This episode of Real Talk from AI and I Care is brought to you by Bardi Software. Welcome to Real Talk on AI and I Care with your host Dr. Scott Morris. And I'm co-host Dr.
Rehan Ahmed. In each episode we discuss current news, adding a little AI education, and debate innovative AI tools and topics that will change the industry. Hey everyone, Dr. Rehan Ahmed here, op-molished and co-editor of AI and I Care.
So welcome to today's quick knowledge bite. And I want to dive into something that's crucial yet often over. It's about crafting better prompts when using large language models like Jack J.P.T. or Clawed.
This is really called prompt engineering and is a subject of a ton of courses and if you google it online or even ask Jack G.P.T. out of the public, it'll give you good information. But I want to focus sort of as I care professionals who are already familiar with LLMS. And I'm sure you play with Jack J.P.T.
and Clawed or Gemini. We know they're powerful but getting precise and actual insights really depends heavily on how you frame and structure your prompts. So I just wanted to do quick framework to love what you're prompting gay. So the first thing to know is that clarity, likened speech is king.
Clarity is king. So when prompting be specific as possible, don't vaguely ask something like tell me about AI and retinal disease. Instead, sharpen it and say something like give me a concise summary of how AI detects early signs of diabetic retinopathy using fungicide images. The specificity here directs the LLMS to produce targeted relevant information.
Next, context matters a lot. Always define the scenario clearly. For example, don't just say explain glaucoma screening with AI. Instead, give it some context by saying briefly explain how AI screening improves glaucoma detection accuracy compared to standard visual field targets targeting ophthalmologists considering AI integration.
I think that context really provides a much more crafted answer. Third, guide the response format explicitly. If you want bullet points say so. If you want emojis say so.
If you just want one, two or three things say it. For example, let's three key advantages of AI assisted OCT imaging and ambient detection. And then you can say, well, briefly describe each and keep responses under two senses. The most important mistake to avoid is being too general.
So you have to specify the tone. Provide a reassuring yet authoritative explanation suitable for patient education. So something like that gives it more information on AI response. Lastly, I think the power examples are super helpful.
Let's say, you know, I do this a lot. Let's say you're asking for some patient education material. You could say, hey, write it friendly two minute patient facing explanation on how on common dry eye treatment questions with the reassuring tone and clear language suitable for a ninth grade reading something like that. You would also say do it in Spanish or Vietnamese.
So providing it some guidance really helps having structured well crafted responses. So to summarize, effective haunting includes being specific. Clearly defining the context and year you would say something like imagine your a seasoned optometrist or ophthalmologist. The third tip is explicitly guide format and tone and then lastly incorporate practical examples.
So by fine tuning your prompt, so consistently get higher quality outputs that are really I think tailored for your needs. I use it all the time from patient education to op reports to just brainstorming and I find that the better the prompt, the better the response. But the best way to do it is to just not play with it and get more experience. I hope this has been helpful knowledge by looking forward to talking to you soon.
Thank you. All right, welcome to real talk podcast. My name is Rayhan Ahmed and I'm excited this week to introduce up the board. She is the nerdy optometrist.
Her podcast has been viewed many, many times is gaining an a ton of popularity and actually I am so happy to have you here with us today. It's a pleasure to be on the show. This week Scott is out this week. He will be back next week and we'll definitely continue our conversations.
The news of the week. So the news to this today. It's it's May 6 and and our vote is going on and starting a couple days ago. I unfortunately am not at our vote.
I went to the Arbor website and I just typed in artificial intelligence of the and I want to see how many searches came in and is phenomenal. I remember my first Arbor that I attended was well, I was actually in medical school. So we're talking almost 20 years ago now. I honestly don't remember any AI related talks.
There may have been a few. But when I did an artificial intelligence in the filter today for search. They were I couldn't even count how many it was just kept on going and going in 20 plus posters, publications, et cetera. There were a lot of mature and that just shows you the breath and depth of AI and I care and I care research specifically for our road that's that's going on.
I've been very, very impressed with that. What would have you sort of seen recently like the in terms of just sort of what what is out there the amount of research. Is this something that is sort of catching your eye as well? Yeah, absolutely.
I just wanted to start by saying I was introduced to the world of technology five years ago and I think like many of us understand what I get in technology is. I myself also experienced a lot of hesitation whether you know this really works. Is this something about the future? Maybe you're looking for something that's going to happen say 15 20 years down the line.
And I think post COVID things have changed drastically pre COVID that probably was still just you know some pockets of innovation happening but post COVID I think one thing that has come out of COVID the good side is the. Incorporation of technology and innovation including AI in healthcare and I care is no different. I've definitely seen a lot more startups lot more AI surrounding conversations happening recently across the industry not just in I care but also in healthcare. Yeah, absolutely and just the amount of stuff it's not just computer vision models it's the most interesting abstract guys reading was about the use of large language models and a large ophthalmology practice in terms of condensing down multi-modal imaging into readable text for practitioners and I think patients as well.
So think about OCT fund a stratography all the other imaging that we're doing the ERG and condensing it down taking all that information and distilling it down into text so that was one example I saw there's there's many many examples definitely a lot on died of that pornography and others are the oculomics type of activities using yeah using retinal for the for for other healthcare related issues. This episode is brought to you by Barty the AI powered all in one each are built for I care with AI scribe voice over internet phones websites payment processing in our cm all bundled into one system party is redefining modern I care spend less time clicking and more time with patients at Barty dot com. One thing and and just applied for you at the end of the is the host of and you haven't checked out definitely check it out nerdy up to is it is it nerdy opt-omperist dot com is that the easiest place for the. Yeah audience to find you so a nodioctologist on all streaming platform with its YouTube or Spotify or Apple music but yeah the easiest is just go to the website nodio.
I'm going to watch it so you have both the office and yeah and definitely check it out I was I was I was I feel grateful gracious enough to have me I last week and it was wonderful to be on and one thing we talked about was how someone suppose I think there's someone who's interested in an AI is and maybe their optician opt-omperist ophthalmologist but they just don't have a tech background. You sort of talked about this at our at our top and I just wanted you to expand a little bit tell us a little about maybe your experience and some pearls and practical advice that you could share with our audience on on exactly that someone who's maybe not having the tech background but very interested and excited about getting into the AI space and some pearls and tools that maybe they can use what advice do you have for them. I think talking from my personal experience and also I had a privilege to talk of talk with lot of entrepreneurs in this space I would say I think the best advice I can give is just get started is what I have also learned and what I heard because honestly if I look at like what is going on in this space people everybody is trying to figure it out so it could be very overwhelming initially so there are lot of courses that are freely available. There are a lot of free available or even some courses that could be like you know at a very minimal cost takes some courses if you know exactly the niche where you want to build your expertise and then get into some role.
You can start with smaller start ups getting collaborating with some start ups are taking some roles where you can get some hands on experience along with some say educational course experience or knowledge that you have gained. But I think the only way to be comfortable with this technology is being uncomfortable and getting involved in it. You just have to get your hands dirty to kind of you know start learning things on the go. I started as a product manager.
I just like learning by doing is super is really the best way and sometimes something so new the only way to learn. Exactly so I actually started working in telehealth space five years ago and I had no idea what this works or what does this mean. One thing that I want to highlight to all the people in the I get industry is we all have the domain expertise. You might not understand the technology you might not understand you know what terminologies technically they are using but you understand I you understand patient you understand care you understand the gap.
And that is the strength that you need to play on because these solutions are being made for you your patients and for you know consumers who can use your expertise. So I strongly feel just go out there and build on your clinical knowledge on your practical knowledge. These things are engineers and experts who can help you learn the technical side of it but you can explain them the clinical side of it and that's how you know you can bring in come up with a great product or great solution. So I'll push you on the practical so give me one or two practical tools maybe it's a tool that you you have used in the past or website.
What is something that someone can do literally the instant they stop listening to us that they could go on and try out what would you recommend. What kind of tools are you looking for? You know, suppose there's someone exactly like I just read maybe they're just like you know they they know a little bit about AI but they want to learn a little bit more they want to as you said get there. They want to get the feet wet with their hands dirty really want to understand that they want to start doing maybe a little bit more maybe there's a neck.
So is there anything in particular that you recommend as a first go to thing you know I'm thinking about maybe one of these large language models that's for what I recommend. Do you have is that would you would you recommend the same. Yes, but I also want to go a step further and I am looking at these tools in two ways one tool is adding it into your practice and you know getting more comfortable with it and understanding how is it solving your problems. So first could be either if you have a practice of your clinic start asking for these companies who have built these AI tools to give you a few demos and get a little more involved to understand the solution how it works and also baby you can start as a consultant just to be little more and you know involved in the technology and the science behind it.
And on the other side of understanding your knowledge caning there are tons and tons of AI related courses free by Google sales force you know by different you demies so you can go to these educational portals and pick up the course that actually is interesting for you to understand different languages to understand like healthcare and AI and there are even books that you can read I think just understanding the. Understanding the evolution of these technologies can also help you understand how we have reached the place we have here today. Yeah that's exactly right there's so many good courses available that I've that I've taken as well I think you didn't you didn't have to answer you didn't hear you to me is fantastic also you know the large language they're amazing at what you just start talking to and asking questions of course it's not the end all be all and it's a good it's a good starting and a good starting point. It's going to get starting point for me but there's a lot of really good information out there thank you so much for that advice I think it's it's a very very helpful again to check out the nerdy app on check out nerdyapplaus.com or search for it on your favorite streaming app this has been a special edition of the AI in I care podcast again please check us out at AI and I care dot com.
Thank you so much for the nerdy app on just for joining us this evening and spending some time with us in our audience. Absolutely it's been a pleasure and yeah definitely I think you guys are also coming up with great AI content so I think just following you listening to you reading all the articles that you're writing is also great start the next step forget everything just start here. Appreciate it as a nice plug at the end I appreciate it thank you so much and thank you for listening good night. Thank you good night.
You've been listening to Real Talk an AI and I care your weekly podcast to keep you informed about AI technologies revolutionizing I care. Thank you thank you very much for supporting this episode of Real Talk for may I and I care.