Hi, everyone. My name is Stefan Buzoiano. I'm a Senior Scientist with Tonisity International. And today we're going to be talking about our journey from Excel to JMP automations. My email is here on the screen, so for any questions or suggestions, do drop me a like.

A bit of background, Tonisity PX is an isotonic protein drink which is administered to newborn piglets and provides hydration, and it provides key nutrients for their intestinal cells. So as a result, the piglets survive better, they grow faster, and they're healthy. So you can think of Tonisity PX as a power bank for the gut as it provides key nutrients to support the cellular metabolism, to enhance the cellular junctions, to make sure that the piglets don't have a leaky gut, to support the immune system, and to support the microbiome. As a main effect, we see improved piglet survival or reduced pre-revening mortality. And this is what we're going to be talking about in the next couple of slides.

We've learned across the way that our clients are starting to look at data more and more, and there's only so much you can achieve by showing one to five trials, because people will ask you about repeatability, will ask you about statistical analysis, will ask you about trends.

So by aggregating the data, our speech, our dissemination this material is much more powerful. So by showing results from 10, 20, 50, 90, 100 trials, our clients start to pay attention, and we're getting to more meaningful discussions and to more value added for our product or for our access to the market. So more data is better. This is the power of aggregated data.

To do this or to enhance our data capabilities, we're running these mortality meta-analysis about once a year. So we have six iterations to date. We started in 2018 with about 29 trials and 33,000 pigs. As of this month, we have 91 trials in our database with 185,000 pigs. So six iterations of this data analysis, and we're adding on the previous is one to where we're at today.

Looking at the effort on the time allocation and what went into this analysis, we see that back in 2018, we did a lot of data or data prep in Excel. The red color indicates Excel data prep. The orange indicates analysis between SAS and JMP, and the blue indicates dissemination, which is the value-added step. Our executives, our Board of Management, our clients are more interested in the dissemination aspect of it, not in the data manipulations, as you would think.

We started off with a very Excel-intensive process. It took us eight days to really figure out the database and work with all these formulas and copy and paste. There was a good bit of human error involved. 2019, what didn't add a lot to the data set, but we still didn't do a lot of dissemination.

We got JMP in 2020. We did an overhaul of our database, and from 2021 then, we started to do much more dissemination, and data manipulation went down drastically. You can see that we were already focusing on this value-added step on the dissemination side of things.

This is another way of looking at it split by the step in the process. So Excel data prep analysis and dissemination. We can see the drop in the amount of time we spent doing Excel. So from 8 days to 0.1 days, that's quite significant, and that adds a lot of value to our process. So less data manipulation, fewer errors, and fewer days spent working on this.

The analysis step We have stayed the same here in the middle, but we've moved from a SAS intensive process to a JMP intensive process, and that has its benefits because we're keeping a lot of the stuff in one place.

But the main focus here is that we went from half a day doing dissemination to about four days. And that adds a lot of value for day-to-day work. For example, we've submitted two abstracts this year. We've done some automations back in the last couple of years, and we've added some new analysis. So again, the time we've gained from not having to work with moving data back and forth in Excel was well-used in submitting abstracts, in disseminating the information, in holding webinars to our clients, and in talking to the academic community.

This is our data analysis process, so it's as ugly as a data analysis process could be. A lot of tables, a lot of interactions, a lot of summarizing, merging, and all of that. The main thing here is that a lot of it, a lot of the heavy lifting is being done in JMP. There's very little Excel, a bit of SAS, but mostly it is JMP.

This is how the process looks like in the background. We get the data set or the database place from Excel into JMP. We run it through SAS, we reabsorb it into JMP, and we come up with this master table, which is the backbone of each of our analysis. This is where everything else gets identified and moves forward. So we have a mortality meta-analysis overall by five geographical regions, by two administration protocols, and by four genetics. So About 13 components of this data table have been automated, and it's being merged seamlessly in the background by JMP.

We have some regular graphs then that are moving from iteration to iteration. The code is already written, JMP does the heavy lifting for us. We just run the scripts. And all these time savings allow us to focus, as I said, on data interpretation and on dissemination. We are able now to look at corners in the data sets, to look at when our product works very well and when things have happened and have yielded unexpected results. And this allows us to troubleshoot with our clients to make sure that we give them value when they're working with our product.

And this is what we basically do in the background. So there's a bunch of boring steps over here. We look at Excel, we bring it into JMP, we do some initial parsing, which we're going to be talking about in a second, and we're going to be demonstrating. We send it to SAS, we analyze, and it outputs in Excel, we bring it back into JMP, and we do a lot of merging mortality averages, these P values, other summaries, and we do some graphing and some PowerPoint expo.

Now, as I said, because JMP does the heavy lifting for us with formulas and concatenations and all of those in the background, we are able to look at data to do dissemination, abstract, webinar, and client training sessions and so on.

We go to our demo, and this is the moment where every JMP user pauses for a short prayer that the code will be working seamlessly. Let's have a look. This is my JMP project. Nothing too exciting, a lot of lines of code. What I really like about JMP is that it gave me this code folding options so as I can collapse entire sections of code. I'm loading everything into the memory, and I'm coordinating them, and I'm running these sections of code from this panel on the right-wing side. We're going to be bringing the tables in and start working with them.

We need to connect to SAS then. It will run some code in the background. We're exporting the table to SAS for it to do its work. SAS then is coming back with some quite ugly tables. There are seven tables with all P values and all sorts of estimates and things that really have to be concatenated and brought over into our master table format. JMP can do that for us, no problem. This is our master table. That's it.

Then we are left with a bunch of tables here on the left-hand side that we have no use for. Of course, we clean them up, we close them. Sorry, we're not good. Then the thing is, we're left with some tables here, I will have to bring the code for figures or for the graphs from an older version of the iteration.

We're doing just that. You can see the codes here, the scripts, and I can play around. These are my standard graphs. I can play around with them as needed. There's another one there. Perfect. Again, this is just an example of one of the graphs. We have mortality of the control on the X axis, mortality of the PX treatment on the Y axis, and these little flags. It allows me to understand where Europe trials are, where each of these dots is a trial in our database. There's a bunch of US trials with really high mortality there. Asia's ones are usually with a low mortality here at the bottom.

So it allows me to look through the data and have this flexibility and to understand what's going on in the data rather than playing around and bring in data via Excel from SAS and doing graphing in Excel. So that's a lot of time saved for us right there.

Perfect. Just a few conclusions on this. So JMP means that we do less data manipulations, so the JMP license pays for itself. We're getting less human error. The JMP software does the heavy lifting with all the formulas, all the concatenations, all the joining of the tables, everything you can think of. It does that in the background, so there's no human error whatsoever. It was all looked after when I put together the code. All this time saving allows us to focus on the value added steps. So looking at the data, writing abstracts, giving webinars, discussing with our clients. So adding more value to our time.

It allows us to do a lot of troubleshooting and interpret the data in various ways, and it allows us to do various different analysis in the data itself. Now, another main thing here is that we did all of this and started in 2020, and we had no previous JMP scripting experience. If I can do it, you guys can do it, too. Thank you very much. Again, my email is on the screen. If you have any questions or suggestions, I'd be happy to hear.