Discussion Gestion:Partenaires/Suivi/Liste/57

À propos de ce flux de discussion

Non modifiable

Antoine Mercier-Linteau (discussioncontributions)

J'ai eu une discussion téléphonique avec ce qui semble être le responsable du projet. J'en ai consigné l'essentiel sur Sujet:Vk7h99adcc0a0v9b. En gros, il m'a confirmé que le contenu était bel et bien CC-BY. Il m'a par contre longuement discouru sur la difficulté que nous allons éprouver à mettre à jour ce contenu lorsque celui de StatPearls changera et qu'il vaudrait beaucoup mieux nous fier au contenu que StatPearls compte traduire en français d'ici quelques années. Dans un sens, il a raison. J'ai tenté de lui expliquer grossièrement ce qu'était Wikimedica, mais j'ai eu de la difficulté à faire passer l'information au travers de notre conversation téléphonique.

Un courriel en bonne et due forme s'impose afin de correctement décrire Wikimedica (en plus de confirmer avec l'équipe éditorial qu'ils sont satisfaits avec la manière dont nous créditons leur travail Wikimedica:Tâches/Liste/258/16/1).

This email is in reference to a phone conversation we had on the 9th of April.

First of all, let me reiterate how impressed I am with StatPearls. Like you, I believe medical knowledge ought to be open access and StatPearls shines as pioneer in this regard.

However, I believe I was not able to properly convey what Wikimedica is over our phone conversation. While it seems we are simply trying to replicate your content in French, let me assure you it is no so. Please allow me to give you an overview of the project.

Wikimedica aims at improving the way we create and disseminate medical knowledge in three ways:

  1. By creating an open access point-of-care evidence based medical resource.
    • Think UpToDate, but open access.
  2. With a dynamic and horizontal peer-reviewing process.
    • We believe knowledge nowadays evolve too fast to be compatible with the standard peer-reviewing editorial process. On Wikimedica, users of the platform also have the possibility of reviewing and commenting knowledge and can verify modifications made to pages on the fly. Safety is guaranteed by a strong versioning system and the fact that editors on Wikimedica are bound by the deontological codes of their respective professions when interacting with the platform.
  3. By providing a knowledge structure that is both usable by humans and computers and with which Artificial Intelligence (AI) systems can interface with.
    • AI is going to be an integral part of how healthcare is conducted in the near future and proof will be soon made that a clinician paired with an AI improves patient outcomes. The private sector is already on it and building AI systems as we speak. Even more so that normal computer programs, AI are black boxes. These systems, if left entirely private, will be fraught with conflicts of interests as we will have absolutely no way of telling how an AI came to recommend such treatment for such patient. Moreover, feeding those AIs with patient data, which they will need to achieve their full potential, will put patients at a great risk of privacy breeches. AIs are knowledge inference machines and if entirely the property of private third parties, they will further dispossess us from the science and know-how that underpins our healthcare systems. I am not saying that the private sector should have no part to play in the AI revolution in medicine, but I believe they should do it in an openly manner that allows all stakeholders, from patient to provider to have complete oversight on all the processes.

Trust me, I know the perils of duplicating data and the challenges of keeping it up to date. Before branching out to medicine, I used to be a computer engineer. Our goal is not to replicate StatPearls in French, but to lay the foundation of a platform we hope can better serve the clinicians and health care systems of tomorrow. Overtime, we plan to supplement StatPearls content with EBM data, semantic data and links.

That being said, here is a link to an article we've translated and imported on Wikimedica. All citations have been replicated and we have added at the bottom a mention that it comes from StatPearls and the required metadata for us to do and check if the StatPearls source has been updated.

Antoine Mercier-Linteau (discussioncontributions)
Michaël St-Gelais (discussioncontributions)

Parfait ! Je rajouterais juste dans le dernier paragraphe une question ouverte : est-ce que cette manière de vous citer vous convient ?

Antoine Mercier-Linteau (discussioncontributions)

Voici le courriel envoyé:

This email is in reference to a phone conversation we had on the 9th of April.

First of all, let me reiterate how impressed I am with StatPearls. Like you, I believe medical knowledge ought to be free and open access and StatPearls shines as pioneer in this regard. Your quiz system is also something to behold and encouraging editors to contribute through CME credits is brilliant. We are at this moment pursuing that goal, but you are much further along.

However, I believe I was not able to properly convey what Wikimedica is over our phone conversation. While it seems we are simply trying to replicate your content in French, let me assure you it is no so. Please allow me to give you a quick overview of the project.

Wikimedica is a platform (run by a non-profit) that aims to improve how we create and use medical knowledge in three ways:

  1. By creating an open access point-of-care evidence based medical resource.
    • Think UpToDate, but open access.
  2. With a dynamic and horizontal peer-reviewing process.
    • We believe knowledge nowadays evolve too fast to be compatible with the standard peer-reviewing editorial process. On Wikimedica, users of the platform also have the possibility of reviewing and commenting knowledge and can verify modifications made to pages on the fly. Safety is guaranteed by a strong versioning system and the fact that editors on Wikimedica are bound by the deontological codes of their respective professions when interacting with the platform.
  3. By providing a knowledge structure that is both usable by humans and computers and with which Artificial Intelligence (AI) systems can interface.
    • AI is going to be an integral part of how healthcare is conducted in the near future and proof will be soon made that a clinician paired with an AI improves patient outcomes.
    • The private sector is already on it and building AI systems as we speak. Even more so that normal computer programs, AI are black boxes. These systems, if left entirely for-profit, will be fraught with conflicts of interests as we will have absolutely no way of telling how an AI came to recommend such treatment for such patient.
    • Moreover, feeding those AIs with patient data, which they will need to achieve their full potential, will put patients at a great risk of privacy breeches.
    • AIs are knowledge inference machines and if entirely the property of private third parties, they will further dispossess us from the science and know-how that underpins our healthcare systems. I am not saying that the private sector should have no part to play in the AI revolution in medicine, but I believe they should do it in an openly manner that allows all stakeholders, from patient to provider to have complete oversight on all the processes.

I am well aware of the perils of duplicating data and the challenges of keeping it up to date. Before branching out to medicine, I used to be a computer engineer. Our goal is not to replicate StatPearls in French, but to lay the foundation of a platform we hope can better serve the clinicians and health care systems of tomorrow. Overtime, we will supplement StatPearls content with guidelines, EBM data (strength of recommendation, strength of evidence, specificity, sensitivity, NNT, NNH, etc), semantic data and links.

Here is a link to an article (Shock) we've translated and imported on Wikimedica. All citations have been replicated and we have added at the bottom a mention that it comes from StatPearls and the required metadata for us to check back if the StatPearls source has been updated:

Cette page a été modifiée ou créée le 11/04/2020 à partir de Shock (StatPearls / Shock (28/01/2020)), écrite par les contributeurs de StatPearls et partagée sous la licence CC-BY 4.0 international. Le contenu original est disponible à https://www.ncbi.nlm.nih.gov/pubmed/30285387 (livre).

This page was modified or created the 11th of April 2020 from Shock (Link to statpearls.com / link to the pubmed page for the article (page version) ) written by the contributors of StatPearls and share under the CC-BY 4.0 international license. The original content is available at https://www.ncbi.nlm.nih.gov/pubmed/30285387 (link to the book on pubmed)

Also, the authors of the source page are cited every time an excerpt is copied onto Wikimedica:

Hayas Haseer Koya et Manju Paul, Shock, StatPearls Publishing, 2020 (PMID 30285387, lire en ligne)

You can also get a glimpse of a content page as seen by a computer program by following this link.

Do you think proper attribution is given?

If also you would like to further discuss how we can join forces in making medical knowledge open access and AI ready, please reach out. We would be happy to collaborate.

Best regards.

Antoine Mercier-Linteau (discussioncontributions)

Aucune réponse reçue de la part de StatPearls... Nous avons fait notre bout de chemin. Je rejette la tâche.

Michaël St-Gelais (discussioncontributions)

Bien d'accord.

Il n’y a aucun sujet plus ancien