hi everyone this is Fahad mza and I welcome you to the channel AI is not just about chatting with your models generating images from text it really warms up my heart to share this project Open Health with you which is a real world use case deriving real value from AI this person who built this project had already spent more than $100,000 plus 30 plus hospital visits on his or her autoimmune disease and then he decided to build this tool because he struggled with his mysterious symptoms and he was getting injured easily during workouts he was
having slow recovery random fatigue joint pain and lot of other uh weird things and he was running from Doctor to doctor hospital to hospital he went through standard treatment to experimental protocols at longevity clinics he changed diets he altered his exercise routine sleep schedules but nothing was helping and the most frustrating part wasn't just that the lack of answers it was how fragmented everything was and if you have or if you know about any prolonged ailments you know what I'm talking about each doctor you see they have only the piece of puzzle most most of
them have no clue what is going on to be honest for example uh in this person's case the orthopedist looked had the joint pain the endocrinologist checked hormones the rheologist ran their own test but no one was looking at the whole picture there was no single pan of glass there was no holistic view it wasn't until this person visited a rumatologist who looked at the combination of the symptoms and genetic test results that this person learned that likely uh he was suffering from an autoimmune condition and then what he did he took the matter in
his or her own hand he fed all the symptoms and medical data from before the rheumatologist visit into the GPT and the model suggested the same diagnosis this person eventually received so after sharing the experience the person dis discovered many other uh facing similar struggles with fragmented medical histories and unclear diagnosis that is why he created this open health and he turned this into an open source tool for anyone to use it is still in early stages but it's functional and it might help others in similar situation but here is a heads up look I'm
really glad that this worked in this person's case but still I believe for these specialized domains and medical where you know there is a matter of life and death and your health I think it's always good to get some holistic View and everything but after taking that instead of acting upon the advice yourself I would suggest go to the physician show him all the findings and I'm sure that the whole picture would help the physician more anyway this is uh my thought on this Open Health let's try to get it installed locally and then we
will see how it works before that let me give a huge shout out to Mast compute who are sponsoring the VM and GPU for this video If you're looking to rent a GPU on very affordable prices I will drop the link to their website in video's description plus I'm also going to give you a coupon code of 50% discount on a range of gpus this is my 2 system and this is my GPU card nvd rtxa 6,000 with 48 GP of vram now for installation make sure that you have a recent or latest version of
Docker installed if you don't know how to install Docker please search my channel I already have covered it in great detail let me first get clone the repo of open health and I will also drop the link to it in video's description they already have provided an environment file which primarily contains some of the uh local information around your database which I believe is being used as the vector store so it contains some of the postest which is being used as Vector store because behind the scene all it is doing is Rag and I will
explain it shortly what exactly it is doing so once you have renamed your environment file all you need to do is to run this command Docker compose EnV file and EnV up so it is going to download everything and then it will set up so let's wait for it to uh get up and running shouldn't take too long in my opinion and it took around a minute or so and now you can see that the application is running locally on our local host at Port 3000 so let me access it in the browser and there
you go open health is running on our local system on the left hand side you can manage your sources through which you can get your data so if you click on it this is where you can add your personal information if you click on add Source here you can upload new files and that sort of stuff let me first show you the interface and then you can click on X on the top right if you look at the right hand side this is where you can select the models you can go with open a model
anthropic Google or so these three are the API based you would need the API key and this is the AMA if you're having a local model you can select that so for example you select AMA and then you see that it has select selected the Local Host here and then it couldn't find the model because it is running in the docker and if you don't know what AMA is AMA is one of the easiest tool to run large language models locally and this is a system prompt which it is using that you a world CLA
doctor with a systematic and patient centered approach it is running in assistant mode and then there's lot of other stuff where you can give it the role like you want just a generic advice or Orthopedic clinical and it goes on and on and then the best doctor and I would suggest you just and then of course the system prompt changes and I would suggest you just go with the system prompt initially and I already have o Lama running let me quickly show you so this is my AMA where I'm running deep C Caron latest you
can use any model you like but to be honest if you are really doing it for very critical Health purposes is I would suggest go with a full model which is API based okay so our uh model is set now and you can ask it questions like how much how many chances do you have for heart attack and that sort of stuff if I click on many sources I just added some fictitious information here about my uh date of birth and some other medical information and then you can add multiple sources like blood reports and
all that stuff and then you can uh ask the question from the model as I just showed you I'm not going to show you all the personal information here so I'll just click on X here and then we just have to wait and for the moment I just selected open a as I said you can select with AMA it's pretty easy to use just select the AMA here or anthropic or Google Now one thing I have noticed is that it takes long time to answer I believe that should be improved and I think that it
is due to the reason and the way they are using postrest database which is quite heavy I think any other Vector store would be good for this purpose also the AMA support is still bit patchy um interface could be improved a lot especially on the right hand site for setting the model maybe we could also set it in some environment variable and that sort of stuff um and that uh kind of thing so I'm not going to run all the medical information here doesn't seem right to me but it's a simple chatbot I would say
with lot of data sources related to medical and I think um the person has tailored it to the medical um side of things which is quite good nothing groundbreaking I would say but still I think these are the real world use cases where people are using it for their own purposes another thing which I have seen uh is that for example if you go to manage sources and and then you click on ADD Source from here you can add your new symptoms today on the daily basis and then it is going to add it to
the context that is quite good so if you're monitoring it day to day this is where um it makes a new difference where it just upd updates the rack database also you can add more sources you can as I said blood report your other medical history and then you can add that data on the Fly uh and then start uh chatting with the model on that context now of course uh if it is you using AP based model there could be some privacy issues for that I would highly suggest if you want to use a
local model the AMA ones use a higher Quant not the Q3 Q4 go at least with Q8 so that the quality of responses of the model should be good enough so that's it I hope that you enjoyed it and if there is one point you want to take away from this video is that try to think of these sort of use cases when dealing with AI not just image generation and a simple chatbot to ask the weather anyway so before I let you go let me also introduce you to the sponsors of the video who
are ENT bot ENT bot lets you effortlessly deploy a personalized knowledge B across platforms like Discord slack and others it is ideal for open source Tech communities and startups that provide user support and I will drop the link to their website in video's description if if you like the content please consider subscribing to the channel if you are already subscribed please share it among your network as it helps thank you for watching
