Hello, i had this idea last year on an alter-psychiatric forum, maybe we can join force here, because i can’t do that alone. And maybe this work could help you to analys your data of survey. For me, spectral analysis may well be “the intuition of the computer”, which could allow us to verify and prove our human intuitions.
the main problem is that to have robust data, it is necessary to be epidemiological (with an end which justifies the means: what one seeks is sponsored by the lab, the means is the money), and it is necessary to justify the why of his research, and without belonging to a research lab, no subsidy, it is very complicated to have access to the database of social security … Otherwise, we do as you will see in 4. a list where it is the bazaar.
here are some links where are available datas, and psychiatric databases around the world, in order to facilitate possible spectral analyzes: Summary:1. statistical tools of spectral analysis2. (co) variances to verify3. data by country4. data by “diseases”, “medicines”, other variables, metadata ---------------------------------------------------------------------------
These statistical tools make it possible to direct calculations and to give tracks thanks to the visual diagrams, which project in an abstract space which makes align the correlated variables, and thus to have ideas of correlations to deepen. To give an example, on an aerial image taken by satellite or plane, of the terrestrial surface, by sorting the distribution, local variations, and proximity / distance, (even very light) in the frequency space of the light spectrum (at each pixel of any photo, is assigned a frequency value of wavelength corresponding to a color) and testing, groping with the computer grinders, it can for example bring out, underground elements (foundations, pipes, etc. ) invisible to the naked eye, yet recorded as a watermark in the image data, a bit like a Fourier transform, http://villemin.gerard.free.fr/Wwwgvmm/Analyse/Fourier.htm but statistical, which decomposes (in several integrals) a single curve, in several sub-periodic and cyclic curves (that we can name harmonic, or polyphony if one wants) which add up to form only one (which apparently has no cycle or logic) in order to see which sub-curves are related.
Another example, admitted unanimously by the scientific community, that I know a little bit: it is the idea that Milankovitch had in 1941 (died in 1968, his theory was proven posthumously in 1976, thanks to new experimental technologies, such as the analysis of the isotope δ18O (which allows to deduce the temperature related to dated sediments) that was finally developed in 1967 by his colleagues, https://fr.wikipedia.org/wiki /% CE% 9418O, and used on boreholes in order to have finally the global curve of evolution of the climate !! and subsequently the satellite observation of the distribution of terrestrial insolation confirmed the theory, it seems to me), which found that the curve of temperature fluctuations over the geological ages was largely EXOGENOUS part. “However, the idea that the Earth’s climate itself can be influenced by astronomical parameters, so exogenous, and in addition cyclical, took a long time to impose, because this idea was not self-evident.” (https://fr.wikipedia.org/wiki/Them_astronomy_of_palatoms), the curve of climatic variations is due to variations of the planet Earth itself (these variations being induced by the interactions with the other planets, and the sun), decomposable into several periodic and cyclic sub-curves, each perfectly corresponding [/ i] and exactly to an orbital parameter of the Earth or astronomical cycle (inclination axis of rotation, equinox precession, solar activity, eccentricity, etc …).
if we statistically analyzed these curves, they would all be interconnected necessarily because they form by adding, one and the same curve, the temperature variation of the Earth, a curve which at first sight has no link, no cycle ( apart from that of glaciations, unexplainable until then), no possible interpretation. We could see that the precession of the equinoxes (cone forming the terrestrial axis in space) is necessarily very correlated with the inclination of the terrestrial axis …, but less correlated with the eccentricity of the ellipse of revolution. of the Earth around the Sun, and that all variables are highly correlated to Sun’s parameters. And to end this example, that the orbital parameters of the Earth are strongly correlated with the fluctuations of the climate. I hope that I can express myself well and make myself understood. So Milankovitch (not alone, of course, but he died just before his colleagues publish the confirmation …) established a thought experiment, a theory, confirmed later by spectral analysis … Transposable, in theory, on presumptions concerning a certain branch of medicine. So, Fourier transform, then AFC, ACP, It’s a bit like “statistical X-rays”. This can apply to things more abstract than a photo, on any data for example. 2. (co) variances to check:
For example, it serves as a basis for epidemiologists to decide on the direction of their study, I think, and if not, well, that’s very disturbing. One could see, in the myriad of factors projected in the statistical matrix, that depression and suicide are correlated, it could be seen that the mood regulator and bipolarity variables are correlated. This is obvious. As diabetes and insulin are correlated, for example. But we get diabetic before taking insulin. If one spontaneously cures a dependent insulin-dependent diabetes, it is impossible to know, because the diag is without appeal, and there is little chance of knowing it, since the treatment is without interruption, and never interrupted. Less obvious, we could see that: . Suicide and AD are correlated (the causal relationship is then to be defined, which of the chicken and the egg? or both, or associated / weighted to other factors, and for what part attributed to each of the factors, and and so on…), . AD and Bipolarity are correlated, causally in a majority sense, then causally circularly … to prove. list of variables to cross: - each “disease” - each “class of” medicine - nocebo effect, placebo effect - in psychiatry or not: . socioprofessional category: (active, unemployed, student, unemployed) . in psychiatry: (disability, ald, esat, hospital, etc …)), duration of “care” in the year (from the first diagnosis), number of different diagnoses posed, number of pharmaceutical classes taken at the same time, alliance therapeutic or not, etc. to be continued…
International Classification of Diseases, ICD-10: Medical coding support site combining CCAM data (managed by AMELI Health Insurance) and ICD-10 (managed by WHO) Here we have a non-raw database, in the form of links and graphics that is very interesting! This site can be very useful, it can make it possible to make statistical links between iatrogenése and pathology, and especially to visualize visual diagrams generated from the statistics of the PMSI (program of medicalization of information systems) French, pertaining to each diagnosis of ICD-10 [pathology diagnosis]: - the statistical proportion of each symptoms encountered associated with [pathology diagnosis] - the statistical proportion of each CCAM Acts (CCAM code) associated with [pathology diagnosis] - the statistical proportion of each homogeneous group of patients (GHM) associated with [pathology diagnosis] (see future post on ):
The Hospital Information Technology Agency (ATIH) provides free access summary data via the Scan Health health facility data retrieval platform: https://www.scansante.fr/
