4 ms·

How to Test More Drugs

23 points by apsec112 2y ago

pr337h4m 2y ago
[flagged]
- seattle_spring 2y ago
  That's how the supplements industry works now. They're almost entirely snake-oil and scams.
- dools 2y ago
  Not sure if RFK satire or genuine ancap point of view
  - pr337h4m 2y ago
    the latter
dang 2y ago
Recent and related: The Case for Clinical Trial Abundance - https://ifp.org/the-case-for-clinical-trial-abundance/ https://ifp.org/the-case-for-clinical-trial-abundance/ (via https://news.ycombinator.com/item?id=42462904 https://news.ycombinator.com/item?id=42462904, but no comments there)
vonnik 2y ago
Rux ignores one of the main bottlenecks of clinical trials: finding the clinical trial managers whose expertise is necessary to manage them and surface candidates for them. Pharma would run as many trials as there are managers. And managers themselves could run trials more quickly if it was easy to find good candidate matches for a trial. LLMs are beginning to help them solve the problem of candidate discovery, since the right patients need to be a high dimensional match, and their fitness can only be discovered by perusing their EHRs for hours.
- forgetfreeman 2y ago
  There is also the minor issue of private equity sucking the air out of the goddamn room for orphan/rare indications. Losing count of the number of investigators that have had their parent orgs purchased by PE firms which then inflate trial fees to the point that they fail BADLY in industry benchmarking software with no rationale provided for the increase and no recourse available. Waiting impatiently for the carnival of bullshit that's going to kick off the first time the OIG flags this shit as a false(?) positive for overt corruption.
- eddsolves 2y ago
  Can you ELI5 how LLMs help there?
  - thangngoc89 2y ago
    EHR is basically freeform text. So you use LLM to parse the text for certain conditions and criteria.
  - selecsosi 2y ago
    An example of a company using AI in the space to help extract data and match patients is this company https://deep6.ai/ https://deep6.ai/
refurb 2y ago
The message needs to be honed. Subtitle of “More trials, not laxer standards” then gives an example of RBM where in-person site visits and comprehensive data validation are replaced by digital “checks”. Ok, so “laxer standards” it is! The key is knows which standards can be more lax without jeopardizing patient safety. But I agree with the idea of the FDA incentivizing sponsors to leverage cost and time saving technologies. They often won’t do it themselves (or at least not quickly), but the FDA is in an ideal position to incentivize it.

23 points by apsec112 2y ago

pr337h4m 2y ago

[flagged]

seattle_spring 2y ago
That's how the supplements industry works now. They're almost entirely snake-oil and scams.
dools 2y ago
Not sure if RFK satire or genuine ancap point of view
- pr337h4m 2y ago
  the latter

dang 2y ago

Recent and related: The Case for Clinical Trial Abundance - https://ifp.org/the-case-for-clinical-trial-abundance/ https://ifp.org/the-case-for-clinical-trial-abundance/ (via https://news.ycombinator.com/item?id=42462904 https://news.ycombinator.com/item?id=42462904, but no comments there)

vonnik 2y ago

Rux ignores one of the main bottlenecks of clinical trials: finding the clinical trial managers whose expertise is necessary to manage them and surface candidates for them. Pharma would run as many trials as there are managers. And managers themselves could run trials more quickly if it was easy to find good candidate matches for a trial. LLMs are beginning to help them solve the problem of candidate discovery, since the right patients need to be a high dimensional match, and their fitness can only be discovered by perusing their EHRs for hours.

forgetfreeman 2y ago
There is also the minor issue of private equity sucking the air out of the goddamn room for orphan/rare indications. Losing count of the number of investigators that have had their parent orgs purchased by PE firms which then inflate trial fees to the point that they fail BADLY in industry benchmarking software with no rationale provided for the increase and no recourse available. Waiting impatiently for the carnival of bullshit that's going to kick off the first time the OIG flags this shit as a false(?) positive for overt corruption.
eddsolves 2y ago
Can you ELI5 how LLMs help there?
- thangngoc89 2y ago
  EHR is basically freeform text. So you use LLM to parse the text for certain conditions and criteria.
- selecsosi 2y ago
  An example of a company using AI in the space to help extract data and match patients is this company https://deep6.ai/ https://deep6.ai/

refurb 2y ago

The message needs to be honed. Subtitle of “More trials, not laxer standards” then gives an example of RBM where in-person site visits and comprehensive data validation are replaced by digital “checks”. Ok, so “laxer standards” it is! The key is knows which standards can be more lax without jeopardizing patient safety. But I agree with the idea of the FDA incentivizing sponsors to leverage cost and time saving technologies. They often won’t do it themselves (or at least not quickly), but the FDA is in an ideal position to incentivize it.