One of the most important cogs of clinical medicine and the health care delivery apparatus is radiology. Recognizing this fact, Covera Health boldly ventures to improve the future of this field by using artificial intelligence and cutting-edge technologies. Today, the company announced $50 million in additional funding to continue this mission. The funding was led by Insight Partners, which has been part of Covera’s story for some time and has also invested in many other popular home products such as Calm and Udemy.
In addition, Covera also announced today that it will acquire CoRead, an AI-driven radiology quality assurance company that has seen rapid growth and success. The CoRead Quality System is used by more than 2,000 hospitals in the United States and provides data-driven methods to improve the accuracy and quality outcomes of radiology workflows.
With this new funding and acquisition in hand, Covera’s mission is bold: to use data-driven insights to improve the field of radiology. To that end, the company’s platform empowers providers with quality metrics for radiology research and opportunities to optimize readings. Ron Vianu, founder and CEO of Covera Health, explains that the field of radiology is unique in that “there are challenges in building a way to measure quality, scale that quality, and really measure the importance of high-impact care on patient outcomes and cost.” Therefore, Vianu explains that one of his main goals with Covera is to deliver a solution that is based on AI and data science that can catch errors and omissions to let doctors and organizations know how they can improve. where more value can be captured and most importantly, improved patient outcomes.
Dr. Lawrence Ngo, MD, PhD, who is the CEO and co-founder of CoRead, thoughtfully explains that quality assurance is a challenge to implement in clinical practices: “In a medical residency, you get continuous feedback on how you can do better and where can you improve with your readings; however, as soon as you finish, this feedback disappears. He further explains that people may intuitively think that this problem can be solved by having a second radiologist validate each radiological examination as a kind of “double check”, but this is extremely impractical, especially given the challenges that arise with personnel and costs. This is where CoRead and Covera’s technologies come into play – “using AI and data, we can now enable a ‘second read’ peer review and quality check process – except on a much larger scale.”
There are many potential applications with this technology. One possibility Vianu and Dr. Ngo discuss is using the technology to gather more robust insights from standard imaging. For example, if a patient receives a CT scan of their chest for an acute complaint in the emergency department, radiologists can usually only interpret that image in the context of ruling out any acute emergencies related to the patient’s particular visit and concern that day. However, Dr. Ngo explains that the same imaging/study, in the right context, can provide additional insights beyond the acute setting, such as screening for other diseases (eg, lung cancer) and thus provide valuable insight into the overall longitudinal picture of the patient’s health.
Indeed, the worlds of artificial intelligence and radiology are increasingly merging. A study was published earlier this year in radiology indicating that AI systems were particularly successful in predicting breast cancer risk. Another study published in Journal of Clinical Oncology showed that a deep learning model predicted lung cancer risk with remarkable accuracy using low-dose CT. Indeed, there are many more such examples that repeatedly test the use of artificial intelligence as a realistic means of expanding the practice of radiology.
Although the results are generally mixed, the technology is still promising, giving hope that perhaps in the next decade, AI could significantly expand the field in the clinical setting. However, the technology does not come without its challenges. Most experts recognize that AI cannot truly replace the clinical acumen and ability of a human radiologist to take into account unique patient factors; rather, AI systems are expected to add a level of verification or quality assessment to a physician’s own workflows and eventually require a human in the loop. Thus, AI is still seen as a way to augment insight generation, data collection and workflow management.
Nonetheless, Covera’s ambitions are bold as it eagerly seeks to use this technology to make a remarkable difference in healthcare delivery. Undoubtedly, the company may also face challenges in this mission; for one thing, the entire landscape of AI and technology, especially as it relates to radiology, is evolving rapidly — meaning the company will need to stay nimble with its own offerings. Also, achieving meaningful quality improvement is not for the faint of heart, as there are so many barriers in healthcare. However, according to Vianu and Dr. Ngo, increasing manpower alone as a means of ensuring quality checks may not be a feasible proposition, especially given the growing shortage of radiologists in the face of an aging population with increasing imaging needs. Thus, if this technology is thoughtfully developed and appropriately scaled, it can truly add value and help improve patient outcomes for years to come.
Follow me at Twitter or LinkedIn.