Good observation regarding learning of Eroom’s law. I learned of it after entering industry and agree that it should be discussed in academic circles as widely as industry to create a sense of “pulling together” from the outset
I continue to be surprised by the relative absence of genuine innovation on the "boring middle bit" of clinical research (the component of R&D that consumes perhaps 66% of total spend). It may well be the case the traditional CRO sector, like all incumbents, will resist radical changes to its business model and other, more nimble and agile players, with less legacy infrastructure will take up the slack.
Let's hope that the adage of necessity being the mother of invention will play out again.
Agree this is a serious issue and while there are some very attractive elements of using AI to enhance discovery. Testable drug hypotheses are in over supply compared to clinical infrastructure. There is real opportunity to improve trial execution (which lowers price and timeline per trial) and trial design (better design can lead to proof of concept faster or a shut down allowing scare resources to go where needed) with AI and the big lever would be improved regulatory processes. Jim O’Neill had some very good ideas on this but not sure the current admin is supportive of implementing these.
Spot-on analysis! The framing of clinical development as the "boring middle" that gets overlooked is so accurate. Everyone wants to talk about AI-powered drug discovery or cutting FDA red tape, but optimizing the actualtesting infrastructure? Way less sexy but probablyway higher ROI. The asymetry between funding basic research vs operationalizing trials is a massive blindspot.
Lack of awareness about Eroom's law is absolutely a problem in academia. I'm currently studying medical statistics at LSHTM and these issues haven't been mentioned. People who should be thinking about this aren't.
One idea I've been mulling over is a weak form of the libertarian proposal: narrow the clinical trial regulator's mandate to focus solely on trials where the treatment is unsafe or has unknown safety. Exclude trials where safety is known and that are only testing efficacy. In other words, the regulatory and ethics scrutiny should be proportional to the reasonable worst-case scenario.
Implementing this would have the benefit of massively reducing costs for a subset of clinical trials. But it would still keep protections where they are needed.
And the great thing is, this scenario - where safety is known but efficacy isn't - is very, very common.
- Any dose-finding study
- Any timing-of-treatment study
- Most repurposed drug studies
- Any trial in populations near the threshold for treatment in clinical guidelines
- Any trial comparing the effectiveness of safe, approved treatments
If the participant information sheets are clear, concise and honest, ethics review should be a breeze. Since the treatments are commercially available, the Good Manufacturing Practice regulatory regime becomes unnecessary. Confirmation that the treatments are common and have been in use for a while will make recruitment easier. Most importantly, safety monitoring in the trial does not need to be more extensive than when that treatment is used in normal practice. This change removes major regulatory hurdles and allows for simpler trials, making them cheaper without undermining patient safety or rigour.
Good observation regarding learning of Eroom’s law. I learned of it after entering industry and agree that it should be discussed in academic circles as widely as industry to create a sense of “pulling together” from the outset
yes it's very weird it's like the industry isn't self aware
I continue to be surprised by the relative absence of genuine innovation on the "boring middle bit" of clinical research (the component of R&D that consumes perhaps 66% of total spend). It may well be the case the traditional CRO sector, like all incumbents, will resist radical changes to its business model and other, more nimble and agile players, with less legacy infrastructure will take up the slack.
Let's hope that the adage of necessity being the mother of invention will play out again.
I have an article about CROs and their incentives which need to be addressed separately.
Agree this is a serious issue and while there are some very attractive elements of using AI to enhance discovery. Testable drug hypotheses are in over supply compared to clinical infrastructure. There is real opportunity to improve trial execution (which lowers price and timeline per trial) and trial design (better design can lead to proof of concept faster or a shut down allowing scare resources to go where needed) with AI and the big lever would be improved regulatory processes. Jim O’Neill had some very good ideas on this but not sure the current admin is supportive of implementing these.
Spot-on analysis! The framing of clinical development as the "boring middle" that gets overlooked is so accurate. Everyone wants to talk about AI-powered drug discovery or cutting FDA red tape, but optimizing the actualtesting infrastructure? Way less sexy but probablyway higher ROI. The asymetry between funding basic research vs operationalizing trials is a massive blindspot.
Lack of awareness about Eroom's law is absolutely a problem in academia. I'm currently studying medical statistics at LSHTM and these issues haven't been mentioned. People who should be thinking about this aren't.
One idea I've been mulling over is a weak form of the libertarian proposal: narrow the clinical trial regulator's mandate to focus solely on trials where the treatment is unsafe or has unknown safety. Exclude trials where safety is known and that are only testing efficacy. In other words, the regulatory and ethics scrutiny should be proportional to the reasonable worst-case scenario.
Implementing this would have the benefit of massively reducing costs for a subset of clinical trials. But it would still keep protections where they are needed.
And the great thing is, this scenario - where safety is known but efficacy isn't - is very, very common.
- Any dose-finding study
- Any timing-of-treatment study
- Most repurposed drug studies
- Any trial in populations near the threshold for treatment in clinical guidelines
- Any trial comparing the effectiveness of safe, approved treatments
If the participant information sheets are clear, concise and honest, ethics review should be a breeze. Since the treatments are commercially available, the Good Manufacturing Practice regulatory regime becomes unnecessary. Confirmation that the treatments are common and have been in use for a while will make recruitment easier. Most importantly, safety monitoring in the trial does not need to be more extensive than when that treatment is used in normal practice. This change removes major regulatory hurdles and allows for simpler trials, making them cheaper without undermining patient safety or rigour.
The EU Accelerating Clinical Trials Act will fix this. After all if it didn't why would they say it does? /s