Wondering What the Confusing Title Means?
In recent, slightly creepy, news: The US Food and Drug Administration has approved an AI (Artificial Intelligence) software. It very accurately predicts whether a patient is about to die suddenly.
Why Is This AI Software Important?
The nature of medicine today is that it involves a huge amount of disparate information. And, most human minds are simply not as good as a computer at processing all that data. In fact, the US has a surprisingly high number of sudden, unexpected deaths in hospitals. The AI program from Excel Medical is able to correctly predict such fatalities up to 6 hours before doctors and with 90% accuracy. Hence, saving their lives.
It crunches through and integrates various metrics, such as vital signs, patient’s previous history, family history, current history, and medications. Medical institutes around the world have slowly built up Standards of Practices over the years to prevent these unexpected deaths. These usually involve the pulling in of many eyes and various experts and expertise, especially pharmacists, to try and proactively catch an issue. However, human doctors have many limitations such as limited time, attention, cognitive capacities, and energy. Ultimately, they are only human and tech can help us better allocate these very precious resources.
Similar Research Projects:
Teams out of Stanford are developing a deep-learning algorithm that can predict death up to a year in advance. This would allow us to properly schedule end-of-life/palliative care and devote resources accordingly. They are actually using the algorithm to learn from it. For example, by following trends, the AI figured out that a specific type of scan is a strong predictor of death. The doctors were able to understand why, which has deepened their understanding of human health.
Image Source: The Dataiku Blog
“We believe that a black-box model can lead physicians to good decisions but only if they keep human intelligence in the loop, bringing in the societal, clinical, and personal context,” Nigam Shah, a co-author of the new study, told Gizmodo.
What This Means:
The quantified self-movement, epitomized by our smartwatches, Fitbits, and Apple watches, has brought a wealth of data to medicine. They constantly monitor heart rates, and will soon be tracking much more. For all this data to be truly useful, it has to be understood and for that, it has to be analyzed. Software algorithms are the best way to do just that. Self-learning AI will parse through these tomes of statistics and identify correlations. The amounts of data we have access to are overwhelming but there is a lot of work being done to make it an asset.
Eventually, and inevitably, this data-centric approach may lead to far-reaching changes. For example, your annual check-up will likely involve a highly accurate life expectancy prediction. Your doctor will very reliably be able to tell you what to expect and how to best improve things. Apple and other major smartwatch companies will become medical titans. They will have troves and troves of data. Whoever manages to integrate best with other electronic medical record systems will likely become the major holder of medical data globally.
Unfortunately, this means that Medical insurance rates can be further specified potentially. Non-discrimination laws would have to be passed to prevent this from leading to discrimination. The flip side of this is that we can implement personalized medicine effectively. Moreover, tailored treatments and doctors appointments for each individual will increase the chances of successful outcomes.
Ultimately, this is just the tip of the iceberg. Medicine, particularly diagnostics, is about to become a lot more data-driven.
When I was teaching freshman Biomedical Engineers, I always told them that if we do our jobs right, doctors will be able to spend more time on the human side of healing than in the robotic procedures and data analysis that is better handled by technology.
Ultimately, I am hopeful that technology will make medicine more holistic globally.
