Researchers in the UK are working on this capability with a system called Babylon, the A.I powered healthcare system has been integrated into the National Health Service which works to find patients with urgent needs and separate them from those with lesser urgent illnesses.
Babylon is similar to other symptom trackers and asks about symptoms while probing with other questions to determine if over the counter medication will suffice or if it is time to seek medical attention, appearing to be especially useful in remote or rural areas where access to doctors is reduced. “Babylon shows how AI can act as a middle ground between self-diagnosis and a visit to the doctor.”
This system could also help reduce costs and save time by helping to eliminate unnecessary appointments and testing. Much like any artificial intelligence in healthcare technology this system will not be able to completely replace the human element, but it can help humans to assist more patients.
At some point or another almost every one has looked online to find insight into something medical, approximately 7.5 million Americans check WebMD each month for one reason or another. Online symptom checking websites can provide accurate diagnosis around 50% of the time, but that also leaves the other half worrying unnecessarily or even worse thinking they are ok when they are not.
With Babylon if you type “I have a cough and fever. What’s wrong with me?” the system will ask for more information and go through a list of other possible symptoms to determine the best course of action. This middle ground diagnosis can help those who are too busy or those who live in remote areas, as well as those with non-life threatening conditions.
These systems utilize extensive databases of conditions, images, and indicators allowing the natural language processing to understand the symptoms being described via chat to translate the corresponding data into instructions, and can an offer remote appointments with clinicians, fill prescriptions, order lab testing, and issue referrals.
This may prove to be very cost effective, developing these techniques may assist in lowering costs and fixing the unaffordability of healthcare. Savings will no doubt arise from avoiding unnecessary appointments, testing, and co-pays as well as possibly catching diseases sooner when it is easier and cheaper to treat them.
A.I can diagnose skin cancer as accurately as a dermatologist by using the ABCDEs of assessing moles, which it’s databases contain thousands of images of allowing the algorithms to classify moles more precisely. A.I has also helped pulmonologists to diagnose lung disease more accurately by identifying symptoms and assessing lung function which may be too small for humans to perceive. A.I has been used in Mayo Clinic trials, the matching program increased the enrollment of breast cancer sufferers in their trials by 80%.
“Supercomputers equipped with AI can provide alternative suggestions to medical professionals, which potentially cuts the amount of time it takes for a patient to be diagnosed and increases available treatment options. Clinical trials allow patients who have exhausted their options to potentially benefit from experimental treatment, increasing the possibility of remission.” says Bridget Rooney.
A.I is far from being perfect, rare diseases can be difficult because inaccuracies are more likely when there is not enough information to make a conclusive decision. Even in those cases this system could help to free up physicians to see patients with more urgent needs by separating them from those with lesser needs, and help doctors to make more precise and personalized diagnosis.
In a pilot study 92% of the UK healthcare providers found it helpful to receive information gathered before a patient visit as it saved time, the teamed approach also prevents doctors and patients from ceding autonomy or control to an algorithm.
“Due to AI having a reasoning process entirely of its own, humans end up in a position where they fundamentally cannot understand why AI makes a decision, even though the decision is correct. That’s problematic because humans will start working with tools that they don’t fundamentally grasp, which is scary at best and dangerous at worst.”
A.I transparency poses problem beyond not understanding how they make decisions such as it being finicky and not working broadly and equally well on all people. “…certain skin cancer predictors are terrible at making accurate predictions on darker skin, since most of the photos used to train the system, to teach it what to look for, are of white skin. These questions of AI fairness are a serious ethical and policy matter, but they’re especially of concern in healthcare.” according to Justin Sherman.
These problems may be fixed over time with more data and more use across diverse populations, with more data collected the better they will be at helping everyone which will take time and in the meantime could result in missed diagnosis and/or advice.
Artificial intelligence may not be able to completely level out the playing field between all patients just yet, but hopefully this is a start in the right direction. Who knows just how far this could go in helping to achieve a shift from reactive-curative to preventative care which will ultimately not only save time and money but lives as well.
