%PDF-1.7 1 0 obj << /Type /Catalog /Outlines 2 0 R /Pages 3 0 R >> endobj 2 0 obj << /Type /Outlines /Count 0 >> endobj 3 0 obj << /Type /Pages /Kids [6 0 R 29 0 R 33 0 R 35 0 R 41 0 R ] /Count 5 /Resources << /ProcSet 4 0 R /Font << /F1 8 0 R /F2 9 0 R /F3 10 0 R >> /XObject << /I1 39 0 R /I2 40 0 R >> /ExtGState << /GS1 25 0 R /GS2 26 0 R /GS3 27 0 R /GS4 28 0 R >> >> /MediaBox [0.000 0.000 595.280 841.890] >> endobj 4 0 obj [/PDF /Text /ImageC ] endobj 5 0 obj << /Producer ( d o m p d f 1 . 0 . 2 + C P D F) /CreationDate (D:20240329103542+00'00') /ModDate (D:20240329103542+00'00') >> endobj 6 0 obj << /Type /Page /MediaBox [0.000 0.000 595.280 841.890] /Parent 3 0 R /Annots [ 11 0 R 13 0 R 15 0 R 17 0 R 19 0 R 21 0 R 23 0 R ] /Contents 7 0 R >> endobj 7 0 obj << /Length 6506 >> stream 0.000 0.000 0.000 rg BT 35.000 736.932 Td /F1 21.0 Tf [(\(21\) Circle Instead of Hierarchy in a Bayesian Analysis)] TJ ET BT 35.000 697.058 Td /F2 14.0 Tf [(Description)] TJ ET BT 35.000 652.299 Td /F2 14.0 Tf [(How the Circular Model Works in Concrete Terms and Why It Is Superior to the)] TJ ET BT 35.000 635.619 Td /F2 14.0 Tf [(Hierarchical One The Example of the Ketogenic Diet)] TJ ET BT 35.000 606.839 Td /F1 12.0 Tf [(Just now )] TJ ET 0.000 0.000 0.800 rg BT 80.336 606.839 Td /F1 12.0 Tf [(a new publication of ours)] TJ ET 0.000 0.000 0.800 RG 0.6 w 0 J [ ] 0 d 80.336 604.259 m 202.316 604.259 l S 0.000 0.000 0.000 rg BT 202.316 606.839 Td /F1 12.0 Tf [( appeared in an online journal, in which we demonstrated how to implement)] TJ ET BT 35.000 592.583 Td /F1 12.0 Tf [(the circular model of cognition in concrete terms [1]. The physicist )] TJ ET 0.000 0.000 0.800 rg BT 359.612 592.583 Td /F1 12.0 Tf [(Rainer Klement)] TJ ET 0.6 w 0 J [ ] 0 d 359.612 590.003 m 435.932 590.003 l S 0.000 0.000 0.000 rg BT 435.932 592.583 Td /F1 12.0 Tf [(, who calculated the analysis)] TJ ET BT 35.000 578.327 Td /F1 12.0 Tf [(and provided the example data, was in charge. I was able to inspire him with my idea that one gains more insight)] TJ ET BT 35.000 564.071 Td /F1 12.0 Tf [(with a circular model of knowledge than with the traditional, hierarchical one of evidence based medicine)] TJ ET BT 35.000 549.815 Td /F1 12.0 Tf [(\(EBM\). The key to this could be the formalization made here, which adopts a Bayesian statistical approach. It)] TJ ET BT 35.000 535.559 Td /F1 12.0 Tf [(sounds more complicated than it is.)] TJ ET BT 35.000 505.343 Td /F2 12.0 Tf [(Background and problem)] TJ ET BT 35.000 475.127 Td /F1 12.0 Tf [(The )] TJ ET BT 56.660 475.127 Td /F3 12.0 Tf [(hierarchical model)] TJ ET BT 148.316 475.127 Td /F1 12.0 Tf [( of EBM, the current mainstream model, assumes that randomised controlled trials, if)] TJ ET BT 35.000 460.871 Td /F1 12.0 Tf [(possible with placebo control, provide the best available evidence on whether an intervention works or not.)] TJ ET BT 35.000 446.615 Td /F1 12.0 Tf [(Implicitly, this sets )] TJ ET BT 129.680 446.615 Td /F3 12.0 Tf [(internal validity)] TJ ET BT 206.684 446.615 Td /F1 12.0 Tf [(, the methodological rigour and validity of a study, )] TJ ET BT 453.668 446.615 Td /F3 12.0 Tf [(against external validity)] TJ ET BT 570.332 446.615 Td /F1 12.0 Tf [(,)] TJ ET BT 35.000 432.359 Td /F1 12.0 Tf [(the transferability, usefulness and generalisability of findings in practice. I have analysed these problems in the)] TJ ET BT 35.000 418.103 Td /F1 12.0 Tf [(first three parts of my methodology \()] TJ ET 0.000 0.000 0.800 rg BT 212.312 418.103 Td /F1 12.0 Tf [(Part 1: Evidence, an Unreflected Battle Cry)] TJ ET 0.6 w 0 J [ ] 0 d 212.312 415.523 m 421.268 415.523 l S 0.000 0.000 0.000 rg BT 421.268 418.103 Td /F1 12.0 Tf [(; )] TJ ET 0.000 0.000 0.800 rg BT 427.604 418.103 Td /F1 12.0 Tf [(Part 2: Hierarchy or a Circle of )] TJ ET 0.6 w 0 J [ ] 0 d 427.604 415.523 m 579.896 415.523 l S BT 35.000 403.847 Td /F1 12.0 Tf [(Evidence)] TJ ET 0.6 w 0 J [ ] 0 d 35.000 401.267 m 79.652 401.267 l S 0.000 0.000 0.000 rg BT 79.652 403.847 Td /F1 12.0 Tf [(; )] TJ ET 0.000 0.000 0.800 rg BT 85.988 403.847 Td /F1 12.0 Tf [(Part 3: The Consequences of the Hierarchical and Circular Models)] TJ ET 0.6 w 0 J [ ] 0 d 85.988 401.267 m 405.584 401.267 l S 0.000 0.000 0.000 rg BT 405.584 403.847 Td /F1 12.0 Tf [(\).)] TJ ET BT 35.000 377.591 Td /F1 12.0 Tf [(In practice, this leads to systematic reviews and meta-analyses disregarding most of the data and often even)] TJ ET BT 35.000 363.335 Td /F1 12.0 Tf [(saying that there is no scientific evidence. Therefore, medical guidelines or meta-analyses often contain)] TJ ET BT 35.000 349.079 Td /F1 12.0 Tf [(recommendations that contradict clinical experience or do not take into account a lot of insights [2]. The problem,)] TJ ET BT 35.000 334.823 Td /F1 12.0 Tf [(as we have analysed in other publications [3, 4] and presented in more detail in )] TJ ET 0.000 0.000 0.800 rg BT 417.596 334.823 Td /F1 12.0 Tf [(part 18 of the methodology blog)] TJ ET 0.6 w 0 J [ ] 0 d 417.596 332.243 m 572.588 332.243 l S 0.000 0.000 0.000 rg BT 572.588 334.823 Td /F1 12.0 Tf [(,)] TJ ET BT 35.000 320.567 Td /F1 12.0 Tf [(is that internal and external evidence are independent of each other. One cannot pretend that one is more)] TJ ET BT 35.000 306.311 Td /F1 12.0 Tf [(important than the other or that one presupposes the other. They dont. Rather, there are studies that maximize)] TJ ET BT 35.000 292.055 Td /F1 12.0 Tf [(internal validity randomized trials and those that maximize external validity all naturalistic studies.)] TJ ET BT 35.000 265.799 Td /F1 12.0 Tf [(Now, if you neglect naturalistic studies in favour of randomized ones, as it is being done at the moment, you risk)] TJ ET BT 35.000 251.543 Td /F1 12.0 Tf [(generating extremely reliable knowledge, but knowledge that either has very limited applicability or that nobody)] TJ ET BT 35.000 237.287 Td /F1 12.0 Tf [(cares about. This is why we have proposed the circular model, which does not favour any type of information or)] TJ ET BT 35.000 223.031 Td /F1 12.0 Tf [(study, but assumes that all studies provide different types of information that are relevant to different questions)] TJ ET BT 35.000 208.775 Td /F1 12.0 Tf [(and should therefore all be taken into account. Which is what happens in the circular model of evidence.)] TJ ET BT 35.000 194.519 Td /F1 12.0 Tf [(Foremost, I think this is a plausible theoretical demand. But the question is: How can this be achieved in concrete)] TJ ET BT 35.000 180.263 Td /F1 12.0 Tf [(terms?)] TJ ET 0.502 0.502 0.502 rg 0.502 0.502 0.502 RG 35.000 781.760 m 585.280 781.760 l 585.280 783.260 l 35.000 783.260 l f 0.000 0.000 0.000 rg BT 416.240 816.290 Td /F2 10.0 Tf [(PROF. DR. DR. HARALD WALACH)] TJ ET BT 377.370 804.386 Td /F1 10.0 Tf [(https://harald-walach.de https://harald-walach.info)] TJ ET q 0.940 0.342 -0.342 0.940 181.247 -73.446 cm /GS1 gs /GS2 gs 0.6 w 0 J [ ] 0 d Q /GS3 gs /GS4 gs 0.502 0.502 0.502 rg 0.502 0.502 0.502 RG 35.000 52.000 m 585.280 52.000 l 585.280 50.500 l 35.000 50.500 l f 0.000 0.000 0.000 rg BT 44.000 31.996 Td /F1 10.0 Tf [(Page 1)] TJ ET q 44.000 17.740 532.280 11.880 re W n 0.000 0.000 0.000 rg BT 263.855 20.116 Td /F1 10.0 Tf [( Prof. Harald Walach)] TJ ET Q endstream endobj 8 0 obj << /Type /Font /Subtype /Type1 /Name /F1 /BaseFont /Times-Roman /Encoding /WinAnsiEncoding >> endobj 9 0 obj << /Type /Font /Subtype /Type1 /Name /F2 /BaseFont /Times-Bold /Encoding /WinAnsiEncoding >> endobj 10 0 obj << /Type /Font /Subtype /Type1 /Name /F3 /BaseFont /Times-Italic /Encoding /WinAnsiEncoding >> endobj 11 0 obj << /Type /Annot /Subtype /Link /A 12 0 R /Border [0 0 0] /H /I /Rect [ 80.3360 605.7586 202.3160 617.6386 ] >> endobj 12 0 obj << /Type /Action /S /URI /URI (https://tbiomed.biomedcentral.com/articles/10.1186/s12976-018-0084-y) >> endobj 13 0 obj << /Type /Annot /Subtype /Link /A 14 0 R /Border [0 0 0] /H /I /Rect [ 359.6120 591.5026 435.9320 603.3826 ] >> endobj 14 0 obj << /Type /Action /S /URI /URI (http://www.rainerklement.de) >> endobj 15 0 obj << /Type /Annot /Subtype /Link /A 16 0 R /Border [0 0 0] /H /I /Rect [ 212.3120 417.0226 421.2680 428.9026 ] >> endobj 16 0 obj << /Type /Action /S /URI /URI (https://harald-walach.info/methodology-for-beginners/1-evidence-an-unreflected-battle-cry-2/) >> endobj 17 0 obj << /Type /Annot /Subtype /Link /A 18 0 R /Border [0 0 0] /H /I /Rect [ 427.6040 417.0226 579.8960 428.9026 ] >> endobj 18 0 obj << /Type /Action /S /URI /URI (https://harald-walach.info/methodology-for-beginners/2-hierarchy-or-a-circle-of-evidence/) >> endobj 19 0 obj << /Type /Annot /Subtype /Link /A 20 0 R /Border [0 0 0] /H /I /Rect [ 35.0000 402.7666 79.6520 414.6466 ] >> endobj 20 0 obj << /Type /Action /S /URI /URI (https://harald-walach.info/methodology-for-beginners/2-hierarchy-or-a-circle-of-evidence/) >> endobj 21 0 obj << /Type /Annot /Subtype /Link /A 22 0 R /Border [0 0 0] /H /I /Rect [ 85.9880 402.7666 405.5840 414.6466 ] >> endobj 22 0 obj << /Type /Action /S /URI /URI (https://harald-walach.info/methodology-for-beginners/3-the-consequences-of-the-hierarchical-and-circular-models/) >> endobj 23 0 obj << /Type /Annot /Subtype /Link /A 24 0 R /Border [0 0 0] /H /I /Rect [ 417.5960 333.7426 572.5880 345.6226 ] >> endobj 24 0 obj << /Type /Action /S /URI /URI (https://harald-walach.info/methodology-for-beginners/18-why-the-hierarchical-model-of-the-evidence-based-medicine-movement-falls-short/) >> endobj 25 0 obj << /Type /ExtGState /BM /Normal /CA 0.3 >> endobj 26 0 obj << /Type /ExtGState /BM /Normal /ca 0.3 >> endobj 27 0 obj << /Type /ExtGState /BM /Normal /CA 1 >> endobj 28 0 obj << /Type /ExtGState /BM /Normal /ca 1 >> endobj 29 0 obj << /Type /Page /MediaBox [0.000 0.000 595.280 841.890] /Parent 3 0 R /Annots [ 31 0 R ] /Contents 30 0 R >> endobj 30 0 obj << /Length 6961 >> stream 0.6 w 0 J [ ] 0 d 0.000 0.000 0.000 rg BT 35.000 748.485 Td /F1 12.0 Tf [(We have now provided the key. With a concrete example of application in a controversial topic. The application)] TJ ET BT 35.000 734.229 Td /F1 12.0 Tf [(of the circular model succeeds if one adopts a Bayesian formal analytic approach that allows studies to change)] TJ ET BT 565.220 734.229 Td /F1 12.0 Tf [(our)] TJ ET BT 35.000 719.973 Td /F1 12.0 Tf [(prior knowledge while giving different weight to different types of studies. The advantage of a Bayesian)] TJ ET BT 536.240 719.973 Td /F1 12.0 Tf [(analysis)] TJ ET BT 35.000 705.717 Td /F1 12.0 Tf [(is that it incorporates our prior knowledge, which is generated by different data, into the analysis. I have)] TJ ET BT 534.908 705.717 Td /F1 12.0 Tf [(discussed)] TJ ET BT 35.000 691.461 Td /F1 12.0 Tf [(this before in a post on methodology \()] TJ ET 0.000 0.000 0.800 rg BT 217.988 691.461 Td /F1 12.0 Tf [(Part 5: On the Relationship between Empiricism and Theory 1)] TJ ET 0.000 0.000 0.800 RG 0.6 w 0 J [ ] 0 d 217.988 688.881 m 517.628 688.881 l S 0.000 0.000 0.000 rg BT 517.628 691.461 Td /F1 12.0 Tf [(\),)] TJ ET BT 524.624 691.461 Td /F1 12.0 Tf [(so I will)] TJ ET BT 35.000 677.205 Td /F1 12.0 Tf [(keep it short with a brief reminder.)] TJ ET BT 35.000 646.989 Td /F2 12.0 Tf [(Bayesian analysis)] TJ ET BT 35.000 616.773 Td /F1 12.0 Tf [(The Irish pastor Bayes had recognized, that we make decisions based on prior knowledge, which changes the)] TJ ET BT 35.000 602.517 Td /F1 12.0 Tf [(more information we have. The analysis named after him asks: if I have certain prior knowledge, how strong)] TJ ET BT 35.000 588.261 Td /F1 12.0 Tf [(must empirical knowledge be in one direction or another to change it? Or, in other words, if all the prior)] TJ ET BT 35.000 574.005 Td /F1 12.0 Tf [(knowledge we have is taken into account, how strong is the influence of a particular study or experience?)] TJ ET BT 35.000 559.749 Td /F1 12.0 Tf [(Bayesian analysis, then, unlike classical frequentist statistics, works with conditional probabilities. It formalizes)] TJ ET BT 35.000 545.493 Td /F1 12.0 Tf [(prior knowledge as the so-called prior probability or initial probability, adds a new study result, and then asks)] TJ ET BT 35.000 531.237 Td /F1 12.0 Tf [(how this new result changes this initial probability into the so-called posterior probability or final probability.)] TJ ET BT 35.000 504.981 Td /F1 12.0 Tf [(We humans are all Bayesians. For we have all had formal or informal experience. Science also implicitly takes a)] TJ ET BT 35.000 490.725 Td /F1 12.0 Tf [(Bayesian approach. The prevailing world view, professional or scientific experience, they all shape an implicit)] TJ ET BT 35.000 476.469 Td /F1 12.0 Tf [(initial probability on the basis of which all available data, new study results or experiences are interpreted and)] TJ ET BT 35.000 462.213 Td /F1 12.0 Tf [(weighted. Bayesian analysis now simply formalizes this procedure, which we always adopt anyway.)] TJ ET BT 35.000 435.957 Td /F1 12.0 Tf [(Classical Fisherian or frequentist statistics imitates the special case, which never actually occurs, that we are)] TJ ET BT 35.000 421.701 Td /F1 12.0 Tf [(completely undecided on a particular question because we have no prior knowledge whatsoever, the prior)] TJ ET BT 35.000 407.445 Td /F1 12.0 Tf [(probability, or the initial probability is therefore 50:50 or . Only if this is the case is frequentist statistics)] TJ ET BT 35.000 393.189 Td /F1 12.0 Tf [(actually applicable in the strict case, otherwise not. Wagenmakers and colleagues rightly pointed this out in the)] TJ ET BT 35.000 378.933 Td /F1 12.0 Tf [(example of parapsychology [5]. However, this should not only apply to psychology, but in principle to all)] TJ ET BT 35.000 364.677 Td /F1 12.0 Tf [(statistics, including the statistics with which clinical studies are evaluated.)] TJ ET BT 35.000 338.421 Td /F1 12.0 Tf [(We have just set up a circular synthesis model with the help of Bayesian statistics for an example that is being)] TJ ET BT 35.000 324.165 Td /F1 12.0 Tf [(discussed very controversially at the moment. Its about the ketogenic diet for high-grade glioma, a difficult-to-)] TJ ET BT 35.000 309.909 Td /F1 12.0 Tf [(treat type of brain tumour with a very poor prognosis.)] TJ ET BT 35.000 279.693 Td /F2 12.0 Tf [(The clinical example: ketogenic diet for brain tumours)] TJ ET BT 35.000 249.477 Td /F1 12.0 Tf [(The term ketogenic diet refers to a diet that mimics fasting metabolism, simply put. I will spare the biochemical)] TJ ET BT 35.000 235.221 Td /F1 12.0 Tf [(and physiological background now. They are set apart in the original publication and in another publication of)] TJ ET BT 35.000 220.965 Td /F1 12.0 Tf [(ours, which is also available online [6]. When we fast, the body breaks down fat. In the process, short-chain fatty)] TJ ET BT 35.000 206.709 Td /F1 12.0 Tf [(acids, so-called ketone bodies, are produced. Most of the bodys cells, including nerve cells, can produce)] TJ ET BT 35.000 192.453 Td /F1 12.0 Tf [(energy from these ketone bodies. And the few that cannot are supplied with sugar, which is formed in the liver)] TJ ET BT 35.000 178.197 Td /F1 12.0 Tf [(from lactate, glycerol or glucogenic amino acids. That is why we do not die or faint when we fast, but can keep it)] TJ ET BT 35.000 163.941 Td /F1 12.0 Tf [(up for quite a long time. People who are not used to it easily fall into hypoglycaemia at first, but that is another)] TJ ET BT 35.000 149.685 Td /F1 12.0 Tf [(story.)] TJ ET BT 35.000 123.429 Td /F1 12.0 Tf [(Anyway, the body can be well nourished in a fasting mode if it feeds on ketone bodies from its own reserves.)] TJ ET BT 35.000 109.173 Td /F1 12.0 Tf [(Every night we do this without having to get up and spread a sandwich, so we dont starve. The ketogenic diet)] TJ ET BT 35.000 94.917 Td /F1 12.0 Tf [(now takes advantage of this, except that the body does not fall back on its own reserves, but on protein and fat)] TJ ET BT 35.000 80.661 Td /F1 12.0 Tf [(supplied through food. The ketogenic diet is therefore a diet in which carbohydrates are largely avoided and)] TJ ET 0.502 0.502 0.502 rg 0.502 0.502 0.502 RG 35.000 781.760 m 585.280 781.760 l 585.280 783.260 l 35.000 783.260 l f 0.000 0.000 0.000 rg BT 416.240 816.290 Td /F2 10.0 Tf [(PROF. DR. DR. HARALD WALACH)] TJ ET BT 377.370 804.386 Td /F1 10.0 Tf [(https://harald-walach.de https://harald-walach.info)] TJ ET q 0.940 0.342 -0.342 0.940 181.247 -73.446 cm /GS1 gs /GS2 gs 0.6 w 0 J [ ] 0 d Q /GS3 gs /GS4 gs 0.502 0.502 0.502 rg 0.502 0.502 0.502 RG 35.000 52.000 m 585.280 52.000 l 585.280 50.500 l 35.000 50.500 l f 0.000 0.000 0.000 rg BT 44.000 31.996 Td /F1 10.0 Tf [(Page 2)] TJ ET q 44.000 17.740 532.280 11.880 re W n 0.000 0.000 0.000 rg BT 263.855 20.116 Td /F1 10.0 Tf [( Prof. Harald Walach)] TJ ET Q endstream endobj 31 0 obj << /Type /Annot /Subtype /Link /A 32 0 R /Border [0 0 0] /H /I /Rect [ 217.9880 690.3812 517.6280 702.2612 ] >> endobj 32 0 obj << /Type /Action /S /URI /URI (https://harald-walach.info/methodology-for-beginners/5-on-the-relationship-between-empiricism-and-theory-part-1/) >> endobj 33 0 obj << /Type /Page /MediaBox [0.000 0.000 595.280 841.890] /Parent 3 0 R /Contents 34 0 R >> endobj 34 0 obj << /Length 4817 >> stream 0.6 w 0 J [ ] 0 d 0.000 0.000 0.000 rg BT 35.000 760.485 Td /F1 12.0 Tf [(nutrition is provided mainly through fat and protein intake and carbohydrates mainly in complex form \(e.g. salad)] TJ ET BT 35.000 746.229 Td /F1 12.0 Tf [(and vegetables\). Such a diet has proven successful for some neurological diseases, such as epilepsy. However, it)] TJ ET BT 35.000 731.973 Td /F1 12.0 Tf [(is also used for cancer [6, 7]. This is because most cancer cells depend on sugar, which they get directly from)] TJ ET BT 35.000 717.717 Td /F1 12.0 Tf [(food, and cannot feed on ketone bodies. So the ketogenic diet is something of a food deprivation programme for)] TJ ET BT 35.000 703.461 Td /F1 12.0 Tf [(cancer cells. It has been proven to work in many ways [7].)] TJ ET BT 35.000 677.205 Td /F1 12.0 Tf [(We have now chosen ketogenic diets for aggressive gliomas as an example because there is very little and)] TJ ET BT 35.000 662.949 Td /F1 12.0 Tf [(insufficient information here, precisely in terms of the hierarchical model. This is because there are only three)] TJ ET BT 35.000 648.693 Td /F1 12.0 Tf [(studies in humans and these are rather small, partly compared with complex other procedures or are only)] TJ ET BT 35.000 634.437 Td /F1 12.0 Tf [(available as observational data, i.e. not coming from a randomized study. Therefore, a classical reviewer would)] TJ ET BT 35.000 620.181 Td /F1 12.0 Tf [(conclude: there are no scientific findings. Therefore, the therapy cannot be recommended. But if you take the)] TJ ET BT 35.000 605.925 Td /F1 12.0 Tf [(existing 17 animal experiments and the data from the human studies together and add the basic physiological)] TJ ET BT 35.000 591.669 Td /F1 12.0 Tf [(considerations, which we cannot simply ignore, then the picture changes.)] TJ ET BT 35.000 561.453 Td /F2 12.0 Tf [(The analysis and the insight)] TJ ET BT 35.000 531.237 Td /F1 12.0 Tf [(If, as the circular model suggests, we take all the data, i.e. the 3 human studies and the 17 animal studies together,)] TJ ET BT 35.000 516.981 Td /F1 12.0 Tf [(we come to a different conclusion. We can formalize different considerations in such an analysis. Here, for)] TJ ET BT 35.000 502.725 Td /F1 12.0 Tf [(example, parameters are used for which assumptions have to be made, the effects of which can then be seen)] TJ ET BT 35.000 488.469 Td /F1 12.0 Tf [(directly in sensitivity analyses. For example, we assumed parameters for how important mechanistic)] TJ ET BT 35.000 474.213 Td /F1 12.0 Tf [(considerations are, i.e. the theoretical knowledge about how ketogenic diets affect gliomas. Then we put a)] TJ ET BT 35.000 459.957 Td /F1 12.0 Tf [(parameter into the model that formalizes how uniform the effects are across certain classes of individuals )] TJ ET BT 35.000 445.701 Td /F1 12.0 Tf [(humans, mice, rats. Finally, we formalized beliefs, such as the sceptical belief that data from animals are not)] TJ ET BT 35.000 431.445 Td /F1 12.0 Tf [(transferable to humans, or even the fundamentally sceptical belief that ketogenic diets are harmful. Finally, one)] TJ ET BT 35.000 417.189 Td /F1 12.0 Tf [(can include other mechanistic considerations, namely the finding that ketogenic diets support other forms of)] TJ ET BT 35.000 402.933 Td /F1 12.0 Tf [(therapy, such as radiation or chemotherapy, in that the diet weakens the tumour and makes it more susceptible to)] TJ ET BT 35.000 388.677 Td /F1 12.0 Tf [(the radical stress of the therapy. If you had very different data from human medical trials for example, large)] TJ ET BT 35.000 374.421 Td /F1 12.0 Tf [(cohort studies, randomized trial and case-control studies then you could have added parameters to control the)] TJ ET BT 35.000 360.165 Td /F1 12.0 Tf [(weighting of each study.)] TJ ET BT 35.000 333.909 Td /F1 12.0 Tf [(All these parameters now influence how the individual studies are accounted for in an overall model and how the)] TJ ET BT 35.000 319.653 Td /F1 12.0 Tf [(statement of the analysis is evaluated. I reproduce the results figure from the original publication here:)] TJ ET 0.502 0.502 0.502 rg 0.502 0.502 0.502 RG 35.000 781.760 m 585.280 781.760 l 585.280 783.260 l 35.000 783.260 l f 0.000 0.000 0.000 rg BT 416.240 816.290 Td /F2 10.0 Tf [(PROF. DR. DR. HARALD WALACH)] TJ ET BT 377.370 804.386 Td /F1 10.0 Tf [(https://harald-walach.de https://harald-walach.info)] TJ ET q 0.940 0.342 -0.342 0.940 181.247 -73.446 cm /GS1 gs /GS2 gs 0.6 w 0 J [ ] 0 d Q /GS3 gs /GS4 gs 0.502 0.502 0.502 rg 0.502 0.502 0.502 RG 35.000 52.000 m 585.280 52.000 l 585.280 50.500 l 35.000 50.500 l f 0.000 0.000 0.000 rg BT 44.000 31.996 Td /F1 10.0 Tf [(Page 3)] TJ ET q 44.000 17.740 532.280 11.880 re W n 0.000 0.000 0.000 rg BT 263.855 20.116 Td /F1 10.0 Tf [( Prof. Harald Walach)] TJ ET Q endstream endobj 35 0 obj << /Type /Page /MediaBox [0.000 0.000 595.280 841.890] /Parent 3 0 R /Annots [ 37 0 R ] /Contents 36 0 R >> endobj 36 0 obj << /Length 3714 >> stream 0.6 w 0 J [ ] 0 d q 490.280 0 0 373.303 65.000 398.587 cm /I2 Do Q 0.000 0.000 0.000 rg BT 65.000 387.182 Td /F1 12.0 Tf [(Figure Estimates of median survival with a glioma on a ketogenic diet \(KD\), ketogenic diet with)] TJ ET BT 65.000 372.926 Td /F1 12.0 Tf [(add-on therapy \(KD+\), or caloric restriction \(partial fasting; CR\) with \(+\) and without add-on)] TJ ET BT 65.000 358.670 Td /F1 12.0 Tf [(therapy. SP: sceptical baseline probability; FSP: fundamentally sceptical baseline probability \(i.e.,)] TJ ET BT 65.000 344.414 Td /F1 12.0 Tf [(assumption of harm from KD\); RP: various assumptions about the associations between the effects)] TJ ET BT 65.000 330.158 Td /F1 12.0 Tf [(of different species \(mice, rats, humans\); MP: various assumptions about mechanistic associations)] TJ ET BT 65.000 315.902 Td /F1 12.0 Tf [(between ketogenic diet and other therapies; EP: enthusiastic expectation.)] TJ ET BT 35.000 289.646 Td /F1 12.0 Tf [(The most important insight from this analysis is probably that all the data indicate that a ketogenic diet confers a)] TJ ET BT 35.000 275.390 Td /F1 12.0 Tf [(slight survival benefit. The estimates range from 1.2 to 1.5 on a ketogenic diet and from 1.5 to 1.7 on a caloric)] TJ ET BT 35.000 261.134 Td /F1 12.0 Tf [(restriction diet with an add-on treatment. So those on a ketogenic diet are about 20-50% more likely to survive.)] TJ ET BT 35.000 246.878 Td /F1 12.0 Tf [(Admittedly, none of these findings is certain in the strict sense, because the confidence interval in each case)] TJ ET BT 35.000 232.622 Td /F1 12.0 Tf [(includes the line of indecision, 1, even if the most optimistic estimates already come close. But it is amazing how)] TJ ET BT 35.000 218.366 Td /F1 12.0 Tf [(close together the estimates are, even when modelling sceptics, i.e. the sceptical priors \(the first three lines in the)] TJ ET BT 35.000 204.110 Td /F1 12.0 Tf [(figure\). The fact that the findings are so widely scattered shows that there is still relatively little data available)] TJ ET BT 35.000 189.854 Td /F1 12.0 Tf [(and the uncertainty is great. However, the fact that the estimation points are all relatively close to each other)] TJ ET BT 35.000 175.598 Td /F1 12.0 Tf [(shows that all the data point in the same direction.)] TJ ET BT 35.000 149.342 Td /F1 12.0 Tf [(So the conclusion of our analysis would be: ketogenic diet and caloric restriction is definitely promising. The)] TJ ET BT 35.000 135.086 Td /F1 12.0 Tf [(therapy promises to increase survival by 50%, and by 20% in the worst case, and should definitely be further)] TJ ET BT 35.000 120.830 Td /F1 12.0 Tf [(investigated. Most importantly, we have shown that and how studies of different types can be combined in a)] TJ ET BT 35.000 106.574 Td /F1 12.0 Tf [(formalized, quantitative analytical model.)] TJ ET BT 35.000 80.318 Td /F1 12.0 Tf [(Now, of course, we hope that the momentum will be taken up and that innovative minds in the Cochrane and)] TJ ET 0.502 0.502 0.502 rg 0.502 0.502 0.502 RG 35.000 781.760 m 585.280 781.760 l 585.280 783.260 l 35.000 783.260 l f 0.000 0.000 0.000 rg BT 416.240 816.290 Td /F2 10.0 Tf [(PROF. DR. DR. HARALD WALACH)] TJ ET BT 377.370 804.386 Td /F1 10.0 Tf [(https://harald-walach.de https://harald-walach.info)] TJ ET q 0.940 0.342 -0.342 0.940 181.247 -73.446 cm /GS1 gs /GS2 gs 0.6 w 0 J [ ] 0 d Q /GS3 gs /GS4 gs 0.502 0.502 0.502 rg 0.502 0.502 0.502 RG 35.000 52.000 m 585.280 52.000 l 585.280 50.500 l 35.000 50.500 l f 0.000 0.000 0.000 rg BT 44.000 31.996 Td /F1 10.0 Tf [(Page 4)] TJ ET q 44.000 17.740 532.280 11.880 re W n 0.000 0.000 0.000 rg BT 263.855 20.116 Td /F1 10.0 Tf [( Prof. Harald Walach)] TJ ET Q endstream endobj 37 0 obj << /Type /Annot /Subtype /Link /A 38 0 R /Border [0 0 0] /H /I /Rect [ 65.0000 398.5868 555.2800 771.8900 ] >> endobj 38 0 obj << /Type /Action /S /URI /URI (https://harald-walach.de/wp-content/uploads/2018/09/ml21-abb1.png) >> endobj 39 0 obj << /Type /XObject /Subtype /Image /Width 767 /Height 584 /Filter /FlateDecode /DecodeParms << /Predictor 15 /Colors 1 /Columns 767 /BitsPerComponent 8>> /ColorSpace /DeviceGray /BitsPerComponent 8 /Length 1486>> stream x 0Y&@U;2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)2)i) endstream endobj 40 0 obj << /Type /XObject /Subtype /Image /Width 767 /Height 584 /SMask 39 0 R /Filter /FlateDecode /DecodeParms << /Predictor 15 /Colors 3 /Columns 767 /BitsPerComponent 8>> /ColorSpace /DeviceRGB /BitsPerComponent 8 /Length 87404>> stream xwx7sz/,rl{B5@B%$GҀHq c۸*uW^#:g<ѝ;3Dk !B}@!RaBB-X Bo!B}V?![A!P߂B!~B!Է`BB-X Bo!B}V?![A!P߂B!~B!Է`BB-X Bo!B}@=wãѣ B 0JW\sBBBOzx>n/(RkP c\zxOc|qF )R vAŹVdx`J^fVRRa]rB 7qh35 m0-miwJ[YY ..:;w$֭K뫟y^K>^[m4k.f6'ܲ0.N8+>Rm۶im[ $-aaK/&ak:A NN) dKT{Wz7_2qxr 7!&}Ez/ @ ,+11)'L9krս8^}pP )Y[ ˀWVSEooVScGE%[pU7\;-y-vRj Ýs'JIk+UJ_ѻ~jpn !i8 pl?nV?'d{6mtd5p_qWf5V|7;B4.!R1d9&O4;kPo()q!@gk૮|/μiM mdw߿XŔ+5;z_qZkMPΓo?O%KeTB) az ugټ}u_{{wse_}odgp쌟 uA( ")H& `r6X @ Y۸3srsKJJV-˖|pۯgx^Vpkܼ7^{-qv)ch@sMݠm[khl: V6p7UI`?V~VS9Æzkoݼ~w?/9m _ZaWۅJ9CGN?:X^߿90YPN Ji7m{rc5Wr/+#1$2{ɶm;~o+y y0Q )RRJ 0B גii c.>oG[