Quadrifolium
Hi, Quadrifolium is an open (formal) language in the social & health domain... that targets openEHR as a natural application. More there : http://www.4folium.org/ Even more as answers to your questions on this list. Even even more during the launch event, August 29 (from 6h pm) in Lyon, at a restaurant located in the Medinfo venue (please tell me if you intend to join). Best, Philippe ___ openEHR-clinical mailing list openEHR-clinical@lists.openehr.org http://lists.openehr.org/mailman/listinfo/openehr-clinical_lists.openehr.org
Re: Science of Machine Learning (was Machine Learning , some thoughts)
BTW, is someone aware of this project by Google? https://ai.googleblog.com/2018/05/deep-learning-for-electronic-health.html Le 03/07/2018 à 12:40, Birger Haarbrandt a écrit : > Hi Philippe, > > I completely agree with your view. This is why data stewardship is > needed before we can make real use of the data: > https://en.wikipedia.org/wiki/Data_steward > > As we use this approach in HiGHmed, I might be able to report in 2020 > about lessons learned :) > > Best, > > -- > *Birger Haarbrandt, M. Sc. > Peter L. Reichertz Institut for Medical Informatics (PLRI) > Technical University Braunschweig and Hannover Medical School > Software Architect HiGHmed Project * > Tel: +49 176 640 94 640, Fax: +49 531/391-9502 > birger.haarbra...@plri.de > www.plri.de > > > > Am 03.07.2018 um 12:21 schrieb Philippe Ameline: >> Le 02/07/2018 à 11:31, Bert Verhees a écrit : >> >>> On 30-06-18 17:16, Philippe Ameline wrote: >>>> (improperly labeling images or adding images of objects that are not >>>> plants) could probably make the whole app plainly crappy. >>> Of course Philippe, but that would be vandalism. Most sensible people >>> don't do that when they stand behind the goal, and a little bit of >>> dirt, therefor it is Machine Learning, it can filter it out. It is >>> part of the learning process. >> If a culture of data quality is properly installed, then it is possible >> to name improper use "vandalism". >> In medicine, since such a culture has never existed, we could name it >> "don't carisme", "no time for thisisme" or "was never thaughtisme". >> >> My point, and what the paper I previously pointed out explains, is that >> trying to get something out of machine learning in a domain of poor data >> quality is a modern kind of magic thinking. >> It just means that any such project should first organize for data >> quality as a first step. >> >> When considering it in hindsight, it makes sense since machine learning >> involves statistics and data quality is paramount in this domain. >> >> >> ___ >> openEHR-clinical mailing list >> openEHR-clinical@lists.openehr.org >> http://lists.openehr.org/mailman/listinfo/openehr-clinical_lists.openehr.org > > > ___ openEHR-clinical mailing list openEHR-clinical@lists.openehr.org http://lists.openehr.org/mailman/listinfo/openehr-clinical_lists.openehr.org
Re: Science of Machine Learning (was Machine Learning , some thoughts)
Le 02/07/2018 à 11:31, Bert Verhees a écrit : > On 30-06-18 17:16, Philippe Ameline wrote: >> (improperly labeling images or adding images of objects that are not >> plants) could probably make the whole app plainly crappy. > > Of course Philippe, but that would be vandalism. Most sensible people > don't do that when they stand behind the goal, and a little bit of > dirt, therefor it is Machine Learning, it can filter it out. It is > part of the learning process. If a culture of data quality is properly installed, then it is possible to name improper use "vandalism". In medicine, since such a culture has never existed, we could name it "don't carisme", "no time for thisisme" or "was never thaughtisme". My point, and what the paper I previously pointed out explains, is that trying to get something out of machine learning in a domain of poor data quality is a modern kind of magic thinking. It just means that any such project should first organize for data quality as a first step. When considering it in hindsight, it makes sense since machine learning involves statistics and data quality is paramount in this domain. ___ openEHR-clinical mailing list openEHR-clinical@lists.openehr.org http://lists.openehr.org/mailman/listinfo/openehr-clinical_lists.openehr.org
Science of Machine Learning (was Machine Learning , some thoughts)
Le 27/06/2018 à 22:26, Bert Verhees a écrit : > On 27-06-18 16:43, Philippe Ameline wrote: >> 1) you can find a bunch of practitioners that agree on working extra >> hours to comment a big bunch of images, or > > Did I tell you about the plant-app? I believe I did. 700.000 pictures > are reviewed, often by volunteers. > > The app recognizes 16000 plants. Important is how you do it, and that > it does not cost effort by the volunteers, for example in relation to > what they do anyway. > > https://plantnet.org/ > > It is a French product. Dear Bert, The plant-app was the subject of your initial post. The math in support of deep learning are being studied. To make it short, it remains somewhat mysterious since such classification algorithms "should not work", but actually, they do ;-) From an article I just read, such NP complete algorithms are similar to finding a needle in a hay stack and shouldn't provide valuable answers... unless the conditions (large enough needle, correctly ordered stack) make the problem handy. To sum it up, data quality (signal over noise ratio) is paramount. In the plant-app you mentioned, adding a certain level of fuzziness (improperly labeling images or adding images of objects that are not plants) could probably make the whole app plainly crappy. Just to say that building a deep learning system starts from making certain that the data it will be fed with are of proper quality. This is usually not the case in medicine, largely because IT is considered a back office concept and there is seldom the kind of feedback loop that could lead to having errors fixed. My point is that you can perfectly (but with considerable efforts) organize a trained network of practitioners to feed a "data lake" in order to train a neural network... but will probably be disappointed if you try to process existing information. Best, Philippe ___ openEHR-clinical mailing list openEHR-clinical@lists.openehr.org http://lists.openehr.org/mailman/listinfo/openehr-clinical_lists.openehr.org
Re: Machine Learning , some thoughts
Bert, I don't think that we really disagree there. As you nail it the dataset comes from people agreeing on building it the proper way. And agreeing with Karsten (who is plainly right), doesn't make that process simple. Means that wether: 1) you can find a bunch of practitioners that agree on working extra hours to comment a big bunch of images, or 2) you expect this process to be(come) part of the usual information recording... and you must instill a culture of data quality and information awareness before the dataset can exist. Le 25/06/2018 à 15:22, Bert Verhees a écrit : > On 25-06-18 14:47, Philippe Ameline wrote: >> Successfully using machine learning demands a prior culture of data >> quality and information awareness. > > Dear Philippe, I read your document later. > > I have to disagree with the word "prior". > > It makes it sound like, is has gone wrong long time ago, and there is > nothing what we can do. > > Big data for machine learning can be build very quick, we have > millions of people in healthcare every day. > > Imagine a GP making a picture of an eye, or a part of skin, and gets > within a second a good explanation about what is there to see. > > It is cheap. If many GP's agree to use an app for classifying viewable > symptoms, the supporting big database will grow fast. > > I also have to agree with Karsten, it is not only a disease which > needs to be cured, but it is a person having that disease. So, age, > weight, gender, ethnicity, profession, social status, country, those > are all factors which limit the search area in which the machine > learning database must find what it sees. > > > ___ > openEHR-clinical mailing list > openEHR-clinical@lists.openehr.org > http://lists.openehr.org/mailman/listinfo/openehr-clinical_lists.openehr.org > > ___ openEHR-clinical mailing list openEHR-clinical@lists.openehr.org http://lists.openehr.org/mailman/listinfo/openehr-clinical_lists.openehr.org
Re: What to call this concept?
Great answer, I also liked Bert's "Only the first man was reflecting the truth because, laying bricks, you do alone, the rest you do together with others" To sum it up (my own way): When asking someone what he does (in life or even for a living), the answer - from people doing the very same thing - can range from very mundane to fully inspirational. In health, all answers are plainly valid, and it would be a terrible mistake to only take into account the technical description. We are all, on this list, working to save lives, to have suffering people heal, to open the way for a better world... and we are also all spending most of our time typing on a computer keyboard (be it code or projects sheet, etc). It is true that typing allows us to benefit from several years of life when compared to those who are pneumatic drilling outside... but it doesn't define who we are. Bert reminds us that the most important part is always done with others... meaning that "what do you do?" could be better expressed as "what is your social network?". It may remain a little bit theoretical or philosophical, but the "individuation" concept states that, to become a better self, one should find her best place as a service provider to her network (hence opposed to "individualism"... and also opposed to finding a comfortable slot in a hierarchy). This to say that a human being is a complex biological entity part of a complex social network and that taking care of her can never be achieved in a reductionist way. So, "what do you do?", "what is your social network?" and "who are you?" are three very connected questions. I stop there... my virtual foreman reminds me that I don't make a living from posting on this list ;-) Best, Philippe Le 21/06/2018 à 19:24, Pablo Pazos a écrit : > And the foreman came along and said "I'm an atheist and you work for > me, stop talking and do your job". > > On Thu, Jun 21, 2018 at 11:46 AM, Philippe Ameline > mailto:philippe.amel...@free.fr>> wrote: > > Le 15/06/2018 à 08:41, Bakke, Silje Ljosland a écrit : > >> >> >> A typical question that would lead to this concept could be “What >> do you do?”. >> >> >> > > A man came upon a construction site where four people were working. > He asked the first, “What are you doing?” and the man replied: “I > am laying bricks.” > He asked the second, “What are you doing?” and the man replied: “I > am building a wall.” > As he approached the third, he heard him humming a tune as he > worked, and asked, “What are you doing?” The man smiled, “I am > building a cathedral!” > The forth was very concentrated, nearly ecstatic, and when asked > “What are you doing?”, looked up at the sky, and answered, “I am > working for the sake of God!” > > ___ > openEHR-clinical mailing list > openEHR-clinical@lists.openehr.org > <mailto:openEHR-clinical@lists.openehr.org> > > http://lists.openehr.org/mailman/listinfo/openehr-clinical_lists.openehr.org > > <http://lists.openehr.org/mailman/listinfo/openehr-clinical_lists.openehr.org> > > > > > -- > *Ing. Pablo Pazos Gutiérrez* > pablo.pa...@cabolabs.com <mailto:pablo.pa...@cabolabs.com> > +598 99 043 145 > skype: cabolabs > Subscribe to our newsletter <http://eepurl.com/b_w_tj> > <https://cabolabs.com/> > http://www.cabolabs.com <http://www.cabolabs.com/> > https://cloudehrserver.com <https://cloudehrserver.com/> > > > > ___ > openEHR-clinical mailing list > openEHR-clinical@lists.openehr.org > http://lists.openehr.org/mailman/listinfo/openehr-clinical_lists.openehr.org ___ openEHR-clinical mailing list openEHR-clinical@lists.openehr.org http://lists.openehr.org/mailman/listinfo/openehr-clinical_lists.openehr.org
Re: What to call this concept?
Le 15/06/2018 à 08:41, Bakke, Silje Ljosland a écrit : > > > A typical question that would lead to this concept could be “What do > you do?”. > > > A man came upon a construction site where four people were working. He asked the first, “What are you doing?” and the man replied: “I am laying bricks.” He asked the second, “What are you doing?” and the man replied: “I am building a wall.” As he approached the third, he heard him humming a tune as he worked, and asked, “What are you doing?” The man smiled, “I am building a cathedral!” The forth was very concentrated, nearly ecstatic, and when asked “What are you doing?”, looked up at the sky, and answered, “I am working for the sake of God!” ___ openEHR-clinical mailing list openEHR-clinical@lists.openehr.org http://lists.openehr.org/mailman/listinfo/openehr-clinical_lists.openehr.org
The Architecture of Open Source Applications
Hi, In case you didn't already know about: http://aosabook.org/en/index.html Best, Philippe
greetings and 2 questions
In Europe (maybe except NHS) this would nowhere be possible. In the USA I believe they have similar regulations. Bert, You can count France as another counter-example. The DMP (Dossier M?dical Personnel aka Personal Medical Record or Shared Medical Record or Public Medical Record) was heavily founded (something around 1B?... and counting). The interesting effect was to shift concerns from reinventing health care (as was in the early 2000s) to interfacing the government platform (that, by the way, never delivered). Such gov behavior doesn't only kills the market, it profoundly destroys the very nature of innovation. Best, Philippe
greetings and 2 questions
Le 21/04/2015 11:26, Bert Verhees a ?crit : So why bother the majority with privacy issues which deliver no advantage for them. IMHO, the interesting question to be asked is what is the reference frame you are working in? The usual reference frame (Cartesian) is centered on the organization, say the hospital, and is fit to tell the story of people passing through (say the patients from in to out). Another reference frame (Polar) is centered on the person, and considers the world as what surrounds Mrs Smith. It is fit to tell the life long story of an individual and to organize the work of her (health) providers. The Cartesian reference frame is local and based on a hierarchy of local roles (homogeneous to a domain). The polar one, on the contrary, travels with the person it is centered on. When creating a Public Health Information System (PHIS), the usual (wrong) way is to extend the walls of a Cartesian reference frame in order to try to cover the whole country. In my opinion, it is like trying to manage a lake in the same way a fish tank would be kept safe, through a giant air pump, a huge filter and a building sized electric heater. It seems to me that open systems and closed ones must not use the same reference frame. In an open world the person herself must be in charge, and privacy issues are more than mandatory, it should be the very starting point of any project... at least any project that makes sense to me, both as a developper and a citizen ;-) Best, Philippe
New paper: Analysis of Clinical Information Modeling Processes
Le 24/03/2015 12:17, David Moner a ?crit : Dear all, My colleagues Alberto Moreno, Wellington Dimas, Marcelo R Santos, Jos? Alberto Maldonado, Montserrat Robles, Dipak Kalra and myself have just published a paper titled *Clinical information modeling processes for semantic interoperability of electronic health records: systematic review and inductive analysis* in the Journal of the American Medical Informatics Association. It can be found in http://jamia.oxfordjournals.org/content/early/2015/03/20/jamia.ocv008.abstract?ijkey=ArRoJ7C7gQpzLCwkeytype=ref We have reviewed existing clinical information modeling methodologies and analyzed them to find out their similarities and differences. We hope this work can be of interest for you. Dear all, It seems to me that the pivotal point of the conclusion is all of these methodologies share the idea of separating the definition of the CIMs from the actual representation and persistence of the data values. As for the persistence, I am not certain that I properly understand the point. According to the 2 levels paradigm, the persistence system, as all other components, are generic regarding archetypes/templates implementation, but still specific to the global information model (typically archetypes syntax). It leads me to another (truly ontological) question. As human beings, we have put the representation of data first when it comes to communicating : we just needs common words (vocabulary), in a specific order (grammar). Obviously, we don't need any Global Information Model. Do you think that the need for a CIM comes ab initio from the lack of a common ontology or from something else (technical, education related...)? Don't you think that, at a time when health is recognized as far more than just medicine, a CIM (with a C for Clinical) is not a synonym for silo? Such questions may seem quite provocative, but they are clearly not. The world of information management is evolving really fast (outside ;-) ) and I feel that it is not useless to question the very root of usual thinking - possibly to strengthen them. Best, Philippe -- next part -- An HTML attachment was scrubbed... URL: http://lists.openehr.org/pipermail/openehr-clinical_lists.openehr.org/attachments/20150326/c337d2a2/attachment.html
Clinical Modeling - A critical analysis
Le 11/03/2015 13:41, Thomas Beale a ?crit : In that case, I suggest a starting point is to dig out the original article and come up with a framework /headings for an article that properly addresses the same questions, providing evidence from the many projects around the place (I meant to mention Linkoping, and I see Mikael Nystrom has chimed in). I would suggest an off-list email loop for this. - thomas Hi all, Being myself pretty ontology based, and maybe more prone to understand what Blobel meant (understanding isn't approving ;-) ), I would be glad to be part of this group writing process. Typical question that could be asked, since we all tell stories in natural language by making sentences made of words arranged according to a grammar (but grammatical concepts are nowhere inside our sentences), is why should we need an external structure such as the one present inside the Archetypes to tell a patient health journey? Answer may be that there is no universal/commonly agreed ontology and grammar... but the structure that compensates for this as a grammatical exoskeleton could appear somewhat dated would the web 2.0 provide patient centered languages. Best, Philippe -- next part -- An HTML attachment was scrubbed... URL: http://lists.openehr.org/pipermail/openehr-clinical_lists.openehr.org/attachments/20150311/13ff1f4c/attachment-0001.html
protocol definition (was maximum heart rate)
Hi Stef, This is correct. Healthcare plans belong in a separate 'workflow management system (WMS)' that 'instruct/ suggest' what should happen and 'oversees' what has happened. This is not part of the EHR itself and should be registered elsewhere. Nonetheless it is an important feature to support and (qualitity) management of procedures, which most healthcare providers might want to have. In my opinion a WMS for routine and clinical trial purposes will function similarly but with different guidelines/ protocols. Cheers, Stef I don't get your point when you write that suggesting what should happen and overseeing what has happened shouldn't be part of the EHR itself. To make it short, what remains if you put this away? Cheers, Philippe
Archetype vs. ontology
a Fil guide will do. You have many Fils guides in a big bag, and when the user is somewhere, you find the more relevant Fil guide (if any) : more relevant means the one whose path is the semantically closest from user actual current path. But the Fils guides are just oppostunistic description support in a non deterministic domain. So the data don't remember the Fil guide they come from. This (too) long description to explain that Fils guides neither belong to the reference model, nor to the ontology, but are interface components in a knowledge management system. Currently, we have nearly 3500 Fils guides, but most of them are used for our report management system and should be replaced with archetypes. By the way, the Fil guide engine, that decides which Fil guide to throw, can also decide to throw an Arcehtype if the user has entered a part of domain where a deterministic description should occur. And you also can go beyond the leaves of an Archetype using Fils guides (or just using the ontology by hand). I hope that all this is understandable ;o) Philippe AMELINE Hi, I just forgot to tell you that our ontology has only 50 000 terms (it means less than 50 000 concepts, since a concept can be represented by several terms). As you may have understood, the ontology contains only basic concepts, since complex concepts are expressed through a small tree. 50 000 is more than a standard medical dictionary (say a dictionary + anatomical terms + drugs international standard names), so it really represent the words used in the medical domain. to clarify a bit: in Philippe's system, there are 50,000 concepts in a compositional terminology, as well as the fils guides (guide paths) whch are little structured post-coordination rules defining legal ways to put the 50,000 concepts together in coordinated tree structures. This combination of a small, clean lexicon, and the fils guides are what makes Philippe's approach to terminology so exciting, in my view. Philippe - one question I have never asked - how many fils guides do you have currently? - thomas - If you have any questions about using this list, please send a message to d.lloyd at openehr.org - If you have any questions about using this list, please send a message to d.lloyd at openehr.org
Archetype vs. ontology
Hi, Sorry not to have been able to make a description of the unified model in english yet. However, I can see it pretty well in the overall direction Gerard described his target system. We use an ontology in order to provide the langage words The patient record is a graph : each document is a tree, and typed links can be established between any nodes inside the trees Archetypes are used as tree molds (for the whole tree or tree parts). So, you express things by building trees with elements from the ontology in the same way you would express yourself in natural langage by making sentences made of words. A tree is a kind of discourse schema and Archetypes are pre-elaborated discourse patterns (classical models) Inside the graph, trees belong to 4 information model levels : the root trees, the system management trees, the data organization tree, and the description trees The root trees are : the root tree (root ontological concept person), the administratrive data tree (name, date of birth, adress...), the health professional data tree (for health professionals : job, abilities...) Each of these trees are linked to the root tree with specific links (ie a admin link between root and admin tree, a prodata link between root and health professional data tree...) The system management trees are : the contribution trees (for a given contribution : who, when, where...), the document trees (meta-data of a document : author, type, title, where and how you can get the document...) There is a contribution link between the root tree and each contribution, and a document link between the root tree and each document. 3 links (created, modified, read) can connect a document and a contribution in which it has been created, modified or read. The organization level currently contains a single tree (root node health index) that contains sub nodes health concerns (health problems and prevention follow up), health goals and treatments This tree is linked to the root tree by a specific link, many other trees and nodes are linked to the health index (for example POMR links between a documents or nodes of a document). The description trees contain structured description documents. The Ligne de vie server allows local systems synchronization of the 3 upper levels (root trees, system management and organization). The description datas are accessed through services under the Ligne de vie server control. As you can imagine the ontology contains medical concepts like diabetis melitus, patient weight, kilogram, colonoscopy, and so on, but also system management concepts as document, contribution, file... We have been working 2 years with a knowledge management research team (from french well known INRIA : national institute of research on informatics and automatics) in order to see where and how we could manage points of view. In the knowledge management field, the point of view is the sum of a view angle (for example the job : doctor, nurse...) and a focus point (in the medical field it can be seen as the speciality : cardiology, gastroenterology...). The INRIA team proposed to specify ontological concepts from a point of view, but I was very reluctant to this idea because it in unmanageable : for example, you could say that migraina means simple headache for a patient, while, for a doctor it is a specific disease. However a patient with a genuine migraina will never make a confusion between both concepts. I just forgot to tell you that our ontology has only 50 000 terms (it means less than 50 000 concepts, since a concept can be represented by several terms). As you may have understood, the ontology contains only basic concepts, since complex concepts are expressed through a small tree. 50 000 is more than a standard medical dictionary (say a dictionary + anatomical terms + drugs international standard names), so it really represent the words used in the medical domain. I hoppe I can make a more structured description soon... Regards Philippe Gerard Freriks wrote: Dear Bernie, I'm not calling for a universal model. What I wrote was that we need a 'language' so we are able to express what we need. We need building blocks only and a evolutionary process so we are able to adopt to the changed perceptions of the real world. And I even agree that the Ontology of everything is utopian. I didn't read Gerard's mail as asking for the utopia of a universal ontology. But some people do believe in this, and in some way things like snomed-ct are an attempt at it. My experience as a software engineer is that universal ontologies don't work for exactly the reason Bernie mentions - everyone is an individual, with his/her own point of view; humans are just not synchronised to the extent that they can create a single ontology or model of the anything which is not extremely narrow in scope, and correctly represent all their points of view therein. During many years