• Center on Health Equity & Access
  • Clinical
  • Health Care Cost
  • Health Care Delivery
  • Insurance
  • Policy
  • Technology
  • Value-Based Care

As AI Utilization Advances in Health Care, Experts at ATS 2024 Share Insights

News
Article

Fatima Rodriguez, MD, MPH, from Stanford University, and Matthew DeCamp, MD, PhD, from the University of Colorado, joined Michael Howell, MD, MPH, of Google, on the stage at ATS 2024 to discuss artificial intelligence (AI) in health care.

Artificial intelligence (AI) in health care can be transformative, but there remain challenges with bias in AI algorithms that could perpetuate disparities, explained speakers in the first keynote session of the American Thoracic Society (ATS) 2024 International Conference.

The keynote was moderated by Michael Howell, MD, MPH, of Google, brought together a pair of experts to discuss the impact of artificial intelligence (AI) in health care. Howell, a seasoned professional with a robust background in health care delivery science, shared insights from his tenure as the chief quality officer at the University of Chicago Medicine and as an associate professor of medicine at Harvard Medical School.

Modern Medical Technologies | Image credit: Prostock-studio - stock.adobe.com

Artificial intelligence can be transformative in health care, as long as bias isn't introduced into the algorithms, explained speakers at the American Thoracic Society 2024 International Conference.

Image credit: Prostock-studio - stock.adobe.com

Howell set the stage by outlining the evolution of AI and discussing the shift from traditional statistical methods to advanced models like transformers, referencing “Three Epochs of Artificial Intelligence in Health Care,” an article he coauthored, published in JAMA.1 Fatima Rodriguez, MD, MPH, FACC, FAHA, from Stanford University, and Matthew DeCamp, MD, PhD, from the University of Colorado, joined the conversation after sharing insights from their latest findings related to AI.

Rodriguez focused on the role of AI in cardiovascular disease screening and patient adherence.2,3 She discussed investigations that highlighted the challenges of preventing heart disease despite available preventive measures and how AI could serve as a tool to address patient motivation.

"Especially in the field of medical imaging, [AI] is transformative,” she said. “It can really help to actually catch things that we may miss with our eyes."

Traditional risk factors like high blood pressure and cholesterol often fail to capture the full scope of a patient’s risk profile, necessitating more advanced screening methods. She introduced an innovative deep-learning algorithm designed to predict coronary artery calcium (CAC) levels with high accuracy, demonstrating its potential to improve patient outcomes through better risk identification and preventive therapies.

Rodriguez further detailed the investigation, which investigated whether incidental CAC quantified on routine non–ECG-gated CTs using a deep-learning (DL) algorithm provided cardiovascular risk stratification beyond traditional risk prediction methods.2 The implementation of AI was effective in increasing adherence to preventive therapy measures among patients with newly identified CAC present in their scans, though Rodriguez acknowledged issues concerning insurance coverage and access.

DeCamp delved further into the complexities of bias in AI, emphasizing that it is more than just a data problem; it is a broader social issue. He highlighted various examples of bias in AI applications, such as disparities in chest X-ray interpretations based on gender, race, and socioeconomic status. Understanding biases embedded in AI algorithms and their outputs is imperative to ethical and accurate implementation, he said.

His proposed solutions to mitigate bias were straightforward: community engagement, promoting diversity in AI development teams, and addressing social and structural inequities. The concept of explainability in AI was also stressed as “trust can be fragile,” and a key ethics principle he shared about the chatbox used in his health system is transparency.

Prompted by Howell, DeCamp and Rodriguez shared their hopes and worries regarding further AI implementation in the next couple years.

“I think my hopes and maybe my worries are probably the same...” DeCamp said. “My hope is that in a few years, some of these health equity metrics will be used to identify and account for these and they’re really taken seriously and used in reporting requirements.”

Both Rodriguez and DeCamp acknowledged concerns about the rapid pace of technology adoption potentially compromising the thoroughness of medical practice, stressing the importance of testing and randomized control trials to avoid unintended consequences.

"My worries, especially where I live and practice in Silicone Valley, are that there's such a need for speed that we're substituting that for rigor and testing,” Rodriguez explained.

Overall, she amplified the transformative potential of AI in fields like cardiology and medical imaging, drawing parallels to the earlier integration of electronic health systems (EHRs). DeCamp added to the discussion by noting the increasing availability of educational resources on AI in medicine, emphasizing the inevitability of AI's integration into EHRs.

Howell closed the session by discussing the FDA's recent activities and the growing presence of AI in health care. He echoed the sentiments for responsible and ethical implementation of AI, and integrating it into the medical curriculum to prepare future clinicians. He called for the continuation thorough investigations that provide good quality evidence on this topic.

“We pulled a thread on some really important topics around AI: bias, inability, [and] the ethics of implementation…” Howell concluded.

References

1. Howell MD, Corrado GS, DeSalvo KB. Three Epochs of Artificial Intelligence in Health Care. JAMA. 2024;331(3):242–244. doi:10.1001/jama.2023.25057

2. Peng A, Dudum R, Jain S, et al. Association of coronary artery calcium detected by routine ungated CT imaging with cardiovascular outcomes. J Am Coll Cardiol. 2023 Sep, 82 (12) 1192–1202. doi.org:10.1016/j.jacc.2023.06.040

3. Somani S, Balla S, Peng AW, et al. Contemporary attitudes and beliefs on coronary artery calcium from social media using artificial intelligence. NPJ Digit Med. 2024;7(1):83. doi:10.1038/s41746-024-01077-w

Related Videos
Milind Desai, MD
Masanori Aikawa, MD
Neil Goldfarb, GPBCH
Mabel Mardones, MD.
Dr Bonnie Qin
Mei Wei, MD, an oncologist specializing in breast cancer at Huntsman Cancer Institute at the University of Utah.
Alexander Mathioudakis, MD, PhD, clinical lecturer in respiratory medicine at The University of Manchester
Dr Bonnie Qin
Screenshot of an interview with Ruben Mesa, MD
Ruben Mesa, MD
Related Content
© 2024 MJH Life Sciences
AJMC®
All rights reserved.