There is progress in the exciting are of advanced therapies (gene therapy/cell therapy products). Even with a good grasp of what might treat or even cure a disease, it's necessary to develop and be able to consistently make a drug product of the right quality to use in clinical trials. An investigational product whose important characteristics (positive and negative) are suitably understood. And, to be able to store, transport, prepare and administer it without those characteristics being negatively impacted. The design of the clinical trial needs to be robust enough so there is unambiguous evidence that it works. That includes showing how the disease is reducing over time because of the product. It also means having clinical sites that have experience, time and skills to conduct the study in accordance with stringent standards so there is no doubt that the results are the results.
I feel optimistic that we have some amazingly talented people in the UK who can do all these things given investment and encouragement.
Fascinating read! I’m Harrison, an ex fine dining industry line cook. My stack "The Secret Ingredient" adapts hit restaurant recipes (mostly NYC and L.A.) for easy home cooking. Dm me if interested in a recommendation swap — we’re growing fast!
Thanks for the comment. Peripheral vestibular and auditory symptoms have been found to be predictive of PD (and incidentally also of Alzheimer's) in prospective studies and are not secondary to PD. In any case, PD cannot explain deafness.
The crucial risk factor is neuroticism, a lifelong predisposition. The best test for N is static ataxia or body sway, a sensitive test for vestibular disorder. The vestibular organs are intimately concerned with autonomic nervous system (dys)function, as any seasick sufferer can confirm.
I am a big fan of Sir Demis Hassabis and admire his ambition of realising genuine solutions by solving fundamental scientific problems ...
like Rory, I am a PwP (Person with Parkinson's') ... and have had a DBS brain implant for many years.
I have been concerned about how to generate sufficient amounts of data from PwPs as without data it is difficult to apply machine learning and build an AI.
There are potentially a number of sources of PD data .... One of the promising ones is voice but there are ethical issues which need to be resolved quickly ... I presented a possible approach at the WPC in Kyoto for collecting such data via voice activated agents ... at the time I was dreaming of an Alexa to have conversations with about PD symptoms ... I believe Alexa may soon be upgraded to Claude, when this becomes a genuine opportunity ... I have been testing having such conversations with Pi, an AI positioned as empathic and safe ...
I wonder whether we could help Sir Demis with setting up a discussion about PD and the data required ... ? Let's connect the dots ...
A good starting point might be the Michael J Fox Foundation’s PPMI database which contains nearly 15 years of clinical, imaging, genetic etc data for several thousand PwPs and controls. Many research teams have been applying classic ML to this data for things like earlier diagnosis, subtyping and analysis of disease progression. MJFF’s Fox Insight database similarly contains symptom data for many thousands of pwps.
However, in terms of developing better therapies, I personally think PD is currently a tough nut to crack because we don’t actually understand what is happening at a biochemical level. If we had a particular pathogen or rogue protein to target then this AI stuff could probably help immensely. But until we identify an objective target I think we’re shooting in the dark somewhat. Just my two pence worth and I hope that I am wrong!
I am currently checking out why neuroticism (N) is a risk factor for PD, all mental illness, medically unexplained syndromes, and many physical diseases, but not, crucially, for definite neurological syndromes. I would like to share some thoughts with a sophisticated audience as I am now scared of being overtaken by AI!
Other risk factors for PD include imbalance; falls; dizziness; vertigo; brain fog; hearing loss; poor speech perception in noise; tinnitus. The only part of the nervous system that can generate all these symptoms in the inner ear. So I hypothesise a variable, hyperactive condition of the auditory and vestibular parts of the peripheral labyrinth.
I would therefore predict an excess of other peripherally-driven symptoms in PD: audiosensitivity; feeling of fullness/pressure in ear(s)/head; susceptibility to drugs, especially alcohol; diplacusis; nausea; transient global amnesia. I would particularly like to hear from anyone who thinks all this is a load of hot air, so that I can move onto some other problem.
To be honest I don’t recognise most of the items you mention as either risk factors or symptoms of PD. Whilst is true that anxiety and depression are common in PD, this tends to be as the condition progresses. In the prodromal and early stages, autonomic issues like sleep disturbance and constipation are often seen. Anosmia is also a good early indicator. These are consistent with the theory of Braak staging whereby the disease starts in the brain stem and moves up through the midbrain and eventually into the cortex. I haven’t seen much evidence that the inner ear is much affected in PD. Hope this helps.
I am curious to know what constitutes a 'solution'. Diagnosis seems to be where the success lies so far, but how will AI be used to identify and describe potential treatments?
I’ve lived through several AI hype cycles over the past 30 years (and indeed even did some research in that area myself in the late 90s) and the impact on society more broadly has always been overstated. However, Demis Hassabis is both amazingly talented and grounded and when he says something, it’s certainly worth paying attention. His Nobel prize interviews are also worth listening to. I really hope he’s right about curing diseases but you raise an excellent point, Rory, about our ability to keep pace from a clinical perspective…
One of the particular challenges with PD is the lack of an objective measure of ‘success’, something that underpins development of AI solutions today This lecture on scientific discovery using AI talks about that and much more… https://youtube.com/watch?v=hHooQmmzG4k
AI in medicine has huge potential, if we could embrace it even in the basic tasks which take up huge resources in NHS now, that would be a great start!
There is progress in the exciting are of advanced therapies (gene therapy/cell therapy products). Even with a good grasp of what might treat or even cure a disease, it's necessary to develop and be able to consistently make a drug product of the right quality to use in clinical trials. An investigational product whose important characteristics (positive and negative) are suitably understood. And, to be able to store, transport, prepare and administer it without those characteristics being negatively impacted. The design of the clinical trial needs to be robust enough so there is unambiguous evidence that it works. That includes showing how the disease is reducing over time because of the product. It also means having clinical sites that have experience, time and skills to conduct the study in accordance with stringent standards so there is no doubt that the results are the results.
I feel optimistic that we have some amazingly talented people in the UK who can do all these things given investment and encouragement.
You should talk to my friend Ron. Used to have Parkinson's. Lots of fasting and low carb diet.
https://www.bc.edu/bc-web/schools/morrissey/departments/biology/people/faculty-directory/thomas-seyfried.html
Fascinating read! I’m Harrison, an ex fine dining industry line cook. My stack "The Secret Ingredient" adapts hit restaurant recipes (mostly NYC and L.A.) for easy home cooking. Dm me if interested in a recommendation swap — we’re growing fast!
check us out:
https://thesecretingredient.substack.com
Thanks for the comment. Peripheral vestibular and auditory symptoms have been found to be predictive of PD (and incidentally also of Alzheimer's) in prospective studies and are not secondary to PD. In any case, PD cannot explain deafness.
The crucial risk factor is neuroticism, a lifelong predisposition. The best test for N is static ataxia or body sway, a sensitive test for vestibular disorder. The vestibular organs are intimately concerned with autonomic nervous system (dys)function, as any seasick sufferer can confirm.
I am a big fan of Sir Demis Hassabis and admire his ambition of realising genuine solutions by solving fundamental scientific problems ...
like Rory, I am a PwP (Person with Parkinson's') ... and have had a DBS brain implant for many years.
I have been concerned about how to generate sufficient amounts of data from PwPs as without data it is difficult to apply machine learning and build an AI.
There are potentially a number of sources of PD data .... One of the promising ones is voice but there are ethical issues which need to be resolved quickly ... I presented a possible approach at the WPC in Kyoto for collecting such data via voice activated agents ... at the time I was dreaming of an Alexa to have conversations with about PD symptoms ... I believe Alexa may soon be upgraded to Claude, when this becomes a genuine opportunity ... I have been testing having such conversations with Pi, an AI positioned as empathic and safe ...
I wonder whether we could help Sir Demis with setting up a discussion about PD and the data required ... ? Let's connect the dots ...
A good starting point might be the Michael J Fox Foundation’s PPMI database which contains nearly 15 years of clinical, imaging, genetic etc data for several thousand PwPs and controls. Many research teams have been applying classic ML to this data for things like earlier diagnosis, subtyping and analysis of disease progression. MJFF’s Fox Insight database similarly contains symptom data for many thousands of pwps.
However, in terms of developing better therapies, I personally think PD is currently a tough nut to crack because we don’t actually understand what is happening at a biochemical level. If we had a particular pathogen or rogue protein to target then this AI stuff could probably help immensely. But until we identify an objective target I think we’re shooting in the dark somewhat. Just my two pence worth and I hope that I am wrong!
I am currently checking out why neuroticism (N) is a risk factor for PD, all mental illness, medically unexplained syndromes, and many physical diseases, but not, crucially, for definite neurological syndromes. I would like to share some thoughts with a sophisticated audience as I am now scared of being overtaken by AI!
Other risk factors for PD include imbalance; falls; dizziness; vertigo; brain fog; hearing loss; poor speech perception in noise; tinnitus. The only part of the nervous system that can generate all these symptoms in the inner ear. So I hypothesise a variable, hyperactive condition of the auditory and vestibular parts of the peripheral labyrinth.
I would therefore predict an excess of other peripherally-driven symptoms in PD: audiosensitivity; feeling of fullness/pressure in ear(s)/head; susceptibility to drugs, especially alcohol; diplacusis; nausea; transient global amnesia. I would particularly like to hear from anyone who thinks all this is a load of hot air, so that I can move onto some other problem.
To be honest I don’t recognise most of the items you mention as either risk factors or symptoms of PD. Whilst is true that anxiety and depression are common in PD, this tends to be as the condition progresses. In the prodromal and early stages, autonomic issues like sleep disturbance and constipation are often seen. Anosmia is also a good early indicator. These are consistent with the theory of Braak staging whereby the disease starts in the brain stem and moves up through the midbrain and eventually into the cortex. I haven’t seen much evidence that the inner ear is much affected in PD. Hope this helps.
I am curious to know what constitutes a 'solution'. Diagnosis seems to be where the success lies so far, but how will AI be used to identify and describe potential treatments?
I’ve lived through several AI hype cycles over the past 30 years (and indeed even did some research in that area myself in the late 90s) and the impact on society more broadly has always been overstated. However, Demis Hassabis is both amazingly talented and grounded and when he says something, it’s certainly worth paying attention. His Nobel prize interviews are also worth listening to. I really hope he’s right about curing diseases but you raise an excellent point, Rory, about our ability to keep pace from a clinical perspective…
One of the particular challenges with PD is the lack of an objective measure of ‘success’, something that underpins development of AI solutions today This lecture on scientific discovery using AI talks about that and much more… https://youtube.com/watch?v=hHooQmmzG4k
If, or when, the problem of disease is “solved,” what will we die of? Perhaps we will be allotted a number of years for our life span.
I hope you get a better night sleep tonight Rory.
AI in medicine has huge potential, if we could embrace it even in the basic tasks which take up huge resources in NHS now, that would be a great start!
I'm surprised people love to throw money at a single type of optimism.