 Okay, then as the last speaker for today, Etienne wants to talk about how to use graph concepts in the medical domain. And as a surprise speaker, I'm really interested in what you have to answer. Thank you for your last minute invitation. This project is... You have to speak a bit louder because the microphone is only for the recording. Okay. This project is an experimental project in the scope of the open source working group of the International Society for Telemedicine, this interactive medical mind maps. You could also call it a look at natural intelligence. The goal is to make a visual representation of medical information. The problem is that I have to know many medical record software where just narrative reports like paper, paper on screen. And my vision is that we have to move to a more task-oriented model. Of course, the goal is to help to solve the patient's health problem, not just to describe what did happen. And I want to have visual support of what people are thinking because it is important to talk to each other across the internet to understand what the other partner is thinking. It can be used for training with students. And this mind synchronization is between members of the care team across the internet in Africa of several thousand kilometers away. I am missing a slide somehow. No, no, no, it was not on the key. Okay, no, no. No, okay, there is a graph but you have seen many graphs today. That is not important. We make a graph with several kinds of things. Observation presented in yellow. What has been observed, measured. There is no discussion about the blood pressure or any measurement. The health issue is the concern requiring attention because medical records cannot contain everything about the patient. What are the concerns for this particular patient? The action is to decide what to do, either more exploration or treatment. And of course at the later stage we will have links to medical knowledge which more or less already exist in database but not necessarily in Neo4G. The attributes of nodes and also of edges are very important because we don't want to have a black or white report. We need the degree of belief, we need the importance of course from the judgment of the author and many other attributes. Attributes can be a long text but I believe the challenge is to make a graph representation in such a way that when a patient is coming, a patient I do not know from a colleague or from one year ago, on one screen I want to have the essential things. A patient has five problems, we are doing that and that and the problem today is maybe something new. Then of course we use navigation, flying over a node will expand the content. Clicking on a node will move to related topics and a right click will make possible to modify and add information. I just spoke about training a student. Of course the medical records today are not in that way but why not to make a transition and use a graph as a kind of table of content and the node may content a discharge letter. Why not? That is the idea. I am looking for a partner helping of this project. Of course medical colleagues but here technical people how to solve many challenges. There is a begin on the GitHub and the system is using the D3 for the visualization with viz.js for the interactive presentation. You will find more information there and please contact it. If you are interested, I need help. I am learning JavaScript and I am not very far. Thank you very much. Thank you for your presentation. You listed the types of nodes on one of the slides like attributes, observations, but there were no mention of the pharmacy. Did you plan or do you intend to go further with the documentation and the medicine to connect the nodes with attributes? The group of nodes called action include many things you can ask for. You can ask for more information, ask for a lab or you can ask for a drug, a prescription. At this time there is not yet very much development but there is an important use of graph about product interaction and there are databases. I do not intend to reinvent that but to make an interface. The goal of this project is to have an interface during a consultation and then you will move on other things but to have a kind of synthesis and one screen graph telling that the patient has hypertension and is taking that drug for them and they have an infection taking some antibiotics and if there are any problems between these drugs we have to show it on the first screen because in many situations you have 5 to 10 minutes for every patient because it must be really efficient and there is of course a lot more to be done about presentation and human interface. This is your graph picture, right? Yes. This is your app actually? Yes. Can I explain it here? Yes, you can move on some... and you get the detail there is no yet real data there and you can click on what did happen on a given day and there will be of course a trade-off. Some information has to be simply as attribute in key value and other have to be real Neo4G but do not think to put everything in Neo4G because then you have a very, very heavy graph. I mean if it's just simple attributes then you can just put them there so it doesn't really matter because it's... I hope you can change it you can make a program modifying some things my vision is to have such a graph and if later it is more efficient to change the structure you can have to change the structure and keep the same information. What's interesting, we had a life sciences workshop in Berlin and there was actually someone who had similar interests to yours from the Netherlands let me find it that was here Substantially improving healthcare proteome here so Peter van den Berg from MSNPL and he had similar ideas to yours I don't know Peter van den Berg it would be definitely... definitely interesting for you to look at what he's done and I can put you into contact with him if you want to Yes, yes please because he also wants to improve healthcare general by adding more connected information to how treatments are done diagnosis are done how providers interoperate how people treat drug dependencies for instance if you have side effects and stuff like that and if you have all these connections available for decision making then it's much easier to make good decisions for your patient and they come from critical wound care clinics in the Netherlands so that's really interesting and of course there is today not yet many data but at a later stage of course pattern recognition will follow first make it usable in the daily work Okay cool Other questions? Other questions? Okay cool then thank you all so much for being here