Whilst AI can provide many advantages to people with MS there is a key issue with digital connectivity. If you live in a more rural/remote area of the UK where there is insuffient connectivity to support a smart phone then this exacerbates the digital divide. There is an urgent need for the NHS and other key providers to insist that this is addressed as an urgent issue. Already the thrust towards digitalisation is important but many people are being left behind. They are missing appointments, letters and key communications because these organisations refuse to send paper based letters or phone patients directly.
The way to get around this is political. High-speed internet access needs to be defined as a utility and treated as such. The government then needs to invest in making sure all people in rural areas have access. This could include covering the subscription for access to StarLink. The technology is already there for internet access all we need is legislation to make it happen.
This is both frightening and exciting. If one could be utterly utterly confident that this information would not be exploited for anyone’s gain except one’s own or the Ms community, then - hell yeah!!
Prof G, thanks for sharing your thoughts on AI. I'd love to share your optimism, but I really can't. When I worked in the nuclear science field, we would talk about public fears about nuclear power in terms of risk perception. The two components of risk are severity and probability; while there's no disputing that misuse of nuclear power can have severe consequences, nuclear reactors are so carefully regulated that the probability of those consequences is quite low. But most people only focus on the severity aspect, of course. Relating this to AI, my concern is that we can have no confidence in our understanding of either the probability or the severity of AI misuse. And so I remain opposed to it, at least for now!
This would be a fantastic idea. For newly diagnosed people it could be a comfort and a helpful learning tool. I myself would be grateful for help with monitoring my symptoms; often I notice things are worse but I don’t know when things got worse or how much worse they really are. Like I know my balance and proprioception have gotten worse because I’m having more falls, but because my strength has gotten better (thanks to working hard in the gym) it’s difficult for me to unpack what part of things has actually progressed and how much. Similarly your points about monitoring bladder and bowel symptoms sound really helpful; I think sometimes it’s easy to miss the early signs that things are getting worse. With my cognition, memory and executive functioning worsening it would also be very helpful to have an AI to help me recognise some of the patterns in things. I have actually been considering getting a smart watch to help me monitor how I’m doing, since most days I get to the point of physical fatigue and so I figured it might help me to notice if that’s happening with less or more activity, and I know there must be other balance functions available because my iPhone often tells me I’m at an increased risk of falls (which I know anyway), based on some data or other that I don’t really understand.
Personally, if this was a tool that was available (and financially accessible to me), I would absolutely want to use it, it’s a fantastic idea!
Ah, that’s really useful to know! I’ve found exercise has made so much difference, and 8 months of consistently going to the gym 2-3x a week has got me from an EDSS of 7 to a 6/6.5, which makes so much difference to my daily life! I’ve heard from some people that the watches are basically an expensive way to feel like you’re being healthy whether or not you are, while others say they’re life changing for their fitness. It’s helpful to hear from someone who also has MS though, as I think our hopes and goals are quite different to your average gym bunny’
As always, Prof. G., penned perfectly! I loved how you emphasized that credible, accurate, and reliable information is vitally important with AI. Also, when MS progresses, IMO, it is impossible to get every question answered thoroughly in a doctor’s appt. Your envision is a great solution for most patients. Yes, there is email, iChart, telephone, etc… But often doctors don’t have time to answer questions for many situations thoroughly. Also, sometimes it gets uncomfortable. Honestly, most of my authentic information is from you, Prof. G. or from my FB group, Solving MS.
FWIW, GPT4 is now at an April 2023 cutoff for me (some claim later than that even). That's pretty decent unless you are going to the bleeding edge.
You can ask it along the lines of "what is your knowledge cutoff"
I agree however, its synthesis capabilities are fairly mixed - sometimes quite good, sometimes rather bad. Because of the hallucinations I generally only use it in domains where I have a fair shot at judging veracity.
I couldn’t grasp it all . Im 62 and I know my cognition is sliding. I’d need help getting to grips with this. How does it monitor cognition problems. I’d like to try FES how could it assess my needs here. I do my daily seated yoga and stretches I don’t need nagging by AI. However I believe for newly / recent diagnosed, those on DMTs it could be brilliant.
It monitors cognition by assessing how you use your phone, analysing your responses to questions or by asking you to download an use an application. MS-GPT is not a static app it will learn and evolve with time.
We will still need the human doctors, God forbid there are cuts. However AI technologies are prime for healthcare. I have a condition and put my symptoms into Isabel (an AI platform) and bingo it came up with my consultant diagnosed condition.
Totally agree re limiting/curating information sources - garbage in/garbage out.
Hi Stephen. I can understand that you may be concerned about cuts in medical services, but I can assure you that most doctors are not of the calibre of Prof G. One can judge this in many ways. Newly trained doctors are nowhere near those who trained only twenty years ago (in my opinion). They have neither the knowledge, skills, and definitely not the empathy now. For a start, they strike; something that would never have happened previously as they cared too much about their patients. This is despite less working hours and higher pay. No longer do the juniors work from Friday night until Monday morning, catching sleep whenever they could. That intensity is one of the reasons they became so well trained. It is not just doctors though. As midwives, we all worked together as a team, only calling a doctor (registrar or consultant) when absolutely necessary. Now, the midwives are not trained to the same level, so do not possess the qualities I describe above. This is why there is an increase in errors that did not occur.
Being in these professions is an honour, but very few now treat it as such. One only has to note the sloppy appearances, tattoos and heavy make-up adorned by medical and nursing/ midwifery staff.
Apologies for a long comment, but sometimes, I think a robot could be better trained, of course, as you mentioned, if the input is correct.
I find this all very sad, as I was extremely proud of my colleagues, and we had fabulous working relationships. Besides the hard work, there was an element of humour, which I think now is absent. Eve
All the reports I’ve seen, such as those from Nuffield Health, The Economist etc, show that pay for most healthcare workers has fallen in real terms since 2010, by around 8-14%, depending on the field and position. Due to massive shortages of staff work is also now harder for particularly junior doctors. As to your old-fashioned views on peoples’ appearances, given that around a quarter of the U.K. population now has tattoos, and that more than 10% have tattoos in visible places like the head and hands, it would make sense that the trend continues with healthcare staff, given that they are not actually fundamentally different than any other section of the population. It’s also a common comment among tattooed people that they enjoy and are comforted by seeing that their doctors, surgeons and other medical staff have tattoos, and again, since that’s a solid chunk of your patients, that should matter to you. I am also of the opinion that putting time and effort into one’s appearance and makeup is actually the polar opposite of dressing sloppily.
I would love a MS-GPT app and think this would be incredibly helpful in managing day to day with MS. There is something more basic than what you envision Prof G but very good nonetheless. The app called BelongMS in which two Neurologists, a Radiologist and Urologist respond to pwMS questions, now has a AI mentor called Sophie who responds to questions about MS in a very personable way. It's quite incredible to see and the response time is practically instantaneous, even to the most difficult of questions. 😊
My job involves me being an expert in finding out accurate information online. But I still find ChatGPT much better than Google for some searches. If have to research something I know nothing about ChatGPT is my first port of call because it gives you the basics a short readable chunk. It could be ideal for newly diagnosed people because if there's scientific word on the page you don't understand you just write that word into ChatGPT and it will instantly give you a clear explanation. If you have any concerns about whether or not the information is correct [it isn't always] then just as for the source of its information.
If there's a way to make a version that's more specific to MS than would be great. And it would be fantastic if could be personalised. Defiantly pursue this idea!
I think a key advantage for an MS chatbot is that it could also act as a journal/diary for PWMS. I would often forget to tell my neuro things when I was first diagnosed, because I was so frazzled with worry and there was seemingly something new all the time to take up the top spot on my worry list. I could never remember all the things I wanted to ask, or all the things I felt, or what happened when.
I absolutely think AI-powered medical assistants are coming. It's a matter of training them on a good enough data set that cannot be fooled by any kind of questioning into giving dangerous answers. "I dont know" should be built in whenever the model thinks it might be going wrong.
But even now, I could probably build a custom GPT on ChatGPT and have it do a very good job of being a first-pass helper to someone who has basic questions about MS.
I'd rather see a real live human neurologist than use MS-GPT for the same reasons that I'd rather go see live music than listen to AI-generated synthesized music.
But what if MS-GPT is better than your neurologist? That is the reality of what is happening. AI will be superior to humans when it comes to intelligence and recall.
AI could theoretically be superior to humans when it comes to anything. One day AI will probably be superior to humans at being life partners. Should I ditch my wife when Partner GPT comes out?
I feel like this put-AI-in-everything approach essentially reduces human beings to appliances. At some point this is about a human connection and not just conveying information and treating a disease.
Great idea that i heard in a while. I think is necessary and to my broad knowledge (including AI . You can check here :) https://liviuf.substack.com including some of the best AI assistant . Sure I am also with MS from 2011 diagnosed) I can help implementation team if you are needed - btw right now can access some AI medical tools but depending on symptoms-. Good luck with not so Sci Fi MS-Gpt ;)
My pleasure and very good information provided by you.MS-Gpt need someone behind that provide accurate info (for example existing chatGpt lack medical accurate info - theoretically are in Palm-2 from google but is not accessible and probably will not be ;) )
Whilst AI can provide many advantages to people with MS there is a key issue with digital connectivity. If you live in a more rural/remote area of the UK where there is insuffient connectivity to support a smart phone then this exacerbates the digital divide. There is an urgent need for the NHS and other key providers to insist that this is addressed as an urgent issue. Already the thrust towards digitalisation is important but many people are being left behind. They are missing appointments, letters and key communications because these organisations refuse to send paper based letters or phone patients directly.
The way to get around this is political. High-speed internet access needs to be defined as a utility and treated as such. The government then needs to invest in making sure all people in rural areas have access. This could include covering the subscription for access to StarLink. The technology is already there for internet access all we need is legislation to make it happen.
Good point. That exists in so many countries.
This is both frightening and exciting. If one could be utterly utterly confident that this information would not be exploited for anyone’s gain except one’s own or the Ms community, then - hell yeah!!
-
Prof G, thanks for sharing your thoughts on AI. I'd love to share your optimism, but I really can't. When I worked in the nuclear science field, we would talk about public fears about nuclear power in terms of risk perception. The two components of risk are severity and probability; while there's no disputing that misuse of nuclear power can have severe consequences, nuclear reactors are so carefully regulated that the probability of those consequences is quite low. But most people only focus on the severity aspect, of course. Relating this to AI, my concern is that we can have no confidence in our understanding of either the probability or the severity of AI misuse. And so I remain opposed to it, at least for now!
This would be a fantastic idea. For newly diagnosed people it could be a comfort and a helpful learning tool. I myself would be grateful for help with monitoring my symptoms; often I notice things are worse but I don’t know when things got worse or how much worse they really are. Like I know my balance and proprioception have gotten worse because I’m having more falls, but because my strength has gotten better (thanks to working hard in the gym) it’s difficult for me to unpack what part of things has actually progressed and how much. Similarly your points about monitoring bladder and bowel symptoms sound really helpful; I think sometimes it’s easy to miss the early signs that things are getting worse. With my cognition, memory and executive functioning worsening it would also be very helpful to have an AI to help me recognise some of the patterns in things. I have actually been considering getting a smart watch to help me monitor how I’m doing, since most days I get to the point of physical fatigue and so I figured it might help me to notice if that’s happening with less or more activity, and I know there must be other balance functions available because my iPhone often tells me I’m at an increased risk of falls (which I know anyway), based on some data or other that I don’t really understand.
Personally, if this was a tool that was available (and financially accessible to me), I would absolutely want to use it, it’s a fantastic idea!
Ah, that’s really useful to know! I’ve found exercise has made so much difference, and 8 months of consistently going to the gym 2-3x a week has got me from an EDSS of 7 to a 6/6.5, which makes so much difference to my daily life! I’ve heard from some people that the watches are basically an expensive way to feel like you’re being healthy whether or not you are, while others say they’re life changing for their fitness. It’s helpful to hear from someone who also has MS though, as I think our hopes and goals are quite different to your average gym bunny’
As always, Prof. G., penned perfectly! I loved how you emphasized that credible, accurate, and reliable information is vitally important with AI. Also, when MS progresses, IMO, it is impossible to get every question answered thoroughly in a doctor’s appt. Your envision is a great solution for most patients. Yes, there is email, iChart, telephone, etc… But often doctors don’t have time to answer questions for many situations thoroughly. Also, sometimes it gets uncomfortable. Honestly, most of my authentic information is from you, Prof. G. or from my FB group, Solving MS.
Yes I also use chat GPT around 30% of the time.
It's best (more foolproof for well founded scientific terms and their classic descriptions/understandings.
It is not current or good at extrapolating from what it's be fed. Or 'sensing' what are yet undocumented findings.
I is a good google amalgamator for the more well founded topics.
GPT lacks knowledge of newer things or so called alternative things beyond the mainstream..
(which it either is not fed by its human masters
FWIW, GPT4 is now at an April 2023 cutoff for me (some claim later than that even). That's pretty decent unless you are going to the bleeding edge.
You can ask it along the lines of "what is your knowledge cutoff"
I agree however, its synthesis capabilities are fairly mixed - sometimes quite good, sometimes rather bad. Because of the hallucinations I generally only use it in domains where I have a fair shot at judging veracity.
I couldn’t grasp it all . Im 62 and I know my cognition is sliding. I’d need help getting to grips with this. How does it monitor cognition problems. I’d like to try FES how could it assess my needs here. I do my daily seated yoga and stretches I don’t need nagging by AI. However I believe for newly / recent diagnosed, those on DMTs it could be brilliant.
It monitors cognition by assessing how you use your phone, analysing your responses to questions or by asking you to download an use an application. MS-GPT is not a static app it will learn and evolve with time.
When I can concentrate for long enough, out of fatigue bubble I’m currently in, I’ll try the App. 👍
We will still need the human doctors, God forbid there are cuts. However AI technologies are prime for healthcare. I have a condition and put my symptoms into Isabel (an AI platform) and bingo it came up with my consultant diagnosed condition.
Totally agree re limiting/curating information sources - garbage in/garbage out.
Hi Stephen. I can understand that you may be concerned about cuts in medical services, but I can assure you that most doctors are not of the calibre of Prof G. One can judge this in many ways. Newly trained doctors are nowhere near those who trained only twenty years ago (in my opinion). They have neither the knowledge, skills, and definitely not the empathy now. For a start, they strike; something that would never have happened previously as they cared too much about their patients. This is despite less working hours and higher pay. No longer do the juniors work from Friday night until Monday morning, catching sleep whenever they could. That intensity is one of the reasons they became so well trained. It is not just doctors though. As midwives, we all worked together as a team, only calling a doctor (registrar or consultant) when absolutely necessary. Now, the midwives are not trained to the same level, so do not possess the qualities I describe above. This is why there is an increase in errors that did not occur.
Being in these professions is an honour, but very few now treat it as such. One only has to note the sloppy appearances, tattoos and heavy make-up adorned by medical and nursing/ midwifery staff.
Apologies for a long comment, but sometimes, I think a robot could be better trained, of course, as you mentioned, if the input is correct.
I find this all very sad, as I was extremely proud of my colleagues, and we had fabulous working relationships. Besides the hard work, there was an element of humour, which I think now is absent. Eve
All the reports I’ve seen, such as those from Nuffield Health, The Economist etc, show that pay for most healthcare workers has fallen in real terms since 2010, by around 8-14%, depending on the field and position. Due to massive shortages of staff work is also now harder for particularly junior doctors. As to your old-fashioned views on peoples’ appearances, given that around a quarter of the U.K. population now has tattoos, and that more than 10% have tattoos in visible places like the head and hands, it would make sense that the trend continues with healthcare staff, given that they are not actually fundamentally different than any other section of the population. It’s also a common comment among tattooed people that they enjoy and are comforted by seeing that their doctors, surgeons and other medical staff have tattoos, and again, since that’s a solid chunk of your patients, that should matter to you. I am also of the opinion that putting time and effort into one’s appearance and makeup is actually the polar opposite of dressing sloppily.
I would love a MS-GPT app and think this would be incredibly helpful in managing day to day with MS. There is something more basic than what you envision Prof G but very good nonetheless. The app called BelongMS in which two Neurologists, a Radiologist and Urologist respond to pwMS questions, now has a AI mentor called Sophie who responds to questions about MS in a very personable way. It's quite incredible to see and the response time is practically instantaneous, even to the most difficult of questions. 😊
I think this is an excellent idea.
My job involves me being an expert in finding out accurate information online. But I still find ChatGPT much better than Google for some searches. If have to research something I know nothing about ChatGPT is my first port of call because it gives you the basics a short readable chunk. It could be ideal for newly diagnosed people because if there's scientific word on the page you don't understand you just write that word into ChatGPT and it will instantly give you a clear explanation. If you have any concerns about whether or not the information is correct [it isn't always] then just as for the source of its information.
If there's a way to make a version that's more specific to MS than would be great. And it would be fantastic if could be personalised. Defiantly pursue this idea!
I think a key advantage for an MS chatbot is that it could also act as a journal/diary for PWMS. I would often forget to tell my neuro things when I was first diagnosed, because I was so frazzled with worry and there was seemingly something new all the time to take up the top spot on my worry list. I could never remember all the things I wanted to ask, or all the things I felt, or what happened when.
I absolutely think AI-powered medical assistants are coming. It's a matter of training them on a good enough data set that cannot be fooled by any kind of questioning into giving dangerous answers. "I dont know" should be built in whenever the model thinks it might be going wrong.
But even now, I could probably build a custom GPT on ChatGPT and have it do a very good job of being a first-pass helper to someone who has basic questions about MS.
Yes, MS-GPT would work as a diary and it will summarise your chats to make a timeline for you to review prior to your scheduled review.
I think this is a fantastic idea as well. I am sure it will be easier to use with plenty of instruction. Thank you
I think this is an excellent idea Professor.
I currently keep diaries for separate issues because I try to recall what helped or impacted me at the time, and it’s exhausting.
Would be super to have one place to put these and possibly get a response if it needs urgent attention as you say!
Happy to help test anything prior to launch!
Great vision! I think taking advantage of the AI evolution mounted with human brains is definitely the way forward.
I'd rather see a real live human neurologist than use MS-GPT for the same reasons that I'd rather go see live music than listen to AI-generated synthesized music.
But what if MS-GPT is better than your neurologist? That is the reality of what is happening. AI will be superior to humans when it comes to intelligence and recall.
AI could theoretically be superior to humans when it comes to anything. One day AI will probably be superior to humans at being life partners. Should I ditch my wife when Partner GPT comes out?
I feel like this put-AI-in-everything approach essentially reduces human beings to appliances. At some point this is about a human connection and not just conveying information and treating a disease.
Great idea that i heard in a while. I think is necessary and to my broad knowledge (including AI . You can check here :) https://liviuf.substack.com including some of the best AI assistant . Sure I am also with MS from 2011 diagnosed) I can help implementation team if you are needed - btw right now can access some AI medical tools but depending on symptoms-. Good luck with not so Sci Fi MS-Gpt ;)
Thank you.
My pleasure and very good information provided by you.MS-Gpt need someone behind that provide accurate info (for example existing chatGpt lack medical accurate info - theoretically are in Palm-2 from google but is not accessible and probably will not be ;) )