Following from several other posts about gender bias, I offer this:
'Calm down dear, it’s only an aneurysm’ – why doctors need to take women’s pain seriously
Female heart-attack victims are half as likely as men to receive treatment. Is ‘hysteria’ still being used to deny women adequate medical care?
Scepticism toward the latter is costing lives: according to a study led by the University of Edinburgh and funded by the British Heart Foundation, women who had gone to A&E after experiencing chest pain (and were later found to be suffering from a heart attack) were half as likely as men to receive the recommended medical treatment. The research comes after it was revealed that entering identical heart symptoms for women and men on Babylon, a virtual GP app praised by the health secretary, Matt Hancock, resulted in different diagnoses. Its artificial intelligence tells a 60-year-old female smoker who reports chest pain and nausea that she is simply having a panic attack. A 60-year-old male smoker with exactly the same symptoms is told that he might be having a heart attack and is advised to go to A&E. Here’s hoping that the researchers from the University of Edinburgh are predominantly male, so that their research is taken more seriously than the anguished cries of women that have rung out since the beginning of time.
Rest of article freely available here:
theguardian.com/commentisfr...
Babylon health have published their own blog in response to the media coverage. Some points I take from it:
First of all, our Symptom Checker is absolutely not a diagnostic tool and we are very clear about this.
It doesn't matter how clear you are, people, including doctors, treat it as a diagnostic tool.
As soon as a doctor sees the short list, they will influenced by what is shown. Conscious attempts to avoid doing so almost certainly can't eliminate the impact of seeing the list.
Our Symptom Checker works on the basis of probability. It makes calculations based on lots of epidemiological data from a huge number of research studies. It then comes up with hundreds of possible matching conditions, from which it displays those which most closely match the symptoms entered.
As soon as Babylon applies statistics to select "best matching" from the list of all conditions, it is making decisions which are beyond questionable, they are life-threatening.
Even if we assume there are no issues with the data, best match is a simplistic, naive, unacceptable basis for making a selection.
I suggest (from my brain, not derived from research) that first cut perhaps should be identifying those conditions which need imemdiate action.
If the probabilities are derived from research, including collation of medicial statistics, it obviously is predicated on the quality of the research. Any bias in the research might well be reinforced rather than eliminated.
It has been a staple of medical literature (as in stories rather than research) that someone is a seemingly obvious case of X but ends up actually having Y, often due to some special factor Z. No, not a cold, but some rare infection because they had been in contact with some far off exotic country where Y occurs. What population statistics do is make it less and less likely that this unusual case will ever be diagnosed. Hence the use of such cases for dramatic effect. Usually hero doctor.
In this case, ever having been diagnosed with anxiety or depression (seemingly regardless of accuracy or identification of cause), would push heart attack off the "most likely" list and raise "panic attack".
I suggest that anxiety or depression caused by organic disease (e.g. hypothyroidism) which is now adequately treated or cured (we can hope!) is almost certainly inappropriate to be used as a guide to whether the person is now suffering heart attack or panic attack.
A patient who is familiar with panic attacks just might be particularly aware that they are currently suffering something which is not the same as their "usual" panic attacks. In that case, the simple fact that they are asking for medical help should be an indicator that is isn't a panic attack.
For every women told she is having a panic attack when really in the throes of a heart attack, how many will actually have the diagnosis reversed? That is, will the misdiagnosis be identified in such a way as to feedback into Babylon (or any other system) and apply a correction?
This goes down the same route as so much else, population statistics being applied to an individual case.
One of the reasons you can prove anything in economics is that so much that is written invokes ceteris paribus. Medicine often avoids saying that - often it appears to ignore it completely. Or assume everyone reading is applying it without having been told.
(Ceteris paribus or caeteris paribus is a Latin phrase meaning "other things equal"; English translations of the phrase include "all other things being equal" or "other things held constant" or "all else unchanged". From wiki.)
So, a woman aged 59 has a risk of a 44 year old man, all other things being equal.
Sorry folks, but there are so many differences between a 59 year old woman and a 44 year old man, this is fatuous. I understand this sort of statement as an illustration. But I simply do not believe you can make anything useful out of that statement.
Full blog freely available here:
babylonhealth.com/blog/tech...
Let is not forget that whatever we are, these issues can be critical. Until recently, there was virtually no mention, anywhere, of breast cancer in males. Statistically very much less likely but, if the patient in front of the doctor has it, that statistic is utterly irrelevant.
[ Have edited to add a few bits and correct some dreadful typing. ]