A model to predict medical equipment failure and predictive maintenance calendar based of a whole lot of data recompile from our own and clients cmms.
Its fun, extremely time consuming, frustrating, but fun. It also provides to our clients with some interesting and useful information
Usually the joy comes after I force myself to build a tool for specific purpose focus on me, and it works. Then if the tool can be scaled for more users, and their feedback is good (good criticism also) that I would consider it a plus.
That would depend on what LLM are you using and the average tokens per answer. For example the average cost of the user query in my system is about 0.07 if you add 0.01 you are adding 15% per query (+-). Then, take into consideration how much value does RAG add to the answer Vs using other method.
If the listing is for all countries
this approach might not be the best, you need to consider that in some countries this data might not translate to the reality of it when it comes from a government source.
I don't have a clear solution either
I our company we use a CMMS to keep track of medical devices and their parts with the corresponding sub parts and so on (much like a tree). Shouldn't a system like this help with the problem?
And why not adopting a CMMS system? I'm missing something? Of course that you need to pay for it, monthly or one time payment, but they are made for this kind of job
It would be wise to not let a theory be affected by the people who preach it. A theory doesn't change because the person telling you about it doesn't have an exemplary lifestyle. Luckily for us, we are all humans and make mistakes
