Agentic Patient 7: How to Use AI as a Caregiver — Without Letting It Diagnose | Pratik Desai

[00:00:01] Dear listeners, welcome to Faces of Digital Health, a podcast about digital health and how healthcare systems around the world adopt technology, and another episode in the special Agentic Patient series where we discuss how patients use AI to navigate their care, get to better outcomes, and find answers that they might not get between visits.

[00:00:29] In today's episode, you're going to learn about a story that ended in an open-source tool that you can use as a caregiver to navigate health locally. So you need to download the tool. I added the link in the show notes. And then you don't have to share any of the medical data with cloud-based solutions.

[00:00:57] The speaker you are going to hear from is Pratik Desai, whose mother was diagnosed with stage 4 duodenal adenocarcinoma, a rare cancer which was terminal. But when the diagnosis was given, the family was handed a biopsy report and was effectively sent home.

[00:01:18] What followed next was not a story about AI curing the disease, but about Pratik using AI to close the gap between a diagnosis and the actions the health system was slow to take and would surely make his mother's life shorter.

[00:01:41] Pratik is an entrepreneur and a New Jersey-based AI practitioner, which also helped him create a tool that helps other people as well. In the discussion, Desai is also candid about the limits of AI, about what we should not do as caregivers or patients, and which boundaries to take into account based on the current capabilities of AI.

[00:02:10] Without further ado, let's dive in today's discussion. And if you want to learn more, make sure to go to the website facesofdigitalhealth.com

[00:02:42] Pratik, Pratik, Pratik, hi, and thank you so much for joining me on the podcast. Thank you so much for joining us.

[00:03:34] Thank you so much for joining us.

[00:04:06] Thank you so much for joining us. Thank you so much for joining us. Thank you so much for joining us. Thank you so much for joining us.

[00:04:35] Thank you so much for joining us. Thank you so much for joining us. Thank you so much for joining us. Thank you so much for joining us.

[00:05:05] Thank you so much for joining us. Thank you so much for joining us. Thank you so much for joining us. Thank you so much for joining us on this website. And so we're going to talk about that. We're going to talk about to you. And I think it's interesting because the way that the results were dripping to us, it was all coming in the portal, my chart.

[00:05:28] So I was receiving it maybe an hour or two hours before the doctor ever even stepped foot in the room. And my mom was there knowing, you know, every bloop, every, every notification that went off on my phone. She was like, what's, what's the update? What's the update? What does it mean? So, you know, in the room, I have to figure out how to communicate this and I don't even know what the words mean.

[00:05:46] So the beginning was just, what does it mean? How do I communicate this to my mom? And how do I navigate this situation? Because yesterday she was throwing, you know, huge parties for Diwali. And today now I have to tell her that she has a terminal illness.

[00:06:01] If we stop there for just a second. So I know that, you know, in the first stages, you basically used AI to even understand what you are up against. Can you take us through a little bit? What kind of prompts did you use? So what did you ask very specifically? And what did you learn based on those questions?

[00:06:25] Yeah. So the first result that actually came back to us, right? We were kind of not using AI at all. You know, she had a GI issue. She went to the GI. They did a endoscopy and they took a biopsy. At this point, we still weren't using AI really until the biopsy came back and there was a piece of paper in front of us. You know, the words, you know, duodenum, adenocarcinoma.

[00:06:49] And so at that point really was, well, okay, AI, here's the report. So we attached the PDF, you know, I'm getting tactical here, but attached the report said, Claude, tell me what this means, right? And that's it. That was the question. Tell me what this means. And it was very holistic and very comprehensive in what that was.

[00:07:08] And even, you know, went as far as saying, hey, who is this? Because I'm going to need to help you. It's very clear, right? Like this, I can't just tell you what it means and not recognize the severity of what I'm about to tell you. So it asked that question and I said, this is my mother. What do I do? Right. That was the next question. And it said, well, we don't have staging yet.

[00:07:30] That was actually the next recommendation from Claude is we don't have staging. It's very clear in this document that they have not staged it. So that's your next step. And you need to push for that. And that's what I did. The doctor came, kind of mentioned that this was the situation. We said, hey, we need to do the CT scan for the staging. And I didn't know what that meant, right?

[00:07:54] I was just saying, hey, AI said this. I'm going to push for it because that didn't come out of the doctor's mouth, right? That was not a clear next step as far as they were concerned. So you said, no, we want to do a same day CT scan. That was all because AI told me that that was the thing I should push for. And it specifically said same day, right? Stat. Use that word. Again, none of these things I knew before, right? I didn't know what the word stat meant.

[00:08:20] So, you know, I want a same day of a stat CT scan and we want the staging. And then we pushed for that and it took a little bit of convincing, but we got the CT scan. So we were able to get the staging that night. And then once I got that, you know, another piece of paper, right? Another PDF in the portal. So I put that into Claude and again, staying in the same thread. What does this mean? That's what I communicated. Okay. You know, this is the situation you're in.

[00:08:47] You're in a state like your mother has stage four. These are the statistics. This is what it tells us. This is what, you know, average life looks like. It was very clear that this is an extremely rare cancer. So statistics might as be, you know, not even listened to because there's just not enough right data.

[00:09:07] I believe, you know, and that was when it has informed me that in the documented world, there's like 3000 cases a year of this cancer. And all of those cases are pretty much found in the later stages because it's a very silent cancer. So that was its communication with me. And it said, now we need to get off of caregiving or rather advocating and we need to focus on caregiving.

[00:09:33] And we, the next few hours, you just need to figure out how to like communicate this to your mom. What does she need? Start thinking about like, what is, what is the rest of her life going to look? I mean, it's a very, this is all 24 hours, right? It's a very intense kind of sequence of events, but that's what it said. It's like pause for a second and take the time you need with your mom because this is heavy. Does that make sense?

[00:10:01] Yeah, absolutely. And so in these 24 hours, when did you get back in touch with the clinicians and with the clinical team that did the exams? What did they say? How much did what they were about to tell you differ from what Claude was saying? Was there any friction there? Did you also say to them that, you know, you learned this from AI? Is it accurate?

[00:10:31] How did that look like? Yeah, I was very clear. I was using AI the whole time, right? So I don't think there was necessary a discrepancy between what AI was saying and what the doctors were saying. The discrepancy was in action, if that makes sense. So it wasn't, you know, I think the medical system very much agreed with every next step that Claude said. It was Claude that told me that this is the next step that I had to then bring up to the doctors to force the next step.

[00:11:00] Right. And so the discrepancy was more whether the medical system was going to move fast enough to do the next step. They agreed with it wholeheartedly. There was never at this point in this 24 hours, there was really never a difference. So I would say AI's power at that point, that 24 hours was making me aware of what was going on and making me aware of what was I supposed to be advocating for.

[00:11:22] Because I truly believe had I not advocated for those moments in the 24 hours, it could have possibly been three days before a CT. And this was a Friday, by the way. So Saturday, Sunday, Monday, it could have been until Monday, where they did the CT scan. And I truly believe that, especially as it relates to the rest of our experience. So I don't think there was a discrepancy in care at this point.

[00:11:51] Now, go ahead. What do you think would be different in a world where AI wasn't present? So how would the outcome for your mother, who eventually passed away after a few months, would the outcome be different? I'm just wondering, like, what's your observation of the impact that AI had on you and the care that you managed to advocate for for your mother?

[00:12:21] And also the kind of the perception that you had of the medical system? Yeah, such a heavy question. So, you know, fast forwarding, she lived 76 days, I was by her bedside for pretty much all of it, 5am to 10pm. Right? I was there using AI, putting information in all the all the things I was seeing. I caught her essentially passing away with due to complications three times that was assisted by AI.

[00:12:47] I do believe she would have passed away had I not been using AI as the assistant. So there's a massive difference there, right? Like, probably on day 20, 20-ish, we would have, that would have been it. That would have been the story. So I think there's a big difference there. The second big difference, in my opinion, independent of those complications, is I don't even know if we would have gotten out of week two. Because we were discharged without an oncology appointment, right?

[00:13:17] It was a, hey, this is the situation. Say your goodbyes. And we said, hey, what's the next step, right? Like, what are we doing here? Are you really discharging us without an oncology appointment? The answer was yes. And we said, that doesn't feel right. It's like, okay, call this office, you'll get an appointment. So we got discharged that Monday, we ended up calling, and they said, hey, yeah, happy to give you an appointment in a month.

[00:13:45] Well, without any intervention, my mom doesn't have a month, right? So, like, that doesn't make sense. And we had that conversation live with the receptionist. I don't think it really changed anything for her, right? She was looking at a calendar and saying, I hear you, but here's my calendar. I don't know what to do. Okay, fair enough. And this is where AI comes in again, right? So what I did was I said, we live in New Jersey.

[00:14:15] We have lots of networks in New Jersey through doctor communities, nurse communities. Like, we have to figure this out. There's a way to figure this out. So I took my entire LinkedIn full of connections, my mother's connections, my dad's connections, my sister, you name it. Anyone that I could, like, just pull out all of their connections. And I fed these massive spreadsheets into AI and said, find me someone, someone that I'm connected to that, like, can get us an appointment.

[00:14:41] And after six hours of deep research, I actually spit out a name that was my ex-girlfriend in high school who happened to work hematology in Hackensack Meridian. And, you know, one way or another, this is all now manual, got her phone number and said, what, you know, this is the situation. What can you do? And she said, what are you doing at 3 p.m.? So we got an appointment that day, right? But had it not been for AI, I mean, how would I even have known that, right? The amount of research it would have taken me to figure out that I knew someone that could get me a 3 p.m. appointment.

[00:15:12] AI did that that same day. We got that appointment. And I think that intervention was the starting point of, again, you know, we weren't asking for the world. We knew we couldn't. We knew this was terminal. But we wanted to give her a chance to say goodbye. The medical system wasn't going to do that for us, right? We had to fight. We had to advocate. And using AI, I think we were able to do that.

[00:15:35] There's one more example, and I don't know necessarily if this changed our care, but I think it changed the way the medical system treated us. Early on in this kind of 76 journey is a moment where we had a CT scan, and the CT scan was riddled in typos. I mean, riddled in typos. And, you know, the doctor came to us the next day after me having put it in AI. And AI was like, I can't read this for you.

[00:16:03] I mean, I legitimately can't read this for you because it has two cancer mentions that she doesn't even have. So, like, that doesn't make sense unless you're telling me something new. There's also mention about, you know, growth in a tumor, but the numbers go in the opposite direction. So, like, is it growth or is it not? Like, I can't read this for you. So AI refused to read it and said, you need to tell the doctor this. They cannot make medical decisions off this. Whatever they do, they cannot make medical. They have to fix this.

[00:16:32] And the doctor came the next day, and I wanted to see if they were going to say anything about it. No. They said, this is what we were looking for. This is what the report said. The one line. They were ignoring the entire report and looking at one line. And we're going to move on to the next course of action because that one line was not what we were looking for.

[00:16:53] And you can imagine my frustration at this point because AI is telling me, I refuse to read this report to you because it's not comprehensive. It's not cohesive. Like, it doesn't make sense. And now I have a doctor here telling me I'm going to make a medical decision off of it. So when I pressed on this, you know, it wasn't received well, I will admit, that I was frustrated and I was very clear on, like, are you seriously making a medical decision off this report? And you stand by this report.

[00:17:24] It didn't go well, but I forced the system to kind of take a step back and look at it and go, yeah, this is we've lost a little bit of trust here. Right. And patient experience got involved and said, yeah, we acknowledge this is not okay. And then a nurse practitioner came later and said, we're going to fix the typo. And then I got a revised report. I never actually got the whole thing fixed.

[00:17:47] But eventually, you know, kind of the medical system over the next 24 hours basically said, we hear you. We understand there's issues. We'll try to fix some of it. I don't know why some of it was the answer. But this right, this thing that we're looking for ultimately is correct. Yeah.

[00:18:06] What did that, like, if you just think back, like, what did that mean for your whole, you know, trust in the system? How did the thinking that you had about the medical system change before you had all these issues?

[00:18:28] And then when you were going through all this and seeing the nuances that it deferred from AI to the healthcare provider? Yeah. Yeah. I think at the end of the day, I acknowledge that we're all humans, right? So we're all humans. We're all going to make mistakes. I don't expect – and I don't think I expected the medical system to be perfect.

[00:18:54] What I did expect was them to acknowledge that we're all humans. This is a corporation. I understand you have an entire floor of oncology patients, including the person that was sharing a room with my mom. And that's a very difficult situation. I understand doctors are all kind of overburdened. Yeah. Totally get that, right? The fact that I asked the question and wasn't immediately met with, oh, let me take a look at what you're saying.

[00:19:24] Okay. Yep. You know what? I see the issue. And it took 24 hours, three calls to patient experience, two calls to someone that I knew in the hospital who I said, hey, this is the experience my mom's getting right now. Is this okay? To finally get the medical system – not to say, okay, yeah, you're right. There's an issue here, which again, like, there was an issue. They eventually acknowledged it.

[00:19:50] But it took almost 12 to 14 hours for them to say, we'll have our resident look into it. Yeah. And then 24 hours later, they finally came back and said, oh, we'll fix some of it. Yeah, there's the typos. But don't worry about those. Don't worry about those things. It's complete loss to faith, right? I mean, trust was eroded. Absolutely, trust was eroded.

[00:20:15] And so now I'm in a situation where post that moment, I'm double-checking everything the doctor's doing, right? Not because I don't trust that they'll figure stuff out and not because I don't trust their education, but I don't trust their good faith, right? Because they've already made a mistake. That's okay. But they weren't willing to actually say, we're going to fix this. So now I no longer trust their good faith.

[00:20:42] And that's a terrible place to be in, especially in a terminal situation. Usually in those kinds of situations, people would either consider a second opinion or changing healthcare providers. What did you do? Did you just stay in the same situation?

[00:21:02] I must say that, you know, when you were discussing how the whole error wasn't kind of admitted and taken into account. In the past, I remember talking to two colleagues. One was a clinician. Both were clinicians, but one was a doctor and the other one was a pharmacist. And they realized how differently the training looks like.

[00:21:30] So doctors are trained to make fast decisions, to make decisions under pressure. Whereas, you know, clinical pharmacists, they have the time or kind of they're trained to take time and consider the options, the interactions and think a little bit more slowly before offering conclusions. Yeah. So what did you do? Did you just stay where you are and despite the lack of trust, try to move forward?

[00:22:00] Yeah, we had. So if you think about when she was diagnosed, November 20th, right before Thanksgiving in America. We dealt with the Thanksgiving holiday to get into a hospital or an appointment. We dealt with then Christmas coming up. So another holiday. We were dealing with holiday shifts and we knew that she didn't have a lot of time. So the amount of fighting we had to do to get into care, we weren't.

[00:22:28] We didn't have the time or luxury, unfortunately, to do it again. That's just the reality. So did we consider it? Yes, absolutely. We considered going somewhere else. I started exploring going somewhere else, but there was nowhere else for us. Everywhere else was going to say the same answer. We can take you in a month. She didn't have a month, right? And unfortunately, the situation at the time was she wasn't even transferable. So we strongly considered it.

[00:22:57] We would have done it had we had the choice. We didn't have the choice. So we had to continue. And, you know, that really it's an unfortunate situation. But the only action I could take from there on was double check, you know, to your point, second opinion, third opinion. I didn't have a doctor to call, but I had three chats on Claude that I was now spinning up and saying, Claude, you know, chat one, second opinion, chat two, third opinion, chat four or three or whatever.

[00:23:27] You're going to now double check all these guys. Right. So that's kind of how I proceeded because we didn't have a choice. Yeah. Yeah. Yeah. How was your mother in all this time? You know, it's not hard to understand that it's very hard for the relatives to accept that, you know, you need to let someone go. Do you think...

[00:23:55] So can you tell me a little bit more about, you know, how her care looked like in those 76 days? We often talk about just prolonging the life. And if the quality of life isn't preserved, especially in cancer care, then what good is it if you're just prolonging the suffering of someone that is suffering? So from that perspective, I don't know what your experience was. What did you gain through all that time?

[00:24:24] So basically, I'm being adversarial here, trying to understand how would the healthcare system see this case? Would they see it as meaningful patient quality time or just rising costs? Yeah. I definitely think they would see the rising costs. So I'll say that.

[00:24:50] On the 75th day my mother was alive, she was able to say goodbye to her family that flew in from California. I'll start there because on the 76th day she passed. So from our perspective, I'm not going to speak for doctors. I'm not going to speak for the medical system on how they would weigh this case. From our perspective, she was able to say goodbye to everyone that she wanted to across the world. They flew in. They said goodbye. She said what she needed to say to them.

[00:25:21] My mother is the bravest woman I've ever met. And the first 24 hours of day zero, day one, what have you, she looked at me and she said, what's the point? It's stage four. Why are you fighting? And we had that conversation overnight on is, is it, does it make sense to fight? And do you want to fight?

[00:25:47] And what, what, what would you want out of your life, whatever time there was remaining at that time, if you do fight? And she wanted a few things. Number one, she wanted to be able to have the conversations before she left. Two, and that's, that's not just with me. That's not just with my sister, but that's with all her family across the world. Number two, she wanted to know that we were going to be okay.

[00:26:15] She wasn't, she didn't, you know, she, we, we were a matriarchy. Our, our family life unit, like the beating heart was my mother. And so it was, it was crazy to us to, to imagine a world without her. And what, what is life going to look like after that? That she wanted to be there to coach us through it, to get ready because we needed to be ready and we weren't.

[00:26:43] And number three, she wanted a few more days with her, with her grandkids. That's my daughter and my nieces. She got those, right? And you, you feel like she got those because you used AI and just helped navigate. Yes. The, um, the care. Um, yeah, it's, uh, I guess we, today we are already talking about, you know, the importance of the patient experience, the quality of life.

[00:27:11] Um, and there's a lot of trauma sometimes if people don't get to say those goodbyes, especially on the side of those who, um, who stay, uh, behind. So, um, I can, I can imagine that, uh, that is still super meaningful, um, to you. Yeah. Yeah. I mean, and, and we define those goals, goals early, right?

[00:27:37] So I think, um, you know, the medical system is going to look at her as a case, uh, as a, as a number monetarily and also statistically and probably look at it and go, was this worth it? Did it make sense? Was the quality there? I think if you look at it from completely just a data perspective, it was absolutely shit quality. The reality is like she hit our goals. We got her what she needed and she was able to say goodbye. And like that, that is huge. Right.

[00:28:06] And, and I, I think she also for herself needed to be able to say, I fought, I tried, I tried everything I could. Um, and so my family will never ever wake up after this and say, what if, and like she gave, that was a gift she gave us. So I, it meant everything. The gift that you gave her basically. Yeah, sure. We, we gave her the gift of fighting, but she gave us the gift to be willing to fight. Right.

[00:28:35] Um, and, and the fact that she was accepting of the chemo and accepting of the trauma of being in the hospital, like 70% of the time through this time and in a bed, staring at four walls and, you know, watching. The same friends episodes over and over. And like, it wasn't great quality. Right. And I, I will admit that it wasn't great quality. She knew that, but that's not what quality was defined by for us. Like that wasn't the total definition.

[00:29:04] The total definition was having those conversations. She needed those conversations. We needed those conversations. So yeah, it was, it was the gift that we got. Can you imagine that the medical system, not you as a caregiver would take care of this? Is it, is it possible in your eyes? Oh man.

[00:29:29] Um, after seeing what I saw, right. I was, I, as I mentioned, she was in the hospital 70% of the time. I was there from basically 5am to 10pm. Um, there was no one else there, right. As it, as it relates to patient advocates, there was a social worker, there's nurses and they are phenomenal at their jobs. So I'm not taking anything away from them. Um, phenomenal.

[00:29:54] Every single patient in there, they had visitors one, two, maybe three hours a day. What I was able to do for my mom is unheard of. And I get that. Um, but I do think it moved the needle. I think it moved the needle drastically. And I don't even think it was just me believing that the nurses told me that some of the doctors told me that the social workers told me that and said, what you're doing for your mom is heard of, but you don't understand the difference it makes.

[00:30:23] Um, so yeah, I think, I think the hospital should think about that, right. As long as the goal is better outcomes. Then you should think about this. And, and I think it's, it's possible to, to, to think about how to get that patient advocate. Do I think it's going to be reality? No, not in the current system we're in, because I don't think, I don't think best outcome at all costs is the goal. Yeah. Yeah.

[00:30:53] Yeah, absolutely. Um, to a degree, it's also what you can afford and what kind of insurance you have, especially in the U S. Um, system. Yes. If we, if we just move back to the technical side of things a little bit in the beginning, we went through some of the prompts that you used. So, um, what does this mean? Uh, what should I do next? You created three personas to cross check what each one, uh, was thinking.

[00:31:22] Um, we also haven't mentioned yet that, uh, by background, you have a, uh, highly sophisticated knowledge of how technology works. So that's worth mentioning. You also vibe coded a tool, uh, for your mother. So can you, can you take us through that a little bit? How did the, the, how does the tool work? Because I, if I'm not mistaken, it's also available to other patients. Yeah.

[00:31:51] I've made it available to friends, family. I'm actually looking to open source it in the next week here. So very good timing on that. But, um, it's a, it's essentially, I'll start with the tool. It's essentially a tool like chat GPT, but the biggest, you know, in, in my journey, post my mother passing away and kind of trying to, trying to move the needle for other patients and other caregivers. I've realized that one of the biggest hurdles on, on using AI is I don't want to share my medical data with large companies.

[00:32:20] And two, I don't want my data training models. And that's fair. That's completely fair. And I don't even think we should fight that. Um, so the tool that I've been building and want to put out there is an open source version where you can load it onto your laptop, your desktop, whatever it is, put in your medical data, have a conversation with AI. But here's the thing, it's entirely local. So your data never gets shared to a corporation.

[00:32:46] Your AI, your data doesn't train the AI because it's just an AI on your desktop. And it's entirely siloed. You can turn off your wifi and it still works because you downloaded it and no one, no one's getting any of this information shared back. Um, so this is in the hands of a few friends, colleagues, they're, they're testing it, they're trying it and, and they're loving it. But I'll, I'll, I'll kind of back up a little bit and answer now kind of your first question,

[00:33:13] which is, um, ultimately I think the sentiment is I, I am a highly sophisticated, um, AI practitioner. And, and I think from all intents and purposes, that's probably a fair sentiment. So I, I, I analyzed that sentiment. I looked as I took a step back after my mom passed and I wanted to kind of do something with her legacy. I, I, I thought a lot about that and I looked at it and said, was this only possible because I have the knowledge I have?

[00:33:44] And the only way I knew how to answer that question was to try to give people the power that I had or what I felt I had over the medical system that didn't have that knowledge. Right. And I, I have, you know, we happen to be part of the, the sandwich generation. We have lots of kids that we need to take care of and lots of parents that we need to take care of. So I have no shortage of friends who are not in the tech world who need this to take care of their whatever. So I went around and helped friends and I said, this is how you do it.

[00:34:13] This is how you do it. This is how you upload. This is how you have, you know, good prompt. This is how you maybe ask a bad prompt, but ensure that there's a prompt at the bottom that says, let me know if I'm asking the wrong question. Let me know if I'm asking about whatever, whatever that process is that I've, I've defined. I've taught all these folks that, and they have completely made it their own and gone out

[00:34:37] and kept the medical system honest to their care and taken control, taken agency of whether it's their caregiver journey or their patient journey or whatever it is, they've taken agency. So I, I've done this now seven times and I truly, truly, I want to kind of break that, that sentiment, which is this only was possible because I am a sophisticated AI practitioner, right? Um, I think, I think that's true. I am probably, but that's not why this worked.

[00:35:06] And, and I have the proof to kind of talk about that. And I'm happy to talk to anyone about these examples. Um, and so, yeah, I'm working on a tool because there's a lot out there, chat, GPT, all these things that are free. And I think if you want to use them, you should. Um, but I'm working on a tool to kind of break the barrier of, I don't want to share my data. Hmm. Yeah. Yeah. It's a, it's a strong bond, especially if the tool is local.

[00:35:32] Cause I, I keep thinking, you know, when you have a small startup in my mind, somehow, uh, the fear related to privacy is higher. There's this false sense that if you use something that's humongous, that you're just going to get lost in the sea of all the data that's out there. And like, why would somebody at open AI look at your data if they've got like millions of

[00:36:01] users every day? Whereas when you've got the, when you have a small company, that idea that somebody is looking specifically at you is, um, is higher. So I don't know, um, how you look at that or what kind of, um, uh, feedback are you getting on the tool that you're giving out? Yeah. I mean, this is intended, not intended to be a startup, right? So there's none of the data streamed back to me, what the tool is.

[00:36:30] It's an open source, uh, platform that is on GitHub. You go to GitHub, you download it to your laptop. It's now entirely siloed to your laptop. It you're essentially creating a copy of it on your laptop that none of your usage, none of your data, nothing gets streamed back to anything because it's completely tethered to your laptop and local. Um, I do agree ultimately to the sentiment that other startups out there that are smaller,

[00:36:58] that are the cloud-based provider of this tool that are, have to get your data back to then present you with the results. They're probably looking at your data because it's their, their subset of actual users is small. And I, and I think that's fair. And I, I, I really, I don't even know if I want to push back against that because I think it's fair. Um, but that's not intending, that's not intending to explain what I'm doing.

[00:37:21] What I'm doing is I'm, I'm trying to break that fear by eliminating the actual streaming, eliminating who owns your data. You own your data, right? Just keep it on your laptop, keep it on your desktop and turn off your wifi. You can use it. Yeah. Yeah. Yeah. That's really important. Where nothing really goes out.

[00:37:49] So that goes, um, in line with the trends. What would be your advice in terms of how people should prompt what they should be wary about, what they should definitely not do? Let's start with a definitely not do. Right. Um, this is not a replacement for medical advice. It, it, it, I don't think that's how it should be considered. I don't think this is a diagnostic tool.

[00:38:15] Um, I don't believe in not doing doctor's appointments or medical appointments in lieu of using, let's use chat GPT as an example in lieu of using chat or in lieu of the doctor appointment using chat GPT is not a replacement. Plain, simple human in the loop. Everything should be corroborated and confirmed and decided by a medical practitioner. I believe that.

[00:38:42] So don't do that in terms of what you can do with this. I think so much of the medical system leaves gaps, right? In between appointments. And again, I'm defining this as you're taking the appointments because you should be taking the appointments. You should be talking to doctors, but in between the appointments, you're going to have questions. What, what just happened? What was said? I, I forgot. I mean, there's things that I, I knew I got explained and then I forgot again.

[00:39:11] Right. And so, Hey, what was the doctor saying again? Or remind me what this meant. Remind me why we're doing this. Cause I have fears. I have concerns. And my mom had fears and she had concerns. So use it for knowledge, understanding. Don't use it for changing the course of your, your journey. And then there's one last thing that you should be using it for is for advocacy, right? Asking the questions. You're right.

[00:39:38] You said earlier, doctors are trained to make fast decisions and move on because there's like 15 other people they need to see in that hour. Unfortunately, or fortunately, maybe at this point, because we have AI as a caregiver, a patient, your job is to advocate for yourself, for your patient. And, and now you have the ability to do so because you can put something into AI, say, this is the appointment I'm about to have. This is the medical situation.

[00:40:07] What questions should I be asking? What should be, what should I be probing for? And it doesn't matter now what the doctor forgets or doesn't forget or how burdened they are or overburdened. How many, like how many hours are they into the shift? Are they burned out? What? It doesn't matter because you have the entire playbook for that appointment in front of you. You can go through and ask the questions. Ultimately, the doctors say that's relevant. That's not relevant. If the AI says, no, push on this, push on it. Right?

[00:40:37] So you, you now have a, an awareness tool, a knowledge tool and an advocacy tool. I think you should use it. Absolutely. Is there like in your specific case, did you also have any negative experience? For example, what, was AI ever wrong?

[00:41:01] Anything that kind of changed your trust in AI itself that caused confusion? Not really. And, and I, I say not really because they're the biggest hesitation I have, I have in saying no, and, and it's not a hundred percent.

[00:41:30] No, nothing, no trust lost is there was a lot of ambiguity in the, in the complication of the case, which meant there was a lot of ambiguity in ultimately the question I kept asking AI, which was what's the prognosis. Right? And AI, I gave it one information, it will go here. I gave it another information, it will go here. Right? And so I'm, I'm, I'm now admitting to your audience that I broke the very thing that I said not to do, which was, it is not a diagnostic tool. Don't do it.

[00:42:00] And I'll admit, come on, of course I'll admit, right? I did it. And I tried and, and I was scared and I was concerned and this was my mother's life. So I tried everything and anything. And when I did that, it, it just was, it, it wasn't good. It wasn't good. Um, but I don't think that was AI's fault. Right. And I was giving it the very thing I'm telling you not to do, I was doing and the very thing that you shouldn't be doing with AI, I was doing because I was scared.

[00:42:29] And so I will admit that. So in those moments, um, it was all over the place, right? It was, it was trying to calm me down while also telling me the statistics aren't there while also telling me it's too rare to tell you the truth and too, but I was pressing it. I was pressing it. Um, cause you can imagine in the situation, like I just, I just wanted an answer so bad. Yeah. So that's where the trust wasn't there. And I, and I think it's because you shouldn't do that. Mm.

[00:42:59] Absolutely. I think, uh, the agreement around AI not being ripe for diagnostic use is, is still there and is commonly mentioned in the discussions, um, around AI. Um, well, uh, Pratik, thank you so much for sharing your thoughts, um, and your ideas. Uh, do let me know when the tool is going to be out. Um, I'd love to share it, uh, with the audience.

[00:43:27] Um, is there anything else that you feel like is important for people to know or any message that you would like to, to share? How do you see that the, you know, the, that medical care is moving forward? Yeah. Yeah. And I'll, I'll try to address this from three angles, right? I think if you're a patient out there, um, you have control. I want to remind everyone of that.

[00:43:51] I think you, you have control, whether it's AI or just Google search or whatever, ask your friends that are doctors and say, what should I be asking for? Advocate for yourself. And, um, the tools you use don't matter as long as you're advocating for yourself. I think as a patient, you know, or sorry, a caregiver, uh, it's hard. It's a hard journey. No one, no one signs you up to be a caregiver. No one says, Hey, you're now a caregiver or a cancer patient. You magically know everything about cancer overnight. Like that doesn't happen.

[00:44:20] It's also a second job. Um, this is a way, a way to alleviate some of the stress and give you a tool to kind of navigate the situation because no one signs up for it, but it is, you know, it is an admiral thing to do. And I think you can be successful at it if you, if you really put all the tools to work as a medical professional, I would say, you know, my, my ask to the medical community is I understand nothing's perfect. Um, AI is not perfect. Neither is the medical system. Not, neither is the process we've built around our patients.

[00:44:50] So my ask is don't fight necessarily, uh, AI usage, fight it being used as a diagnostic side care, um, but encourage it being used as an awareness tool, as an, as a knowledge tool and as an advocacy tool. And of course, encourage caregivers to get involved, educate themselves and figure out how they can help their patients. That's my ask of everyone.

[00:45:15] Um, if we, if we all kind of lean in on the areas where it clearly makes sense, I think there could be so much power here. Absolutely. And this is exactly why we're here having these discussions and sharing it with the broader audience. So Pratik, thank you again. Um, and good luck with the tool and the development that you're working on. Thank you so much. You've been listening to Faces of Digital Health, a proud member of the Health Podcast Network.

[00:45:43] If you enjoyed the show, do leave a rating or a review wherever you get your podcast, subscribe to the show or follow us on LinkedIn. Additionally, check out our newsletter. You can find it at FODH.substack.com. That's FODH.substack.com. Stay tuned.