Does anyone know if artificial intelligence is being used for finding hep b cure?
Looks like AI is being used for hep b evaluation but how about a cure?
A machine learning can expedite the process of finding the right method for finding a cure I would assume. If there is none, how can we promote and push for the pharma and government to explore that? How can we create an incentive for them?
I don’t know about it that much, but I think it’s quite possible and that it makes sense for the labs to use AI in order to expedite HBV cure. And I remember that I have written something regarding to this somewhere on the internet. Like what kind of combination drugs could do the trick for HBV cure.
Actually, there are a lot of powerful tools has developed for drugdiscovery and protein designing, but apparently Hbv has less proiority in world, Idk what is the reason? I dont want to judge but I investigated a lot, even current company in development are not eager to use tools.
Maybe some startup or company is using this powerful tools, but me as I have been researched in this field(AI and finding cure for Hbv) for more than 3 month recently and there is not much thing about it.
I really wants to ask experts what is the problem, why so reluctent?
Yes, I have seen a couple of papers using AI (and I think there will be more in the future), but I think for the most part this technology is still developing and scientists are just getting to develop approaches to use this powerful method to research hepatitis B.
One other problem is just collecting the amount of good data to be able to answer what questions we have. We need actual observations of the world to feed into AI algorithms and a lot of them, and in some instances this has been done haphazardly and not systematically, which makes them much harder to use. A lot of the assumptions we have in Hepatitis B biology may not be fully justified so gaps in our knowledge remain and stop us from building frameworks to train the AI on.
These are just my impressions of why we’re not seeing some of these studies out yet.
Dear @ThomasTu, thanks for your reply and I am really happy that at last some experts join the conversation, because I don’t know why experts do not find these kinda conversations useful, we are a community right!
May I ask some question?
Do you mean AI technology? Or hbv researches?
I am sorry to ask too much question, could you please explain it more to me?
What do you mean by systematicaly? Because as far as I know there is not much use in field of hbv for combinaning data science and observation.
There are a lot of research but I think lack of modeling or data science approachs feels.
Please feel free to continue the conversation and tell me what I could not understand,
This community is the only place I am really comfortable to talk in.
I can not thanks enough to you @ThomasTu.
To some degree I mean both, but it’s really about the HBV researchers knowing exactly how AI technology can be used effectively and without bias to get accurate answers. I have seen ChatGPT produce outputs that sound very confident, but are actually factually incorrect. It would be a real shame for patients everywhere if we followed a line of research and devoted significant time and effort based on misinterpretation or an AI being confidently wrong.
I mean that different groups may be:
Measuring different values and readouts
Using different ways of measuring them (different platforms, assays, kits, samples)
Using different definitions
Designing trials with different timelines…
… for different patient cohorts…
… in different disease phases…
and not necessarily explicitly describing all of these differences in the reports.
All of this makes it hard to aggregate the observations of different studies all into the same model fairly and without bias. We also have bias of who gets to be in a trial and what we can truly measure with the tools that we have.
Some things where we do have a fairly clear-cut human interpretation (e.g. liver pathology) and high amounts of data (thousands of scanned liver sections), have been subject to machine learning and AI development, but still need validation in different cohorts.
Actually for a good and reliable result we need to maintain a good AI team besides researches to tune the data and collect good quality data, it is a misunderstanding that we can have good models after we found what we want, so we have to setup a good team to fine-tune the data in result of collaborations of both research and AI team, thats how a good outcome could be very helpful, but it is not a good way to not find a solution to integrate results, as you have mentioned in
There are still some milestone in aggregating data and classificate them, and if we do not start using it we are just delaying and loosing the chance of great possible progress.
So the first outcome of changing this prespective is to achieve a good platform and convention to make the model more reliable and more aggregatable. And I dont think it is so hard to acheive.
And also just to clarify, ChatGpt is a LLM(large language model) that is generative in NLP(natural language model) and in a breif explanation, it has some Hallucinations(seems right but it is not totally real). It is in some generated results which is clearly exist and a part of NLP and is correct but if you are using this result as a outcome for special problem(such as solution to a problem) you are making mistake, because the purpose is not this.
For some special fields such as research in Hbv you need to develop a good LLM specific to this fields and related fields.
for all, This LLM needs to be feed by papers, notes, datas of that science and it also needs to be fine tuned by regular testing and feedbacks from experts(Human), and I think it is very essential to be supervised model.
That is the way of developing a good tools to accelarate the way of research, investigations and finding solutions.
Then you can use the right model to accuratly and reliably test those thing.
I really like to talk and discuss about it with experts and I am happy at least someone listen to my words and can not be appriciate enough.
Thank you again for listening but I am really sure that there is a high chance in making a breakthrough by collabration and working
Please @ThomasTu let we continue our talk through such approach, we could have a good talk, and sharing some Ideas, worth to shot.
Thanks and sorry if it goes so long comment, I try not expand it😅