Search results

I wonder with results like this if repeated tests on the same patients whilst monitoring activity and PEM would help. If this represents metabolic stress wouldn't the levels change as severity of symptoms change?

Its on my reading list - they had interesting results I think its significant the Microsoft have picked it up in one of their small models (one Phi-4-mini) although I've not tried that model yet. https://arxiv.org/abs/2312.00752

There is lots of work on LLM architectures although it tends to be small adaptations on the basic Decoder based arch. For example, Microsoft have released a model using Mamba that uses recurrent layers to change linearize the computation in the prefill stage. Google have done some interesting...

This is a potentially interesting paper (I've only read the abstract). They seem to be fine tuning LLMs with time series data from various organs and then have an agentic approach to simulating the whole body based on these organ models. Could be an interesting approach to modelling the body as...

I was more thinking of guidance to people like web designers, service providers etc to make things usable. I think patients sharing experiences can be very positive as well (although not via HCP). [Added] Also from a public policy perspective acknowledging the importance of internet provision...

I could see some interesting research in the area. Given PwME are often house bound and find talking with people in person often leads to PEM I can see interest in looking at how social media use can help create connections and how this can be done to avoid PEM. Also from a usability/experience...

They are what they are. Where they can help is things like summarizing the literature. So I could see an LLM being used to search the literature around potential candidate hypotheses to look for other diseases having similar mechanisms. However, I suspect its better to recruit the experts in...

Is there something that could be done statistically to see if the data was consistent given a higher degree if error? It may depend on whether the sampling methods are biased or errors at random?

It the electronic health records are there could they be used to reassess the diagnosis?

One of the things I'm wondering is what next. Given the genes of interest that have been found what are the next steps to understand potential mechanisms; what would help test the hypotheses; and which would be good research groups in the right areas? Is there some low hanging fruit that could...

One think that struck me on a quick read of the decode paper is how little I feel I understand about the methods and genetics in general. Are there good resources to understand more about the details?

I wondered if there was a way to use the UK Bio bank to see if common genetic variants being talked about had predictive power. I assume the Biobank data was recorded sometime ago and prior to the pandemic. So could people with these variants be contacted to see their current health and whether...

The point is how do we influence people to do the right thing. We can challenge people but we are less likely to get success than if we nudge their existing positions.

no I don't think that came due to political pressure to help patients. You could argue they made a case the the DWP but I think this was more they had a true believer in the DWP (can't remember his name) who just added funding. It is different suggesting to politicians that if they want to...

In the end I think change will come if enough MPs push for it. It is easier if a senior minister pushes for change. One issue that is clear with the delivery plan is how difficult cross department working is on issues that aren't getting ministerial attention. I'm thinking with Labour MPs we...

This was quoted as an example in the research hub proposal to demonstrate that the MRC can do such things. They never addressed that point - or gave a formal answer as to why they won't fund a research hub for ME. I have the impression that where there is a political will (lots of people...

There is intended to be a showcase event that the MRC/NIHR will do which should help. I wonder about follow up after that if we could do something to interest researchers.

So government funding won't be forthcoming without lots of work it doesn't work that way. The process requires academics in the right area to put in research proposals to explore areas identified by decode results and this these proposals go to review etc. The problem as I would see it is...

My concern is that Decode will show something interesting but research doesn't shift to investigate as there is no incentive to do so.

Do we need modern AI for CBT wouldn't a variant of Eliza work (https://en.wikipedia.org/wiki/ELIZA)?