 There we go. Okay, I'll turn it over to you. Thank you. So hi everyone, I'm Ciaran Bromley. I'm from Keel University in the United Kingdom. And I'm going to be talking about a simple shiny application called SS Progress, which is looking at sample size, calculation and evaluation of progression criteria in pilot and feasibility studies. And the QR code in the bottom left side of the screen is a QR code to take you to the application. So to start with just a little bit of background, in clinical research, there's often uncertainty about whether a study design is feasible or not. And that is why pilot and feasibility studies are designed to provide early indicators of potential success of a main trial. In these studies, we're not interested in clinical outcomes, but instead focus on understanding whether other aspects relevant to the delivery of the trial are suitable or not. Progression criteria are used to decide whether to continue the study, whether to implement remedial actions or whether to stop. There is, however, no consensus on estimating the sample size suitable for these studies, with various recommendations in the literature, ranging from 10 to 70 or even greater, with a variety of different outcomes, which are suggested focusing on precision of clinical parameters, feasibility parameters or even event rates. Lewis et al, a few years ago, proposed a hypothesis testing framework to determine thresholds for progression criteria based upon a traffic light system. So this is a simple system where we have an unacceptable region, which is our red region. We have an amber region where we specify we need amendments for our objective to make it acceptable. And then finally, we have our green already acceptable region. And using a one-sided hypothesis test, we are focused on row, which is a predetermined cutoff for the red zone within this traffic light system. And this red cutoff is the maximum level that we would still reject an outcome at. So in this hypothesis test, we use the null hypothesis as the red limit not being greater than row, and alternative hypothesis being the red value is greater than row. However, the progression criteria are considered at different levels, different hierarchical levels. So the feasibility outcomes can be seen at what we classify as the population level, the participant level or the treatment level. So some examples of these at the population level, we may be interested in looking at the rate of recruitment into a study based on our population of interest. At the participant level, of those that have been recruited, we may be interested in knowing the follow-up rate to our data collection procedures, or if the blinding of participants to the randomized are successful or not. And then finally, the treatment level, we may be interested in things like our treatment fidelity or monitoring the percent of adverse events which have been observed. So the aims of the project was to produce a simple and freely available application implementing this methodology that would allow researchers to easily determine sample sizes for their studies and to evaluate their outcomes. Throughout this, we used an example feasibility trial, which was looking at the oral protein, energy supplements and its flavor drinks to improve the nutritional status in children with 50 fibrosis. And within this example study, we had three progression criteria. The first at the population level was looking at recruitment uptake where we wanted at least 35% of children screening to be ineligible with 20% being unacceptable, 20% all over, sorry. At the participant level, we are focused on our follow-up where we want to see 85% all more children being retained within the study where 65% or less is unacceptable. And then at the treatment level, we have our treatment fidelity where we want to see 75% or more children being given the correct treatment plan. Anything less than 50% would be deemed unacceptable. The sample size is a calculated fruits of the outcomes, but the total sample size needs to be extrapolated through the levels of the study which I will get onto in a second. So here is a screen grab of our application. So there are multiple tabs across the top but this is the first interactive tab in our application and this is a sample size calculator. The application, this is the first version of the application which has been created by the research team and we had some additional testing from independent researchers from other UK universities. And when the user comes to this tab, it would normally be blank but I've put some information in here for illustration purposes but the user can interact with going across the top row. We have the test that they want to use for the hypothesis test whether this is the normal approximation or the binomial exact test. They're alpha and beta values which are used within the sample size calculation. The allocation ratio which shows the proportion split to the intervention or the control arm. And then finally the expected recruitment which is the proportion of participants who are recruited to the study from the population of interest. We allowed up to six different objectives for researchers to include where they can specify the level as either the population, the participant level or the intervention or control level. And the user inputs their red upper limit, their RUL and their green lower limit, their green and their GLL. These are the examples from before and the application will calculate their SS in the first column on the right hand side which is their sample size for each of the objectives. To the right of that we have our SS tot which is our total sample size needed throughout the study. As I said before the sample size needs to be extrapolated through the study and the application will determine which is the optimal sample size required across the whole study. As we can see at the bottom level we need 35 participants to have 90% power for our hypothesis test. However, if we were driven by the hypothesis test at level two we would have 44 participants at the participant level but given we have an allocation ratio of one to one we would only have 22 participants at this level and we would not have sufficient power in order to satisfy our objective. If there are multiple levels which are included the application will figure out which is the maximum sample size required and this is shown in the recommendations box below so it's easily noticeable to the user. We've also included a simple boot marking tab so researchers can return to the state that they've been working on. The second interactive tab within the application is our evaluation tab. This tab allows users to check the outcome of their study either through the monitoring or the final stages and the user would import their objectives and values from before as well as importing their numerator and denominator. The application would simply then calculate their current proportion which is then all shown on the right hand side in a reactive plot. So the number of objectives shown will be displayed within the plot on the right hand side and as you can see we have our red, amber and green regions, the name of the objective and the current proportion which is also shown by the black arrow. And what we're interested in knowing is where this black arrow sits in relation to these cut-offs to know whether our outcome is acceptable, requires amendment or is unacceptable. The user can also include custom titles, change the plot height and width to make it scale better within the application and then also add grid lines to easily read off where the proportion sits on the zero to one scale. From a more statistical perspective the user can also enter the alpha value which they used within their sample size calculated in order to be consistent as well as displaying a critical value through ejecting the null hypothesis based upon the hypothesis test through each objective and this is marked by the red cross. They can also display a lower confidence interval through each objective and the distribution for this constantable can be changed from the normal approximation and the binomial exact to be consistent again with their sample size calculations. The red cross is the cut-off for statistical significance and if the arrow sits above the red cross then this provides statistically significant evidence that the outcome is acceptable and if it also sits in the green zone requires no further amendments. If it does sit above the red cross but is in the amber region then we recommend that there would need to be smaller amendments to be acceptable. So the recommendations from this would be that recruitment uptake was better than expected as it was in the green region but follow-up and treatment of reality would need some minor work to be improved as they sit in the amber region. However, all objectives sit above the critical value and provide statistically significant evidence that the outcomes are achievable. So the recommendation from this application would be to recommend progression to a main trial given some minor amendments. So to conclude this application the simple application allows the newly proposed methodology to be easily implemented as it's not available in any of the packages. It will facilitate improved design monitoring and evaluation of pilot and feasibility studies which in turn will promote better research which helps to better inform stakeholders regarding the future progression of the study. Thank you for listening. Thank you, Kiran. Okay, we have one question in the chat which as we're getting our next speaker lined up Peter asks, particularly in the feasibility phase there are cycles in which you alter the protocol to improve feasibility. Is the goal here to provide sample size after you have finalized the protocol what some would call the pilot phase? So it's at the beginning really to get an idea about what sample size would be suitable and also accessible. Of course there are going to be changes as things change but it is to primarily get an idea. So you have more sound evidence moving to the next stage of the study but I'm happy to talk more about it in an email or something if you would like. Excellent, thank you, Kiran. We'll go ahead and move on to our next speaker. We can continue questions and discussion in the chat. Thank you so much.