Automation of Technical Documentation drafting, compiling, and updating

[aNiceCupOfQualiTea]

Hi all, here to ask about your insights and experiences with document generation and editing tools.

Now that the world is exploding with new language-model-based AI services with supposed application for the medical industry, I believe there should be safe and effective ways out there to use GPT-style bots to read information (for instance analytical results or regulatory updates or whatever) and suggest concrete text updates to existing technical documentation, including risk management files, post-market surveillance reports, and what not.

I wonder about the contradiction between protecting sensitive data versus allowing the bot to train on your files and connect it to web-trained parameters.
On the one hand, general text bots such as DocsBot are available, where I wonder about its suitability/safety for medical industry/technology... on the other hand we see specialized environments such as MedDevo, where it is unclear to me whether the benefits outweigh the costs for small organizations.

Any experience with automation of Medical Device File compiling/updates? Anyone here who might be developing their own tool using Microsoft Azure, for example.

Looking forward to learn and share thoughts!

 

[v9991]

reality check.,
a) one of the key aspect of automation and digitisation to be truly leveraged is to be standardized; i.e. at a data, information, knowledge ; etc.,
forget it;
b) even the organisation of files ( both and soft copies) is in itself a journey of evolution and maturity;
c) contents and context of the documents are also not standardised uniformly.
simply put, that is an work in progress still; ( read the automation or digitisation levels

with that context; the current status or challenges or practicalities
1. accuracy and validation to be balanced with use-case and benefits
2. data privacy / protection and security with governance and oversight

now, given the background.,
our experience has been to try to (solve)evolve an inhouse tool, which is used at business process levels, not directly integrated into the QMS/final submissions level; just say, tier-1 and tier-2 levels;

More efforts required to deal with points listed above than actually solving the problem; and often the the 1. and 2. seen as burden or overload; than appreciating the long term impact it has on focussing on core-science. and as per the solution it has mix of many things put together; require a leader who trusts the value and leverage of technology to make it happen.


here's short summary of GPT; and hope to hear back from the leaders