Gregory Jones MD

Something here about ketamine, psychedelics, and immunology research.

All opinions expressed are my own

[Gregory Jones MD @GjPsych]

https://t.co/rJyg0x2m2h

Our letter to the editor in molecular psychiatry. We take a look at this historical trajectory of translational oncology research to make the case for continuing to reverse engineer the "nonspecific" effects of psychedelics.

[Gregory Jones MD @GjPsych]

@psybalazs @JAMAPsych Very interesting, but I wonder about the many failed open label ssri studies, which we will never know about, due to lack of mandatory registration...and whether these estimates (NEJM) could be used to adjust your data.
https://t.co/pTG7R1LxLr

[Gregory Jones MD @GjPsych]

@Sanbomics So what about throwing all the pipeline annotations + normalized prediction scores + qc metrics into a random forest with undersampling? If you already have the data setup it would be interesting to see A) what's the high feature importance B) how good the consensus does

[Gregory Jones MD @GjPsych]

@cremieuxrecueil 10 day post-flight endpoint is not super relevant for the type of neuronal damage they would incur. Wouldn't put much stake in these findings

[Gregory Jones MD @GjPsych]

@VPrasadMDMPH @Nature That's fair, but you also have to acknowledge that it was multiply-recurrent TNBC (until the final histology after she self-treated). Most people would probably be more concerned than just characterizing it as a "tiny local recurrence", no?

[Gregory Jones MD @GjPsych]

@MarkLRuffalo @TheBJPsych Is that the case? (Honest question). Anticonvulsants with -1.11 SMD and i2 = 24% fairly respectable? I don't put much stake in "very low certainty", would need n>400 and >10 studies not to get discounted by them. Also treatments specifically for BPD basically don't get developed

[Gregory Jones MD @GjPsych]

@suzannegildert @davecurtis314 @StuartHameroff Good discussion about that here
https://t.co/L64sosReIF
My understanding is the CNNa generally outperform everything else in modeling, at least for visual system
https://t.co/0wcH62ZUOF

[Gregory Jones MD @GjPsych]

@NTFabiano https://t.co/TmNZX9E4Y1

[Gregory Jones MD @GjPsych]

@TheBorisLab @EikoFried @psybalazs UMbias = SMDsaline -SMDactive-plc. From Fig 1, we can infer active plc contribution would be drastically different D1 v D3 without much ket change. Would guess rates change v D1? Corr w smd? Ought to be modeling both effects across multiple TPs if we truly care about unmasking.

[Gregory Jones MD @GjPsych]

@TheBorisLab @EikoFried @psybalazs See supplementary figure 1 here for an example...not about the AEs, but perhaps important to know what the active placebo response (and blindedness) looks like at multiple time points, if one truly cares about capturing unmasking?
https://t.co/IVVkzumW7M

[Gregory Jones MD @GjPsych]

@alieninsect @torjpay Same glutamate surge downstream of "primary" receptors, including scopolamine. All blocked in vitro by rapamycin. Subjective experiences for others vs DMT just as distinct as ket to any. All however have rapid therapeutic effects unlike any other drugs. Should be grouped IMO.

[Gregory Jones MD @GjPsych]

@cremieuxrecueil That you think you can have any certainty of harms (real or not) for such a widespread, poorly understood issue at this juncture is perplexing. I would love to see some robust human studies that give you confidence of lack of harm.

[Gregory Jones MD @GjPsych]

@mindslikeseeds @trpwolff Also I can, in theory, "nocebo" you to have lower efficacy of statins (certainly this has been shown with opioids and many other drugs). So it's not so clear that anything has biological activity independently of context. Hence double blind trials, even for imperceptible drugs.

[Gregory Jones MD @GjPsych]

@mindslikeseeds @trpwolff I guess my "generalizing" point would be statins only lower cholesterol if the experiencer utilizes context-dependent plasticity to faithfully ingest the pill everyday. My main point was that it's very far from clear that therapy is necessary for benefits from psychedelics.

[Gregory Jones MD @GjPsych]

@trpwolff @mindslikeseeds Honestly curious where the evidence for solely context interaction (and not drug) effects comes from? Unless you're making that case more generally in which case most things are dependent on context of conscious perception, drug or otherwise.

[Gregory Jones MD @GjPsych]

@doodlestein @langofmind I think Chomsky's main argument is that the transformer architecture required massive compute training on highly cleaned data (also with human feedback for the models that impress). Whereas my 2 year old with a "10 watt" brain and no clean data learns basic grammar intuitively.

[Gregory Jones MD @GjPsych]

@Neuro_Skeptic @Unity_MoT 99% just bad translation. Curved surface region for AUC, arbitrary forest for RF, K closest neighbors for KNN, etc. Some of the most amazing researchers, even in English speaking countries, don't have the best English proficiency. I wouldn't judge scholarship on that assumption.

[Gregory Jones MD @GjPsych]

@EikoFried Given ~40-60% comorbidity and likely bidirectionally causative over time, is it useful to segregate on population level? So much noise. Also in US reimbursement/insurance linked to billing codes, so many Dxs liberally specified to obtain specific complexity or get drugs covered

[Gregory Jones MD @GjPsych]

@Plinz Respectfully, it's convincing to the guy with a nobel prize in physics. From a medical research standpoint, his claims are more falsifiable in the next 15 years than any other theory. That MT-stabilizers (taxanes) increase anesthesia tolerance also very much worth investigating

[Gregory Jones MD @GjPsych]

@drmichaellevin @karina__kofman Very interesting. They have metabolic modeling tools for RNAseq. It would be interesting to see if someone has a similar rnaseq pipeline for Ion channels (even a focused network analysis). You could really test a lot of these hypotheses on public data.

[Gregory Jones MD @GjPsych]

@cremieuxrecueil Agree somewhat, but perhaps consider the immense advances from retrospective analysis . Biologics came from post hoc tissue analysis in autoimmune disease vs hc. "Tissue Agnostic" (basket) trials came from 50% pooled, post hoc data on cancer treatment response. Many more examples