As Chief Information Officer of a national medical group, artificial intelligence (AI) presents the opportunity to contribute toclinical solutions that enable new treatment options for patients. During the past six months responding to COVID-19, our group has deployed more than 200,000 virtual health solutions to keep our clinicians safe and reduce the number of in-person patient encounters while continuing to provide care. Radiology departments have been indispensable allies in COVID-19 detection technologies. When looking at the big picture, I think about the significant strides AI has made in our radiology group and the many ways it can help improve the delivery of more accurate, timely and patient-centered care now and in the future.
"You can have the best AI on the market, but if clinicians can’t access it, it isn’t that beneficial"
My approach focuses on what technology is good at, making binary decisions quickly, which is why AI and radiology make a great pair. Radiology images are many shades of gray, and physicians can evaluate hundreds or even thousands of them when assessing a patient’s condition. Envision clinicians read more than 10.4 million radiology studies a year. Utilizing our data infrastructure, we can train AI to identify conditions that are more black and white like perforated ulcers, brain hemorrhage, rib fractures and pulmonary embolism. Our AI flags these potential conditions for our radiologists, who then make the final diagnosis.
Once a technology has a reasonable test case and has been proven effective, it needs to integrate into our data infrastructure to continue providing value. Essentially, you can have the best AI on the market, but if clinicians can’t access it, it isn’t that beneficial. As a national group, Envision is building a central data source and platform for all of our clinicians and tools while continually optimizing to create the most efficient workflow. Our team is continuously pushing updates, patches and testing new notification systems and processes to improve the clinician and patient experience. The average read time using AI technology can be reduced by 36 percent with earlier notification. As this technology improves, it will continue to provide benefits.
In medicine, these notifications can be lifesaving. With AI, we can flag potential COVID-19 in the lungs, even if the scan is for another ailment. We are working to ensure our clinicians have the option to jump to that case once the AI identifies it since we have a process to flag that suspected COVID-19 to our partners in the hospital in real-time, activating isolation procedures and other potentially lifesaving measures.
Beyond COVID-19, AI assistance in reading images helps cut down on diagnostic wait times. Shorter wait times mean a quicker transition to treatment. AI systems can also flag and prioritize serious diagnoses quicker, enabling our teams to quickly provide the appropriate care to those who need it most. AI detection also cuts down on missed detection rates. These systems can detect abnormalities with more precision. While there are some productivity gains with AI, the greatest advantage is likely on the accuracy and quality front.
Using AI to support binary decision-making processes gives our team the time and ability to provide the final diagnosis and enact a treatment plan. Precision medicine is still an emerging field. The path forward should focus on applying AI to the areas where it can have the most impact--in areas where there are black and white options, not complex, ambiguous, multi-factorial problems and conditions.