Despite an area under the ROC curve of 1, Cassandra’s
prophesies were never believed. She neither hedged nor relied on retrospective
data – her predictions, such as the Trojan war, were prospectively validated. In
medicine, a new type of Cassandra has emerged –
one who speaks in probabilistic tongue, forked unevenly between the
probability of being right and the possibility of being wrong. One who, by conceding
that she may be categorically wrong, is technically never wrong. We call these
new Minervas “predictions.” The Owl of Minerva flies above its denominator.
Deep learning (DL) promises to transform the prediction
industry from a stepping stone for academic promotion and tenure to something
vaguely useful for clinicians at the patient’s bedside. Economists studying AI believe that AI is revolutionary,
revolutionary like the steam engine and the internet, because it better predicts.
Recently published in Nature, a sophisticated DL algorithm was able to predict acute kidney injury (AKI), continuously, in hospitalized patients by extracting data from their electronic health records (EHRs). The algorithm interrogated nearly million EHRS of patients in Veteran Affairs hospitals. As intriguing as their methodology is, it’s less interesting than their results. For every correct prediction of AKI, there were two false positives. The false alarms would have made Cassandra blush, but they’re not bad for prognostic medicine. The DL- generated ROC curve stands head and shoulders above the diagonal representing randomness.
The researchers used a technique called “ablation analysis.”
I have no idea how that works but it sounds clever. Let me make a humble
prophesy of my own – if unleashed at the bedside the AKI-specific, DL-augmented
Cassandra could unleash havoc of a scale one struggles to comprehend.
Leaving aside that the accuracy of algorithms trained
retrospectively falls in the real world – as doctors know, there’s a difference
between book knowledge and practical knowledge – the major problem is the
effect availability of information has on decision making. Prediction is
fundamentally information. Information changes us.
In 2015 I think there is a good chance we’ll see a major security incident along the lines of this month’s Sony hack. This event will be like 9/11, in the sense that there will be a before and an after, and life as we know it will change forever. This has been coming for a long time. We’ll finally see how vulnerable we are and there will be a public outcry, most likely leading to some kind of government action. Up until now, most incidents have been security breaches of the disgruntled employee and clueless user variety, which are a huge big deal as far as HIPAA lawyers and privacy advocates are concerned, but not a very real threat to anybody.
This will be the real thing, with potentially disastrous results. I don’t know if this will be an attack on Healthcare.gov by a politically-motivated hacker group. (I’m surprised it hasn’t happened already. ) Or an attack on a academic hospital system designed to acquire potentially valuable patient and research data. Or a hack of a health insurance company, intended to Wikileak financial and (possibly damaging) patient claims data. For an insurer with a poor track record, this could cause serious problems.
88.2 % of all statistics are made up on the spot
– Victor Reeves
There’s a growing movement in medicine in general and imaging in particular which wishes to attach a number to everything.
It no longer suffices to say: “you’re at moderate risk for pulmonary embolism (PE).”
We must quantify our qualification.
Either by an interval. “Your chances of PE are between 15 and 45 %.”
Or, preferably, a point estimate. “You have a 15 % chance of PE.”
If we can throw a decimal point, even better. “You have a 15.2 % chance of PE.”
The rationale is that numbers empower patients to make a more informed choice, optimizing patient-centered medicine and improving outcomes.
Sounds reasonable enough. Although I find it difficult to believe that patients will have this conversation with their physicians.
“Thank god doctor my risk of PE is 15.1 % not 15.2 %. Otherwise I’d be in real trouble.”
What’s the allure of precision? Let’s understand certain terms: risk and uncertainty; prediction and prophesy.
By certainty I mean one hundred percent certainty. Opposite of certainty is uncertainty. Frank Knight, the economist, divided uncertainty to Knightian risk and Knightian uncertainty (1).
What’s Knightian risk?
If you toss a double-headed coin you’re certain of heads. If you toss a coin with head on one and tail on the other side, chance of a head is 50 %, assuming it’s a fair coin toss. Although you don’t know for certain that the toss will yield head or tail, you do know for certain that the chance of a head is 50 %. This can be verified by multiple tosses.
As the new year started, all kinds of predictions come to our attention, mostly of things that will enter our lives.
How about things that will dissolve from our lives ?
Of all species that became extinct the Dodo has become sort of synonymous with extinction. To “go the way the Dodo”means something is headed to go out of existence. (picture and quote source The Smithsonian)
So this goes not only for species but also stuff we use or things we do.
You might want to have a look at the extinction timeline and find things you did, ‘some’ time ago, and don’t anymore.
But what about health care? What will vanish, will the doctor due to all of this new technology disappear, or the nurse? Will we no longer go to a hospital or to the doctors office? I don’t think so.
We still will be needing professionals with compassion and care. However shift is happening and some things will start getting obsolete. In the following I am in no way going to try to be exhaustive, so feel free to add in comments or thought on what you think will disrupt from our lives in terms of health(care).