by: Leo Zhang
WebMD is a health information services website that contains health care publications, physician blogs, and a symptoms checklist that tries to diagnose medical conditions based on the symptoms the user describes. There is a running joke on the internet about how using this symptom checker for something like a mild headache will inevitably lead to diagnoses of cancer or brain tumors. While this is not to belie the possibility of a seemingly innocuous symptom stemming from a more serious problem, it does bring up a point on the current state of accessible self-diagnostic tools, as well as how they can be improved.
There are many benefits to having more reliable, easily accessible medical diagnostic tools. For physicians, it helps with double checking their diagnoses and prescription decisions. For patients, it provides an accessible and quick way to gauge their health to see if a trip to the doctor is necessary or not. How could such a system be implemented? One way is to use machine learning, or artificial intelligence to manipulate massive quantities of data and make complicated models. The Human Diagnostic Project, or Human DX, is a recent attempt to enable accurate, accessible, and affordable care for all by combining the intelligence of medical communities across the globe using machine learning. Proposed nearly a year ago, this project has since garnered the support of many top medical professional organizations, including the American Medical Association and Association of American Medical Colleges.
While the Affordable Care Act has reduced the number of uninsured people, there are still approximately 30 million Americans without health coverage. These people turn to “safety-net” health systems which offer care regardless of patients’ ability to pay. However, these hospitals often lack access to medical specialists, so the patient has two choices: pay an expensive fee for immediate consultation or wait several months to see one of the few specialists working at public hospitals. Human DX aims to overcome this specialist shortage and provide higher quality of care for all by creating a remote, accessible source of information obtained from many physicians around the world. For example, if a doctor needs diagnostic help, he or she can post the case in the application and have other physicians provide their inputs. Artificial Intelligence (AI) then aggregates this information and returns a single report. Studies on Human DX have shown that it can tell the difference between diagnostic abilities of medical residents and fully trained physicians, demonstrating that Human DX can also be used to assess the strengths and weaknesses of doctors and subsequently improving their quality of care.
Of course, there is still much to be improved on before incorporating AI into healthcare. Large tech companies such as Google and Microsoft have begun to join AI guided medical diagnosis startups and we can expect them to expand more into the health field in the future to help develop the software. On the more human-related side of this project, there’s the issue of getting physicians to volunteer time and submit their inputs as well as regulations of extremely inaccurate data. There may also be possible issues of privacy as patient information will be shared across the world. In any case, the usage of AI will likely introduce a new paradigm of healthcare that can potentially have very positive results. It is also important to note that this will not be heading into a future where diagnoses are solely carried out by programs. Human doctors should always be the basis of healthcare, but the cooperation between humans and machines can be used to solve problems more effectively.