This morning a friend sent me a link to a story on ABC from the 2nd of June 2010 about the EVestG, a new device developed at Monash University in Melbourne, Australia, for “electrovestibulography”. This device has been developed by Brian Lithgow, an adjunct professor of Electrical & Computer Systems Engineering, and Jayashri Kulkarni, a professor of psychiatry. The ABC site says:
EVestG is a new diagnostic technique that measures the patterns of electrical activity in the brains vestibular (or balance) system allowing Doctors to dramatically fast-track the detection of mental and neurological illnesses.
Sounds great.
… But hold on. Why do I, a practising psychiatrist, first hear about this incredible psychiatric breakthrough from a friend who saw it on the ABC? That’s the wrong way around, and it should always raise one’s sceptical hackles. So what to do? Well I popped along to Pubmed and did a little search for evesg: 5 results. Did another search, for electrovestibulography: 6 results. Not exactly a wealth of citations for a device that says (852 kB pdf) it:
… can objectively and in real time:
o Separate a wide range of Central Nervous System [CNS] disorders from healthy normals, including schizophrenia, depression, Parkinsons disease, Menieres and Vertigo , and
o Measures the effects of drug and physical therapies on sufferers of CNS conditions.EVestG has application to multiple sclerosis, Parkinsonian type disorders, and autistic spectrum disorders as well as to Alzheimers disease and dementia.
That pdf is from the website of Neural Diagnostics – a Melbourne-based company that is:
…commercializing a breakthrough technique for assessing Central Nervous System (CNS) disorders. This non-invasive technology is called ElectroVestibulography (EVestG) which has the potential to enable clinicians to diagnose disorders such as schizophrenia, Meniere’s, Alzheimer’s and Parkinson’s disease, as well as depression and bi-polar disorders, in a shorter time than currently, using new bio-marker correlations objectively measured in one simple non-invasive and radiation free test.
Ok. So it’s on ABC, and it’s being “commercialised” as a “breakthrough technique” that “can objectively and in real time” give definitive diagnoses of various neuropsychiatric disorders. Awesome. It must have some pretty damn good evidence in those few papers then, right?
Oh, wait … papers? No … conference proceedings, not peer-reviewed published papers. Sorry for the confusion … In any case, what do they say? I only have the abstracts, as I’m not a member of the IEEE at which conference this device has been presented. The IEEE by the way is not a medical organisation; it’s a group concerned with various technological developments, though they do have an “Engineering in Biology and Medicine Society” at the 2007 and 2008 meetings of which this device was discussed. The psychiatrically relevant citations appear to be two: one for depression which compared 5 (five) depressed patients with 10 non-depressed controls, and one schizophrenia, which compared 4 (four) people with schizophrenia to 10 healthy controls. Some arcane differences in changes in various vestibular “dynamic measures” are said to differentiate between the groups. There are some figures in the Neural Diagnostics pdf linked above. that are worth looking at. They purport to show how distinguishable are various diagnostic groups, and healthy controls. I suggest having a look, and in all seriousness, if you can explain to me what either the x “biomarker-BGi-onBB correlation (L+R+LR)” or y “biomarker-FP interval change correlation (L+R+LR)” axes mean, please do. One of the graphs has different x and y axes: “biomarker related to dynamic range performance” vs “biomarker related to firing interval pattern”. Seems odd to me. While I’m not a neurophysiologist, I would have thought that if there were a strong relationship between the neurotransmission in the brain responsible for these neuro-psychiatric disorders and the vestibular system (which is the basis on which they present this as working), you wouldn’t need such jiggery-pokery to demonstrate it.
It’s perhaps also worthy of note that in the report on electrovestibulography in schizophrenia, they mention that electrovestibulography is “a new technique similar to ECOG”. With a minute of searching I found this interesting statement about ECOG (electrocochleography, or ECochG):
ECochG testing is technically difficult, has a questionable rationale, lacks reliable norms, lacks standards, and is vulnerable to operator bias.
Given that the technique as described seems very similar, one has to wonder how they would avoid those same problems.
Look, if this turns out to work, I’d actually be very pleased. It’d be great to have an objective test for some of the disorders we treat. However, our desire for that should never be allowed to overtake the science. Here what we have is a rather left-field idea, with very little (and not at all replicated) evidence to support it, that is being hawked via news media, and already has a company that is “commercializing” it. Not exactly auspicious. Certainly in reality there’s nothing to get excited about – despite the impressive credentials of the psychiatrist involved (and possibly those of the engineer – I just don’t know of him).
Popularity: 15% [?]
Electrovestibulography is a wonderfully long word. I'm not even sure it would fit on a scrabble board, even if it were allowed.
I had an idea you might like the word
Electrovestibulography – 'to wear a vest with an electronic picture on the back' (kind of like those tshirts nerds wear to raves)
Jiggery pokery is right. Where would you start to deconstruct this nonsense? This post will be a bit of a mish mash, dare I admit, a ramble.
The figure 700 has been bandied around. Well apparently of "schizophrenics" there have been 22. All being treated with meuroleptics that cause some degree of dystonia and akithesia.
Depressives? Last night Kulkarni (who I've worked with… sheesh) blurted out – 100% sensitivity and specificity for depression. Lithgow (who seems the be the brains(?) of the outfit because Kulkarni certainly isn't) then piped up with this 700 figure. Puhleeze.
Let them even dare suggest any kind of sensitivity and specificity can be applied to diagnoses such as schizophrenia and bipolar disorder, treatments for which are often imposed by way of force and which can totally destroy a persons life.
Just a bit about the curves. If you ran the tilt chair on all members of the St Kilda and Collingwood players and then "averaged" the curves by team you would have two different curves. How can these curves have any predictive value? If you tilt test a rookie player in a VFL club will you be able to predict that he will one day be recruited by St Kilda rather than Collingwood, or vice versa? This is what these people are claiming.
OK they say that they also have bio-markers which demonstrate the validity of the curves as having predictive ability. But they have arrived at these bio-markers by lengthy trial and error of the mechanics and timing of the tilt sequence and the selective selection of subjects.
The average person thinks that lengthy trial and error increases validity when it is actually the reverse. The researchers are actually systematically indulging in data mining and confirmation bias. And the gullible public lap it up.
And everywhere I look these commercial researchers have already lumped behavioural disorders in with known neurological conditions… I gotta stop here… uuurrgh.
I whole heartedly agree with Dr Fraser’s comments on this machine. Out of the blue comes a miracle diagnostic technique that is set to revolutionize diagnosis in Psychiatry. On the basis of one curve we are now going to be put into neat diagnostic categories. Maybe if it is true it will be for the best but somehow I don’t think so. The etiology and manifest symptomology of psychiatric conditions is far more complicated than that and leaving it up to a mere curve generated by a machine that has not as yet been subject to proper scientific scrutiny is gravely concerning.
Do I detect quackery here, not necessarily, just that I think the whole notion of a simple machine diagnosis too good to be true and the onus is definitely on the inventors to prove humankind otherwise. Bye the way I suffer from depression. Would I submit myself to one of those tests? Put it this way the doctor may be bemused by my impolite refusal.
….we already look at the function of the balance system using electrical potentials; vestibular evoked myogenic potentials and electronystagmography. All this time we have been misdiagnosing patients as having vestibular dysfunction when we really should have been sending them straight to a psychiatrist.