A 2018 study, performed by researchers from UCSF, UC Berkeley, and Beth Israel Deaconess Medical Center, suggests that artificial intelligence (AI) can significantly improve the quality and efficiency of echocardiogram studies. Learn more about how biomedical research in the Bay Area is revolutionizing cardiology and medical treatment below.
What Is Deep Learning and How Can It Improve Medical Care?
Today’s physicians have access to more patient data than ever before, but there is also higher demand for efficient, personalized care. AI and deep learning can help balance these expectations and improve the patient experience.
In the last ten years, machine learning algorithms have become increasingly adept at interpreting complex information and large data sets. Deep learning is a form of AI or machine learning that uses convolutional or recurrent neural networks. These algorithms are modeled after the human brain, using layers of inputs, outputs, and “hidden layers” of nodes to evaluate and interpret data. Deep learning algorithms are especially good at image processing and recognition.
While all healthcare professionals have access to record amounts of data, some medical practices are particularly data-intensive. Researchers have demonstrated that properly-trained machine-learning algorithms can assist oncologists, dermatologists, and ophthalmologists in the diagnoses of specific disorders and health conditions. However, any healthcare practice that uses diagnostic imaging studies and well-defined diagnostic criteria could potentially use machine learning to read and interpret patient data, supplementing the physicians’ expertise and insight.
Importantly, these technologies will augment the healthcare experience, not replace physicians and other vital care providers. It’s nearly impossible for physicians to sift through every piece of patient data today. Algorithms can give providers time to fine-tune diagnoses, provide meaningful patient counseling, and create personalized treatment plans.
AI Can Improve the Efficiency and Accuracy of Cardiology Studies
In a 2018 study, published in the online journal npj, UCSF Health and other researchers trained a machine learning algorithm to read echocardiogram images. The software learned how to identify commonly diagnosed cardiac conditions by reviewing more than 180,000 echocardiograms images.
Once the training process was complete, the researchers gave both the algorithm and human technicians a series of echocardiograms to assess. The technicians accurately evaluated the studies 70.2 to 83.5 percent of the time. In comparison, the machine learning algorithm was 91.7 to 97.8 percent accurate.
Additionally, the researchers discovered that the machine learning algorithm was equally effective in interpreting both color and black-and-white images. Because black-and-white images are smaller and easier to send, this could make echocardiogram analysis not only more effective, but also more efficient.
UCSF Health: A Leader in Biomedical Research
UCSF Health is one of the world’s preeminent medical research institutions and is one of the country’s largest recipients of National Institutes of Health (NIH) research grants. In 2018 alone, UCSF won more than $593 million in NIH research funding.
With thousands of ongoing research projects and numerous Nobel Prize-winning scientists, it has revolutionized medical and healthcare research. Its researchers have found new ways to fight cancer, identified agents responsible for brain disease, and were vital to the discovery and use of embryonic stem cells in the fight against congenital disabilities, cancer, diabetes, and heart disease. Currently, more than 1,600 clinical trials are occurring at UCSF Health facilities.
Learn More About Canopy Health’s Alliance of Cutting-Edge Providers
Canopy Health’s alliance of award-winning physicians, hospitals, and care centers is one of the Bay Area’s newest and largest single-provider, high-performance network. If you’d like to learn more about our remarkable network and carrier partners, contact us online or at 888-8-CANOPY today.
Johnson, K., Torres Soto, J., Glicksberg, B., Shameer, K., Miotto, R., Ali, M., Ashley, E., Dudley, J. (2018, June 12). Artificial intelligence in cardiology. Journal of the American College of Cardiology. Retrieved from http://www.onlinejacc.org/content/71/23/2668
Madani, A., Arnaout, R., Mofrad, M., Arnaout, R. (2018). Fast and accurate view classification of echocardiograms using deep learning. npj. Retrieved from https://www.nature.com/articles/s41746-017-0013-1
Research (n.d.). UCSF Health. Retrieved from https://www.ucsfhealth.org/research/
Weiler, N. (2018, March 6). UCSF retains position as top public recipient of NIH funding. UCSF. Retrieved from https://www.ucsf.edu/news/2018/03/409956/ucsf-retains-position-top-public-recipient-nih-funding