported by the HiPerGator, the most powerful university-owned supercomputer. Among the processors Included in the HiPerGator supercomputer are more than 30,000 Intel® Xeon® processors, according to UF.

To expand upon its AI Initiative, UF announced its Artificial Intelligence Academic Initiative Center in 2022. UF students can now take AI-related courses and complete degree programs and certificates through the Artificial Intelligence Academic Initiative Center with the goal of preparing interested students for computer and information science positions.

Medical InvestiGator

The UF AI Initiative allows student and faculty researchers to explore new, AI-powered technologies in a variety of fields. The University of Florida Health system is no exception. Researchers are developing SynGatorTron, the successor of GatorTron, to quickly analyze and extract data to accelerate medical research.

SynGatorTron can create synthetic medical data, untraceable to real patients, to train healthcare AI systems to recognize medical terminology and language. SynGatorTron removes the time-consuming process of removing patients’ personal data from electronic health records before AI applications can use them. UF researchers believe using synthetic patient data better protects patient privacy, yet still maintains accurate medical information.

The SynGatorTron natural language processing model accesses and understands medical language from electronic health records, including clinical data. SynGatorTron allows users to analyze large amounts of medical information quickly, expediting medical research and decision-making, and ultimately improving patient care. For example, SynGatorTron could help researchers select patients for clinical trials in minutes rather than months. Both GatorTron and SynGatorTron were trained on UF’s HiPerGator.

University of Florida's HiPerGator supercomputer.

Image credit: University of Florida

Reduce Postoperative Complications with AI

Healthcare professionals estimate that, on average, 12.5 percent of patients nationally experience postoperative complications. Postoperative complications can increase a patient’s mortality risk, and they often also increase their medical costs.

UF researchers are also working on an AI system that helps medical professionals predict and manage possible postoperative complications. The system, known as MySurgeryRisk, uses machine learning to analyze patients’ medical data and notifies doctors of possible complications, including sepsis and blood clots, and their associated risk levels.

MySurgeryRisk enables doctors to monitor and care for possible postoperative complications before they become a health or mortality risk to the patient. The system uses data from more than 70,000 surgical procedures and nearly 60,000 patients. Although still under development, UF researchers working on MySurgeryRisk claim that the system predicted complications at least as accurately as physicians in recent testing.

The information from MySurgeryRisk is available to doctors on their mobile devices, allowing them to easily access their patient’s risk data. The system’s patient risk analysis can also inform surgeons of possible complications during surgery, making them better prepared for the procedures. The UF team working on MySurgeryRisk hopes that the machine learning system will reduce the likelihood of postoperative complications and improve patients' overall medical experience.