Listen on the Almanac or Anchor, and find us and subscribe on Apple Podcasts, Google Podcasts or Spotify.
Don Harting, MA, MS, ELS, CHCP: Hello and welcome back to the Alliance Podcast, Continuing Conversations. My name is Don Harting. I'm a member of the Alliance PEERS section. And I also am a medical writer who specializes in needs assessments mostly in the oncology space. And I'm very interested in learning more about generative artificial intelligence, and also sharing with you what I learned along the way. With me today, I'm delighted to welcome our guest Alexandra Howson from the West Coast. Alex, could you just say a little bit more about yourself briefly to our listening audience?
Alexandra Howson, Ph.D., CHCP: Oh, sure. Well, thanks for the introduction, Don. As I said before, I'm really happy to be here. I'm a writer, educator, researcher and podcaster. I have a previous professional life and expertise as a trauma OR nurse. That's how I started my professional working life. And as an academic at universities in Scotland — University of Aberdeen, University of Edinburgh. But for the last 20 years, I've been self-employed and working as a freelance medical writer specializing in continuing medical education content and helping education providers develop content and evaluate that content. I'm also a qualitative researcher. In the last year, I've really shifted my business model a little bit and turned my attention to developing training and resources for medical writers and others who are looking to specialize in creating continuing medical education content. So I run courses and professional development, membership and coaching for people who are kind of looking to do that.
DH: Great! Well, thank you. Thank you for that intro. And yes, I've been watching you from afar and admire the work you're doing now, in your efforts to bring along the next generation of what I like to call CME writers. Just to be clear, when I use that term, I use it as a kind of a shorthand, or a term of art to mean writers who develop needs assessments and instructional content for accredited continuing education in the health professions. On our last episode, we heard from Andy Crim, who discussed generative AI, and its use in developing needs assessments and business proposals. During this episode, I'm eagerly looking forward to our conversation about using generative AI in the sphere of content development. And then in our third episode of this mini-series, we'll be taking a closer look at how to use generative AI in outcomes reporting and analysis. Before we go on, I want to thank you, Alex, for your service to the Alliance so far, by serving on the research committee, and also for your work behind the scenes raising awareness of what I like to call the content development piece, as opposed to program management and compliance and outcomes and learning management and funding and these other aspects, which are very, very important within our profession, but certainly not the whole story. And so before we go on, could you say just a little bit more about your feelings, and thoughts about the importance of content within the world of accredited CME CPD, and the people who work with the faculty to actually develop it?
AH: Yeah, that's such a great question. Thank you for asking that. So I've been involved in the Alliance since I thought it was 2010. But I recently, you know, was reviewing all my activities over the years for a couple of different things. And I think it's actually since 2008, 2009. And as you say, I've been involved in the research committee and other things as well. I was involved in the very early competency review in, I can't remember when that was, it was something like 2008 or 2009. It was a long time ago. And other things over the years. And one of the things that I always find striking was that people would talk mostly about things like program management, accreditation, outcomes, learning management systems, funding these sorts of things. And we talk about formats and deliverables, your wonderful term and the work that you've been doing in establishing, you know, competencies for medical writers who specialize in CME, but I rarely heard people actually talking about how we make the content: the stuff, the words, the images and the information that goes into slide decks, and learning materials and online activities, patient cases, these kinds of things that always kind of puzzled me and then I realized, Oh, actually, there's a perception that faculty do all of that. That they actually put all of this content together, when in fact, there are lots of people helping faculty do that and I have come to see. . . So I want to talk about that piece a little bit. But I have come to see the work that medical writers who specialize in CME, CPD, CEE, all the terms that we use to describe continuing education for health professionals. I’ve come to see medical writers as a little bit of hidden labor. Because a lot of people really don't know what medical writers do. So, you know, in relation to that content piece, content is everything. We know, because Graham McMahon talks about this all the time and other industry leaders. Content has to be evidence-based. It has to be trustworthy, it has to be free from commercial bias or even the perception of influence. And I feel that medical writers particularly are key pieces in making sure that that happens. We talk about making sure that there's a firewall between industry and education. And there are lots of processes and protocols in place to establish that firewall, I feel like writers are bricks in that firewall. And everything that we do is maintaining integrity and independence, according to the 2020 standards that ACME released. So how do we do that? I think there are two main ways. One is we work with faculty in terms of what they already have, and what they bring to the education activity. So the content that they are delivering, whether that is verbally in kickoff calls, or whether that is in the form of slide decks, and who among us have not had to work with 180 slides for you know, 60 minutes, AMA, PRA credit activity and reducing that by, well, you do the math, or bringing other types of assets that we that we need to work with. And often what happens in that process that is that our faculty are experts, and they are the experts, they have too much, or they have too many, you have too many words, too many slides too much information for a 30 or 60 minutes, activity or even less, you know, we've moved increasingly to microlearning. Some of those activities are, you know, 15 to 20 minutes long. And there is this tendency to stuff activities with information. So one of the things that I think that we do on the content side as writers, but also instructional designers, education strategists, graphic designers, is work with faculty to print and shape their content so that it adheres to adult learning principles. So that it is restructured into content that is presented in a way that supports learning through techniques like using white space and bullets, creating those pauses in the content so that people can stop and think and really reflect on what they're learning, using things like plain language to make sure that you know that that content is presentable. And also fact checking, you know, I think it's probably a fairly common experience for writers to get a slide deck that has literally been, you know, our brain dump over the years from faculty of all our expertise. But there, you know, there's there are no citations there. And so you have to kind of backtrack and figure out where pieces of information came from and update them if they're out of date. And then we're doing other things like creating outlines, finding resources, searching and synthesizing the literature to make sure that that content is current, as well as evidence-based arranging, presenting identifying key messages. As well, sometimes, you know, those messages can be a little bit lost. So I think the first way that we work with content is working with faculty to build out what they already have to prune and shape and restructure. And then the second way is building from scratch. And I think the same process applies where often we are, you know, having those conversations with faculty and learning how to use good questions to get the information that we need to build outlines, to create content, to direct us in our search in the literature and all those sorts of things. But we are often building things from scratch and I think I'm interested in your perspective here Don. One of the things that I've been seeing this year in particular, is that writers are increasingly being asked to provide some strategy in terms of content development. They know some providers are leaning very heavily on writers to bring that piece to the process. And I think this is something that maybe we need to start thinking about a little bit more carefully in terms of how we develop that skill set for ourselves.
DH: Sure, sure. I'd be happy to. And I did just get a query from a colleague yesterday asking me for help. Her client needs help with strategy. And I declined because I don't, I can't speak to strategy in the Alzheimer's disease space. You know, I only ever write in hematology oncology. So that's a great question. And as you've been talking, for our listeners, who can't see us on video, I've been just nodding my head. Yes, I get it, Alex. Yes, absolutely. We live and work in the same world. And I could, as you were talking, I jotted down several questions I wanted to either come back to or come to if we have time. But for the sake of staying focused on the topic of this episode, which to my understanding is about using Generative AI in content development. Let me go on to our next question. And just ask you, you personally, Alex, I think you're admired for the work that you do, and also your tech savviness. If that's a word, could you tell us a little bit more about when and how you got started using GenAI in your work? And what specifically does AI help you do faster or better? We'll get to the drawbacks later. But if you could just tell us about . . .
AH: Absolutely, yeah. And for our listeners, I’m blushing. Thank you for these kind, kind words. So I ,you know, like everybody else, I heard about GenAI or ChatGPT, specifically in late 2022. Probably through Andy Crim, I'm not entirely sure what the origins of that was for me. But I started playing with it pretty quickly, particularly in the needs assessment space. That seemed like an easy place to start. And I kind of asked it, I think probably some of the questions that Andy was asking, asking ChatGPT to create kind of background information. I think I started with a piece on MS. It wasn't for a project. This was just out of my own interest. My dad had MS. So you know, MS is a disease of interest for me and my family. But I asked it to create some background information. Like I imagined myself as somebody who didn't know anything about MS. working on this needs assessment. Where do you start? That was the kind of mindset that I took into it. And what I found pretty quickly was that, and I wrote an article in Medium about this and the pros and cons. The information provided was pretty generic but decent. If you were using it as a Kickstarter, if you didn't know anything about the topic, there was a bit of hallucination for sure. ChatGPT’s, not mine. And you really, you know, it told me right away that you really need to fact check, you know, the information that you're getting back from ChatGPT. This was way back in late 2022. And I was already using, you mentioned that I'm tech savvy, I don't actually see myself as tech savvy. I think what happened during the pandemic is like everybody else, I had to kind of figure out a lot of stuff. I also teach yoga, and so I had to pull all my teaching yoga teaching online, and that was very tech heavy. But I was already using things like OtterAI for meeting summaries and that kind of thing. And of course, it's using AI. And so I was somewhat familiar with, you know, AI a little bit. And pretty quickly Otter began to add things into its transcription process, like creating summaries of meetings, and I find that really useful. And so what does it help me do faster or better? You know, on the content side, I tend to still use AI, Generative AI as a kind of assistant and interestingly, slight digression here but still relevant to what we're talking about. A team out of, I think it was Harvard Business School, published something, at least in the last year, about the ways that we use AI. We use AI as either centaurs or cyborgs. We use AI either as kind of separate assistants, or as extensions of ourselves and our own kind of capacity. So we might circle back to that, but I think I use AI largely as a kind of assistant. It's good for things like brainstorming, ideation. If you are in meetings with clients, or, I mean, doesn't matter if you're in meetings, you can, if you have permission to go record, you can transcribe that record, you know, use AI to create very rapid transcriptions. I mean, as we're speaking, and in fact, Zoom has added AI Companion to its suite of AI tools, that will immediately generate a kind of summary and you can ask it to create, you can use other tools to ask for action steps. So I find meetings as one of the places where I find AI, Generative AI, especially helpful, because I get the transcript, I get a summary. I can use other tools to pull out the action steps that we talked about pretty quickly. Now, you still need to fact check. And I'm going to say that periodically throughout our conversation today. But it's a great start. In theory, you could also use AI. And I've played around with this a little bit in my own teaching. I teach at the University of Chicago in the Medical Writing Professional Certificate Program. You could use AI to, in our world, do things like create learning objectives, so you could create personas, and there are different prompting strategies. Persona is just one of them. But you could create different personas to ask AI to generate learning objectives in a very detailed way. So you could feed it, what Bloom's Taxonomy is, and what level of Bloom's Taxonomy you want to focus on. And Bloom's Taxonomy is not the only learning objectives model, but you know.
DH: But it’s very well known and widely used in our exact discipline.
AH: Exactly. So you could use that and train the model on what you need to create learning objectives, or even things like education activity agenda, or, you know, anticipated outcomes, and these sorts of things. Now, if you are doing that kind of thing, then of course, everything that we do in this world is proprietary. And so you need to be very transparent with the people that you're working with, if you're gonna go down that road. But in theory, you could use Generative AI to brainstorm some of those things in the needs assessment world. And I'm interested in your perspective here, because I know that you are the needs assessment king. You could ask, I was talking about this recently, actually, with Nuria Nigro, who has been using Generative AI a lot. And one of the things she suggests is that you can use Generative AI to do things like check your work to identify sources of bias, or if you're missing something, so you could ask it to identify urban and rural bias, like, you know, depending on the topic, or if there are economic or racial or ethnic disparities that you haven't included in your content that you should really be thinking about, or to pull in international perspectives. You know, a lot of the work that we do is very U.S. focused and again, depending on the programmer, the project, that's appropriate, but sometimes we are blindsided by our own bias. And don't necessarily search the literature for, you know, some of those kind of wider issues that we should be thinking about. So you can use Generative AI in that sort of way as well. Just to finish off that, that kind of piece about what I'm using it for? I have tried using ChatGPT Plus for literature searches in experimental modes. I have the paid version. ScholarAI integrates with ChatGPT plus. And there are a couple of other tools out there like Graphene AI which is designed specifically for healthcare, copywriting and consumer health writing. I think Shawn Soto was on the right medicine podcast a few weeks ago. Yeah. So there are tools that you can use to support you know, Searching the Literature, personally, I think if you have the skills that you need to search the literature, you you know which keywords to use, how to string keywords together, how to use Boolean search MeSH terms, those kinds of things, then you're much faster just using PubMed. Then using and also you still get that hallucination, you still get stuff turned back, oh, this is a great article, I really, this is going to be really useful. And you click on the link, and that is, yeah, there's nothing there that is really meaningful because the search is so wide. And you can get around that by being more specific with your contextualization and persona and so on.
Alliance Ad Break: Being an Alliance member has its perks. From discounts to industry leading events like the Alliance Annual Conference to members-only access to the Alliance Communities, the Alliance is where healthcare CPD professionals come to learn. Visit acehp.org to join today.
DH: You know, I asked you about the advantages of using GenAI in your work. You've touched a little bit on the disadvantages. I know you have pretty strong feelings about that as well. Could you say just a little bit more about the pitfalls as opposed to the promise of using GenAI for content development?
AH: Yeah, I'm sure the pitfalls have been, you know, a lot of people have been kinda focused on this. So I think there's a few things. And incidentally, there's an article in Frontiers of artificial intelligence, which very nicely sets out a lot of these pitfalls, you know, issues like security, which of course, the Bletchley Declaration has been focusing on, you know, political, economic and cultural education.
DH: For those of us who aren't totally up to date on this. You mentioned the Bletchley Declaration. What is that?
AH: Yeah. So various governments around the world met on November 1 to focus on the kind of security, safety and human rights issues that are potentially foregrounded by artificial intelligence, and Generative AI. I haven't had an opportunity to dig into the declaration itself. But I think what is significant is that so many countries are on board, in terms of trying to kind of figure out a way to, you know, create some sort of framework to support safety. Obviously, in our world, and adjacent to our world, other organizations have been doing that, as well. Earlier this year, The JAMA Network, the Journal, the American Medical Association and the Nature Network, both issued consensus statements around using Generative AI to create content, particularly in relation to author relationships. ISMPP, the International Society for Medical Publication Professionals, I think that's the acronym, they actually released a statement on November 1. Also, and I'm not a member of ISMAPP, so I'm not able to read that statement yet, but I know that they are also kind of focused on, you know, some of the similar things to JAMA and Nature.
DH: And having just gotten back from Baltimore, where the American Medical Writers Association was meeting for our annual meeting. Boy, the sessions on artificial intelligence, they were standing room only. I had to sit on the floor in the aisle, because I got there, like five minutes late, and there were no places to sit anymore. So then I was remarking to a colleague about how well attended that session was, and she said, ‘You should have been at the first session, the first session was even more crowded.’
AH: Yeah, and that obviously tells you, you know, something about the interest and anxiety, particularly in, you know, if you're in the writing field, but to circle back to your question about, you know, some of the issues, you know, we've talked about hallucination, we, there are other issues like copyrights. You know, there are there's a group of writers that have sued OpenAI, the platform that created ChatGPT because, you know, there's a question mark over all the information that's being pulled into what ChatGPT and also Claude and Casper and Jasper and Copy.ai and Perplexity and Inflection AI, what all these platforms deliver to you in response to the questions you're asking it. There are issues around bias and reproducing bias, you know, we know the world is biased. There are all sorts of disparities in how we think about, you know, everything, really, and we know that a lot of content that is already out there in the world is gendered. It reinforces racial and ethnic disparities. There's an argument that Generative AI is only reinforcing and reproducing that bias and disparities. So that's definitely something that we need to be really, I think, dialed into in the interests of diversity, equity and inclusion. There are other issues I mentioned, the JAMA and the Nature Network and their statements, there are issues around authorship. And I think where ISMPP, JAMA and Nature have somewhat coalesced is in relation to the issue of transparency. So if we are using Generative AI, then it's on us to be completely transparent about how we're using that information. And I think so those are some of the challenges, I do think we are kind of reaching a point in our world where continuing education for health professionals, our community needs to think about what its position is going to be on the use of Generative AI, because I think at the moment, we're still a bit Wild Westy. There are lots of us who are using it in different ways, and you know, with varying degrees of enthusiasm. But I think it's going to be pretty easy to get carried away by that. I do think we're going to need some mental benchmarks and parameters pretty soon, to make it clear. And we see this, you know, the National Institute of Health, I think, created some kind of policy or protocols, maybe about six months ago, high efficacy probably knows about this, to kind of create, you know, a foundation for using Generative AI in relation to grant processes. So I think that's going to spill into our world at some point or another. That's probably a longer answer than you needed.
DH: Oh, thank you. Thank you. I'm learning so much as we're speaking, I realized we're kind of diverting a little bit from the script that we prepared in advance, because I was going to ask you about some of the platforms that you would recommend for us to use. And you've so kindly already mentioned, quite a few that some I'd heard of, and some I'd never heard of I, I've heard you mentioned several I'm just kind of reviewing, but I've heard you mentioned of course ChatGPT but also Claude, and also Otter, and also Jasper and Inflection and Grafi. What have I missed?
AH: Perplexity. I mean, there's so many, Don. And in fact, there are some great resources. And I will include these for your show notes. But there are some great resources and newsletters that you can sign up for that are, you know, providing information about all the new stuff that's coming onto the market all the time, it seems just now. And I have to say being in the podcasting world, in the podcasting world has kind of switched pretty quickly to using AI tools. And so that's where I get a lot of my information. And that is mainly where I use AI to speed up the workflow.
DH: In your podcast operations. Yeah. Okay. And have you mentioned those platforms? I'm sorry, if I missed them, the platforms you've used in your podcast operations, because podcasting is becoming more and more of a thing in the continuing education in the health professions.
AH: Yeah, absolutely. And I'm seeing this and it's interesting. There's a couple of research studies, which I can also share for you for your show notes, which actually track how podcasting has grown in medical education generally, and also in CME and continuing education for health professionals. Yeah, there's a ton of platforms and tools out there, if you are thinking about developing your own podcast that can really help the process. On the recording side, you know, Margaret and I were talking before the recording of tools platforms like SquadCast and Riverside.fm. And they integrates with tools like D script, which is an AI tool that allows you to actually edit the podcast through text, rather than audio. So if you don't have audio engineering, on your staff, and who among us do? It is a really easy way to edit.
DH: Okay, my ears are picking up that, you know, editing in text as opposed to WAV forms on a computer screen.
AH: Exactly. And it's very simple to do and Descript integrates with all sorts of other platforms like YouTube and whatnot. And then on the, you talked about show notes, there are now several tools to help with show note production and also generating all the other materials and collateral that go along with the podcast that incidentally, I think you can also use to generate material to support other learning activities, materials like handouts, cheat sheets, templates, but platforms like SwellAi.com, Cast Magic, I love Cast Magic. I use both these platforms, by the way, because I'm still tinkering and experimenting, but I do love Cast Magic. There's some others as well. And what these do is they create transcripts, they create show notes. They create title suggestions like maybe 10 for each episode. They create social posts across most of the social platforms. They create timestamps, obviously. And now most of these platforms also have an additional piece of AI magic baked in where you can ask open-ended questions. You can ask conversational questions and ask the tool to create other types of content. So for instance, I want to write a blog article based on the conversation that we've had today.
DH: Kind of repurpose, or adapt a content form? To go from blogs, blog posts to podcast or from podcast to blog post? I think is where you're headed with that.
AH: Yeah. And so if you are repurposing your original audio content, then these sorts of platforms definitely help in that process. They speed it up, they help you to streamline your workflow, but, and I said, I would return to this at many points in our conversation. Yes, it's just a starting point. You, you need to fact check. But the other thing that you really need to bear in mind, and I think this is what assuages the anxiety for me, as a writer and a content creator, you absolutely need that human input. You need the stories, you need the context, you need the richness of language and the different ways in which we put words together to create that richness as humans. And you need to be able to, I believe very strongly that you can build empathy into your content. Generative AI doesn't do that. And so being aware of who your audience is, what their concerns are, what they need to hear from you in the content that you're creating is still very much going to be something that comes from us as humans.
DH: Looking ahead, what excites you the most about Generative AI and its applications in the CME CPD content development space?
AH: That's a good question. I think there's a couple of things. One is the potential for wider inclusion in creating and consuming learning content. I think that, you know, there are opportunities for people to kind of get started in the content creation space in continuing education, who might not currently see themselves in this space, or in this kind of role, I think as one thing. But inclusion also in terms of generating and developing, creating, designing different kinds of content for people to consume in different spaces. And through different devices and technologies. We're all pretty phone bound these days. And we know from some research that's been done that health professionals really consume a lot of their accredited CME on their phones. And so how can we use Generative AI to create and deliver content that is accessible for all different types of health professionals? You know, through those kinds of devices? I think there's opportunity there. Second, I think there's opportunity, or one of the things that is, you know, potentially exciting is the use of Generative AI to reinforce learning. We talked about repurposing content. And of course, at the moment, there are, you know, many activities that are accompanied by different assets or materials to try and help reinforce learning, downloadable slide decks, you know, those kinds of things. I think the opportunities for developing other types of assets to reinforce learning are also, you know, potentially captured by Generative AI. So those are the kind of two things I guess I think are interesting.
DH: I heard you say, the very first thing that you mentioned was inclusion. You're excited by the prospect of wider or broader inclusion that Generative AI presents? At first I thought you were talking about people, then I thought you were talking about ideas then I thought you were talking about different kinds of medicine. Are you talking about all of those? And more? Or what when you say inclusion? What do you actually mean by that?
AH: Yeah, so I think I probably meant it in terms of, in the first instance, people. People who are looking to build their skills as you know, writers, designers, strategists getting into the CME field. I think Generative AI can be supportive of that process. Well, I think the very first example that I used at the beginning of our conversation in terms of, okay, I have to write a needs assessment on multiple sclerosis, I don't know where to start. There are all sorts of other ways that you could start that process. You could use UptoDate, you could use other tools to, you know, start reading the literature, pulling pieces of information together. Or you could use Generative AI just to create a kind of foundation, that gives you a little bit of direction, so that now you have a template, okay, I need to know about this, this and this, I will use this information to help me really structure my search on PubMed, or an UptoDate to find the right. So I think when you're learning, as a writer, or as an editor, you know, at the very beginning of that journey, then I can see that there's potential for Generative AI to include people who might otherwise see themselves as not having the skills to to create content, whether that's a good thing or not, I'm not entirely sure, but I think there's potential there. I was just gonna say I really enjoy the way that you put those different things together. People ideas and whatever. The third thing was, is really interesting.
DH: Well, obviously, he took therapeutic areas.
AH: Yes, yes. Particularly in the rare disease space, I mean, obviously, rare diseases, having a moment at the moment, and has been over the last few years, but rare diseases, so tricky. To get good quality, you know, sort of evidence-based information. Obviously, there's journals like Orphanet, which is so central to the rare disease space. But if you are, I think for anyone who's you know, starting off in the rare disease space, where do you start? It's just hundreds of rare diseases. So, you know, I think that to your point, in that sort of context, yeah, there's the possibility to use Generative AI. But again, for assistance, for brainstorming, for ideation, we need to fact check. We need to be diligent, transparent, all those things.
DH: Thank you. Alex, before we end up, is there anything else that you'd like to say to our listeners on the topic of using Generative AI in content development?
AH: Well, I think we've covered a lot of ground. Don, you've asked a lot of really great questions. I think, you know, I want to emphasize, we learn by doing and so play with these tools, experiment. But really think about how you're using them. I think, you know, the question for me at the moment is less about which platforms to use, then how to use these platforms. It's a process issue. I mentioned, you know, earlier in our conversation, some research out of Harvard Business School, who that, you know, suggests we're either using AI as centaurs. We see AI as separate from ourselves, we approach it as a kind of assistant in their system model, or we see AI as cyborgs, as being integrated with ourselves as an extension of ourselves. And so I think we're going to have to start thinking about how we're using AI. And that is definitely going to have an impact on our search strategies. The prompting strategies that we use, persona is one of them, but there are lots of prompt strategies like flipped interation, asking Generative AI to ask us the questions, question refinements you know, dialing down to really kind of identify questions that are better than the original question you came up with. Asking Generative AI to be your reflection partner, to reflect on, to explain reasoning. You know, these are all about process and so, you know, experiment, play. Think about your process. And be transparent, if you're working on the content side especially and you're working with partners and collaborate. Collaborators, be transparent about how you're using Generative AI and be mindful about its potential for not only hallucination, but especially I think bias. We live in a world that is already biased. It's already very difficult for us as individuals to see and do anything about our own implicit bias. We need to not be baking that into the ways in which we're using Generative AI.
DH: Great! Well, thank you so much, Alex, for those concluding remarks. And thank you for joining us for today's discussion. Please be sure to tune into our next installment of this podcast mini-series on Generative AI, when we'll be speaking with Greg Salinas of CE Outcomes, who will discuss AI’s use in producing and analyzing outcome reports. Also, for even more education on Generative AI and healthcare CPD, register to join us for the Alliance 2024 Annual Conference taking place February 5 through 8 in New Orleans. Visit acehp.org/annual-conference to secure your spot and view the full program including sessions from Alex and myself. Thanks again for joining us.