The Role of Exascale Computing in Predicting Perioperative Outcomes

 

Computers have facilitated and revolutionized the way we study and understand the complexity and multiscale nature of biology. The trend towards technological innovation has indicated that output data rates and data volume are growing at an exponential rate. To convert the raw data into useful information, computationally intensive algorithms that require high memory and disk space are employed. High-performance exascale computing is a crucial system in this new age of data (3). 

Exascale Computing is the development of computing systems that are 1,000 times faster than existing petaflop machines (10¹⁵ calculations per second), capable of the sustained delivery of at least 1 exaflop (10¹⁸ calculations per second). This has tremendous potential for computational data analysis and modeling, furthering our understanding and the development of new predictive multiscale models (3) 

In the realm of perioperative outcomes, exascale computing coupled with learning algorithms in artificial intelligence (AI) and machine learning (ML), could greatly improve the advance of personalized medicine and predicting perioperative outcomes. As with AI applications in other areas of clinical research, the growing amount of data being collected or generated is inputted to the most challenging AI problems, which require the most capable supercomputers (1). 

There are multiple areas within anesthesiology in which AI is being studied, including depth of anesthesia monitoring, control of anesthesia, event and risk prediction, ultrasound guidance, pain management, and operating room logistics. The future of exascale computing, as demonstrated by the scheduled delivery of Aurora in 2021 (capable of a billion-billion calculations per second), combined with AI algorithms on large datasets will help advance predictive perioperative outcomes (2, 5). 

Exascale computing could be applied in other areas of predicting perioperative outcomes. For instance, within hemodynamics, which involves the flow of blood within the organs and tissues of the body, exascale computing is especially relevant to guiding surgeons during procedures on delicate arteries (6). Tracing the movements of several million blood cells through the arteries supplying oxygen to the heart muscle is an extremely difficult and time-consuming task. With an exascale computer, simulation of just one second of blood flow–the duration of a single heartbeat–could be reduced from five hours on one of the world’s fastest supercomputers a few years ago, to potentially a few minutes. This research is currently underway at Duke University, where biomedical engineers are developing a massive fluid dynamics simulator, HARVEY, that can model blood flow through the full human arterial system at subcellular resolution (4). 

Other ways that exascale computing is being incorporated into predicting perioperative outcomes is through imaging. Adaptix, a medical imaging company, in collaboration with the University of Manchester, is using enhanced imaging to allow surgeons to effectively differentiate between healthy tissues and tumors in cancer surgery. 

The possibilities of exascale computing are tremendous, especially when it comes to predicting perioperative outcomes. This is due primarily to the fact that as technology becomes more complex and datasets grow, exascale computing provides much-needed computational power to tackle science’s big questions. 

Sources: 

1. Daniel A. Hashimoto, Elan Witkowski, Lei Gao, Ozanan Meireles, Guy Rosman; Artificial Intelligence in Anesthesiology: Current Techniques, Clinical Applications, and Limitations. Anesthesiology 2020;132(2):379-394. doi: https://doi.org/10.1097/ALN.0000000000002960. 

2. Fitzpatrick, Mary, and John Spizzirri. “Aurora Exascale Supercomputer.” Sciencesprings, Sept. 2019, sciencesprings.wordpress.com/tag/aurora-exascale-supercomputer/. 

3. Lee, Christopher T, and Rommie E Amaro. “Exascale Computing: A New Dawn for Computational Biology.” Exascale Computing: A New Dawn for Computational Biology – IEEE Journals & Magazine, 2018, ieeexplore.ieee.org/document/8452060. 

4. Kingery, Ken. “Virtual Reality Blood Flow Simulation To Improve Cardiovascular Interventions.” Duke Pratt School of Engineering, 15 May 2020, pratt.duke.edu/about/news/vr-harvis. 

5. Johnson, Rob. “Aurora Supercomputer to Assist in the Fight Against Cancer.” Informatics from Technology Networks, 2019, www.technologynetworks.com/informatics/articles/aurora-supercomputer-to-assist-in-the-fight-against-cancer-321907. 

6. Michael R. Mathis, Timur Z. Dubovoy, Matthew D. Caldwell, Milo C. Engoren, Making Sense of Big Data to Improve Perioperative Care: Learning Health Systems and the Multicenter Perioperative Outcomes Group, Journal of Cardiothoracic and Vascular Anesthesia, Volume 34, Issue 3, 2020, Pages 582-585, ISSN 1053-0770, https://doi.org/10.1053/j.jvca.2019.11.012. (http://www.sciencedirect.com/science/article/pii/S1053077019311590)