 Good morning everybody. My name is Rudolf Kruppmann and this is my colleague Markus Kalkbrenner. Good morning. What? Yeah, I already pressed it. Okay. I think it's recording now. Turn to red. Yeah, it's red. Okay. Yeah, we are happy that at least some of you showed up. I think there was some party yesterday. And we hope that you're all well caffeinated, so your body is somehow prepared for the topic today. Some are called logical. And yeah, the topic is content management as a critical success factor for personalized medicine. And then we like to take you into a journey into human genetics and personal medicine and talk about how Drupal can be part of it. So Markus as a CTO and we as a team leader on bioinformatics of the biologist genetic information management company based in Frankfurt. We are part of a team of highly motivated people from different disciplines with the aim to improve healthcare by bringing knowledge about human genetics into clinical practice. So before we talk about the technical aspects here, we start with a kind of an introduction bringing up some of the biological background and the motivation. So pharmacogenomics is basically tailoring the drug treatments to the genetic makeup of a person. So and very important here is that most of the drugs are effective only for a part of people. So for the majority at least, but there is a substantial part where the drug is not effective or even worse you can have some severe side effects sometimes. So typically often you have a mixed pool of responders and non-responders. So and of course the goal is to make all of them happy. So by adjusting the therapy for those where the drug is not efficiently first place, we have to find maybe alternatives here. So as we are talking about pharmacogenomics, of course the main elements are here. Our genes, our individual DNA which has a strong influence on how a drug is metabolized in our body. So I don't want to go into detail too much into detail here, but you can end up having an inefficient, efficient or adverse drug reaction mode of action. So to illustrate that with a kind of radical comparison here, you can compare the normal situation where you have a pair of chromosomes which has two functional alleles, which is the normal case to a bungee jump where you typically are secured by two safety straps on your feet. We have an efficient compatible mode of action of your drug, but imagine you have a mutation, so you end up having just one functional alleles. And also the situation for the bungee jump is not really safe anymore. So it's partially efficient and problematic and even worse. There are a few people which have a mutation on both alleles, so you end up having an inefficient, incompetent way and in this case also the bungee jump is not a good idea at all. And you really want to know upfront before you take the drug if you're a person that has this kind of genotype. So basically there are four kinds of phenotypes we have in metabolization. So on the very left you see the normal metabolizer which has two copies of the gene. And below you see the concentration of the drug which is shown in red and it's metabolite in green. So if you're a poor metabolizer which is one of the extremes, then you don't have a functional gene at all. So then your drug or your active ingredient which is part of the drug has a very steep increase as you can see in the red curve. And as you are not capable of metabolizing this drug, your metabolite curve stays very flat. So in this case it's the effective component is in the primary drug. You definitely want to lower the dose at here because otherwise you can have side effects because the drug is accumulated in your body and it's not metabolized at all. But there's a second case where the drug not itself is the effective component but it gets metabolized to the metabolite. So the metabolite itself is the active component and as you can see here the curve is very flat and you definitely want to push this a little bit higher. So then you would kind of increase the dosage above the normal levels which is also an alternative maybe to change the drug at all because it's not working in your case. And the other extreme is to the very right, if you're an ultra rapid metabolizer your body is really strong in metabolizing the drug. Then as you see in the red curve it gets degraded very fast and the metabolite rises very steep. And as an example if you have codeine for example which gets metabolized to morphine then you have really high morphine levels here which is not a really good idea. So in this case you definitely want to lower the dosage. So there are a bunch of guidelines around which are curated by consortiums worldwide. Here in the Netherlands they are really kind of leading in Europe. They have the Dutch pharmacogenomic group which brings out recommendations and these recommendations are based on studies. And as you can see for example in Ypres-Bien which is a pharmacone where you have, if you're a pro metabolizer you should increase the standard dosage to 28%. So as I said the other way could be of not modifying the orange dosage but you can also change maybe to another drug which works in your body. So generally it's a really good idea to know what's going on in your body. So you can be checked genetically because the development achievements during the last years were in this case so that it gets really cheap to be sequenced. And as we can see here in the speaker in the upper part the whole amount of genetic data has increased substantially here together with all the other data we get from other sensors. So you have this situation in the middle of the figure where you have a bottleneck and you have definitely a need to bring down all this information to those information which are really usable for the patient. And there's a need for systems who can do that and yes one family of systems who can support this are the clinical decision support systems which are defined as kind of linking health observations with health knowledge to influence health choices by clinicians for improved healthcare. And as these systems can be an answer to solve this question. Okay so let's talk about how to build such clinical decision support system and what are the building blocks of it. Yeah if you have a closer look at such systems that deliver the information one example was to identify the right dosage of a drug for a specific person or patient then we will recognize that everything is about content. In the first step we have to aggregate all kinds of content from different sources. Once we aggregated it we need to manage it continuously. We need to filter the content for the important information that is needed right in time. And finally we need to deliver this content to the point of care where it's really needed. So let's have a closer look. First of all we have a lot of different kinds of content to manage. We have clinical observations so if you have a look at a patient there are facts like the age, the gender, measurements like the blood pressure or different symptoms like a headache or what could happen so these are clinical observations and this is data. Then you get the data from the laboratory, measurements like standard data from blood like iron or cholesterol or on the other end more complex data like human genetic data. And the third kind of data you have to handle is knowledge. As Rudolf already mentioned we have different kind of data sources, public and private ones that deliver this kind of knowledge is built by experts and like the Dutch pharmacy group or others and so the starting point is that we have three different kinds and sources of content. So once we aggregated this content we need to manage it. First of all we need to provide the possibility to review and annotate such a content and also to rate the content and do an evaluation. For example you often will face the situation that you get knowledge from different sources and they don't agree on a special fact. You need to discuss, you need to decide on it and maybe also the kind of information changes regarding where you are in the world because we are kind of different. And if you did that once you need to update your judgment, your rating because there is new knowledge generated all over the world every day. And you also need to track all the changes and you need to build a history of these changes. And also an important fact, especially here in Europe is translation. So we mentioned the Dutch pharmacy group two times now and they provide a lot of recommendations but they are all written in Dutch. So how should a patient or a physician in Italy or Spain should use this information. So translation is an important fact. And finally you have to establish workflows and a lot of quality management and everything what you do. Because all the laboratories and if you deal with medical content and medical information systems you need a lot of certifications. You need to prove that every change in the system is wanted and not just happen. So when you manage your content and you can think of it that even if in personalized medicine you can create a lot of information for one specific patient. But for example in the case of emergency you won't read information of thousands of pages. You need the information for one special drug or one special circumstance. So filtering down the content is really essential and you need to be prepared for that. So once you have the content from different sources you need always to apply kind of normalization and clustering of the data. I already mentioned that we have translations and if you want to do it interactively in this filtering. You need to be able to search for this content and you need to be able to search in different languages. To go through this data you need to organize it. You need to build and apply ontologies and storage graphs. And also here is the possibility to automate a lot of things. And therefore you need to build rule-based systems, rule engines. And there is also a possibility for artificial intelligence to point you to the right data, to the right content where you can base your decision on. And once that's done this information needs to be transported to the place where it's needed. So it's a hospital or it's a laboratory and there are special kind of systems established in this field. So there are hospital information systems from different vendors. There are laboratory information systems all come with their own kind of interfaces and you need to interact with them. And where those systems are not in place you need to find different alternatives to deliver the data to the point of care. One possibility could be a mobile application or a website. And what we as a company do, we provide a product that covers all this area and deals with the content across the whole pipeline. And that's what we call Jim's, which means a genetic information management system. Like we have hospital information systems or laboratory information systems. Oh, sorry. So let's focus on some points here. I mentioned that the building block content management and obviously we are here at the Drupal Con. And there's an existing system that could deal with this job. And I think it was nine years ago we selected Drupal to be the system for this part. And so all the requirements are fulfilled and it makes a lot of sense to not build such systems from scratch but leverage what's existing on the market and even better if it's open source. Regarding content filtering, there are also systems in place, especially for search. There's solar, there are other alternatives and for dealing with graphs and also to prepare the data for artificial intelligence. There's a system called JanusGraph which is built on top of solar and the scene and also different graph databases and things like that so that you have a single point of contact with all the systems. Regarding the aggregation and the content delivery, you have to deal with different standards and most of the systems in the healthcare market are based on HL7 which is an old standard. It's not the best, it's not comparable to newer things based on JSON or XML. And so recently there was a new approach from the HL7 consortium which is called FHIR which leverages new interfaces based on REST for example and I think there are new players in the market who have a look at the delivery for example AppleHealthKit and internally they also use FHIR as a protocol but if you want to interact with the systems, if you want to get the data and deliver the data you definitely have to deal with all the standards, parts of them are 30 years old so it's a little bit of complicated to deal with them. And on top of that you also have proprietary appies you have to deal with. As a short example of this whole tool chain we can switch to my mobile phone and have a look at content delivery. That's live. Do you find the customer of us could use? It's called the pharma sensor and so I have to leave the microphone now. I can start a barcode scanner. So this is the desk here. So there's the drug I just scanned and I get an alert for a special gene and in real time I get the information for example that I have to lower the dosage of the drug by 50%. And this is a kind of delivery maybe for the patient but also for the physician at the point of care to quickly deal with the information. And I think this is really impressive and important because normally people would start at 100% and see what happens and then adjust the dosage over weeks, months, whatever. And you can predict a lot in this field. What's also worth to mention is that our product got certified. It's a certified medical device in Europe, a medical device of Class 1. And an interesting fact for all of you especially if you deal with Drupal that Drupal is part of the stack and got indirectly certified in Europe as a medical device which means that maybe compared to other systems we have proven quality here. And you should leverage or probably leverage that kind of information for marketing Drupal. So our system is in place with several customers but I want to emphasize on one of the very prominent projects where GIMS is used. There's a ubiquitous pharmacogenomics project funded in the EU within the Horizon 22 program. And this is currently in place. So our system supports pharmacogenomic consulting at seven selected hospitals across Europe. And that was very interesting and very challenging. But it also makes use of all the good features we have from Drupal like strong support for translation and also you can imagine if you have seven different countries with different regulations, different modes of content they want to deliver. So some content could, for example, in Austria the genotypes were not allowed to be in the reports. So a task like that could be managed really well by using all these features as shown here and so as well as you can imagine there are different setups in the laboratory you have to deal with different kind of data and we have really good feedback from these different countries and they really want to use our system beyond this project. So here we have again this figure with the bottleneck and I think we have shown that the answer can solve this problem by bringing down the content to the very useful information which helps the physician to consult the patient so that you in general improve healthcare here and we have shown that personalized medicine is impossible without an efficient and effective content management and it is also a good example to use best practices developed in other technical areas and they can really apply it also in healthcare. Okay, as a reaction of, I don't know if everyone attended the Drees Note so we want to tell you a little bit more what we are doing at Biologis and we consider ourselves as makers not as takers as Drees introduced because we are not just leveraging Drupal, we are not just leveraging solar we are giving back, we are actively developing these systems and so I think this is two years in a row or three years in a row we as a small company are listed as the one in the top 30 contributors to Drupal and we are directly maintaining and sponsoring various projects on Drupal.org and I just want to highlight some. For example, we are entirely maintaining the search API solar module for 8.x and these are just the numbers for the 8.x versions there are 15,000 active installations measured on Drupal.org and as you can see all three people actively maintaining it are working at Biologis and underneath or under the hood of the search API solar there is the solarium PHP library and it is also maintained by me at the moment we have more than 5 million downloads via packages and it is also used by other content management systems like Type 3 or WordPress and there is also symphony integration but all this is driven by the need of Biologis and the systems we build also one of our team members is Worstow-Johnow he is an active entity API maintainer of the Drupal core and also created all the latest creations are the autosave and the conflict module which we need for content management and I think these will be also interesting for the core and future for the standard distribution and I just want to mention that's how we think of it it's a kind we want to give back and we want to increase the quality of our software by a wide usage of our modules and every bug report or every improvements are just that improves our products another thing I want to highlight is during the development of Drupal 8 we were really active in the multilingual initiative because translation really matters for us and as an example it took months of work but the result was that we managed to get proper revisioning with a new translation model introduced in Drupal 8 so this was built at Biologis one more thing you want to mention besides thanking you for your attention one of our customers is providing such genetic tests and unfortunately we have two of these packages available one we want to give away, I don't know if you attended our booth already where you can play in a card game and the one who gets the high score until I don't know this afternoon or noon we'll get one of these packages and the other one will be referred between everyone who wants to participate so thank you thank you for everything you're doing I'm curious to ask you about the app that you're using and what are you using there for technology and is this then a decoupled Drupal? that would be the first question so about the app and the second thing would be how are you handling security as obviously with your app everybody can get to this information so what are the security standards that you're using there? okay first the technology that is used for the app if you look back in our short history we were always early adopters or as I showed inventors of some things and so the communication layer is entirely built on JSON API before 8.8 so we started earlier and found a lot of bugs right? and security is set up also on established technologies obviously everything is encrypted the communication is encrypted the data is stored encrypted so let's say you have to use or ensure that all the available technology is simply used you must not forget to turn on some things but another key thing is you add some additional features or you have to ensure that the patient could only access his own data and if you talk about data privacy it's up to the patient to decide depending what kind of app you are using if he wants to share this data for example with a precision or if it should be stored in a national health record as well so there are so many things to do but we always use standards like JSON API or at least try to answer everything My question is how do you handle the content aggregation updates into the content management block to avoid potential conflicts? Yeah that's one of the main parts of our system to deal with the conflicts and that was also the driver why we invested so much in having a proper translation revision system in Drupal 8 basically you must not delete the old data and simply store the new data you need to maintain all the relationships and you need to update the data and you need to track conflicts that occur for example and build a kind of work list and let humans decide in the case of conflict because it's a fact that even at the starting point of the chain and the data aggregation the data is not always correct so you also have to not just deal with updates with errors and the difficulty here is to build a system that enables people to do their work and to focus on what they are doing and to hide the difficult technology behind it but the entire workflow could not be explained in a few sentences because this is the result of years of development and parts of it we are giving back you mentioned the conflict module or the auto-save module that no information gets lost when you work with the content anything else? you have said that it's kind of sponsored by the European Union this is just like a project or it's like ongoing also said like the possible it's one to yes make progress or keep it ongoing and yes this is one question is it just the only project or it's ongoing and how can maybe other hospitals apply to this project or ongoing project because I'm thinking there's a great vision behind it maybe for Europe or the world to have this database so that the drugs can be prescribed according to the patient's DNA and so I think it should be ongoing absolutely so first of all we are a company so we are selling products and services and beside the hospitals Rudolf mentioned or shown on the slides we have other customers and the nature of this Horizon 2020 project of the European Union is the goal is to have a commercial output from it we are pretty sure that the hospitals within this project will stay as a customer and because this European Union project was set up to demonstrate the importance and how it affects the quality for patients in the hospitals regarding less complications and I think the goal will definitely be achieved and we already have customers influenced by this project for example one of our recent customers well not recent customers but they started last year already as the biggest hospital in Stuttgart in Germany for example and we also have customers in Switzerland and they are not part of the European Union so it is available I think it's also affordable and the next step would be to also we want to establish a kind of collaboration because we have experts giving information in the system and the bottom line is that we think that participants of the healthcare market could learn from the open source scene on collaboration, on knowledge and I think there are a lot of possibilities and we also have things in mind and on our roadmap to go in this direction Questions? Is the app available? Can I download it and use it? Yes, it's called Pharma Sensor or if you look for it I think it's listed as Bielogus in the App Store I installed it long time ago so I don't know but while I'm waiting for others You can access the Play Store from your computer and just throw it I got it You got it? Pharma Sensor So this is one example there are more possibilities because this application is currently or this app is currently focusing on pharmacogenetics or the effectiveness of drugs but that's not the limit We also support use cases for different customers like nutrition or prevention so there are so many useful things you can do and this app is specific just for one use case but that's definitely a field of growth for the future as well and what you can see as information you can play with it then the information depends on the laboratory that provides your data because from different laboratories you get different amount of data and for example if you are in one of these hospitals then you get the information for the panel they agreed on or the medication they agreed on for the European project but there's more to deliver so that will grow So, thank you for your time enjoy your last 8 years of Dr Pepper