{"id":202,"date":"2025-08-21T09:04:50","date_gmt":"2025-08-21T07:04:50","guid":{"rendered":"https:\/\/dali.science\/?p=202"},"modified":"2025-08-21T12:44:19","modified_gmt":"2025-08-21T10:44:19","slug":"data-science-sante-machine-learning","status":"publish","type":"post","link":"https:\/\/dali.science\/index.php\/2025\/08\/21\/data-science-sante-machine-learning\/","title":{"rendered":"Faut-il (vraiment) faire du machine learning sur les donn\u00e9es de sant\u00e9 ?"},"content":{"rendered":"\n<p>Travaillant dans le secteur de la sant\u00e9, nous nous sommes demand\u00e9 s\u2019il est devenu trop courant d&rsquo;entendre : <em>\u00ab On pourrait faire du machine learning sur les donn\u00e9es de sant\u00e9\u2026 \u00bb<\/em>. La proposition semble prometteuse, avec un c\u00f4t\u00e9 moderne, voire incontournable <em>(et c&rsquo;est sans parler d&rsquo;IA g\u00e9n\u00e9ratives qui attirent beaucoup d&rsquo;attention, on parle ici du machine learning \u00ab\u00a0traditionnel\u00a0\u00bb)<\/em>. Et pourtant\u2026 trop souvent, ces projets sont lanc\u00e9s sans r\u00e9el besoin d\u00e9fini, ni question pr\u00e9cise \u00e0 r\u00e9soudre. Comme si le simple fait d\u2019utiliser un mod\u00e8le complexe \u00e9tait en soi un gage de valeur ajout\u00e9e.<\/p>\n\n\n\n<p>Nous avons appris au fil des projets que <strong>le machine learning n\u2019a de sens que s\u2019il est ancr\u00e9 dans un besoin de terrain, dans un besoin bien formul\u00e9.<\/strong> Le risque est que le r\u00e9sultat soit d\u00e9ceptif pour tout le monde, et que personne ne s\u2019empare des r\u00e9sultats.<\/p>\n\n\n\n<p>C\u2019est ce que nous souhaitons partager dans cet article. Nous allons donc poser une question simple :<\/p>\n\n\n\n<p class=\"has-text-align-center\"><strong><em>\u00c0 quoi \u00e7a sert vraiment de faire du machine learning en sant\u00e9 ?<\/em><\/strong><\/p>\n\n\n\n<p>Nous verrons pourquoi l\u2019engouement est r\u00e9el (et souvent justifi\u00e9), mais aussi pourquoi il faut rester lucide. Et surtout, comment construire des projets qui visent rigueur scientifique et finalit\u00e9 concr\u00e8te.<\/p>\n\n\n\n<p class=\"has-background\" style=\"background:linear-gradient(90deg,rgba(7,146,227,0.26) 0%,rgba(155,81,224,0.29) 100%)\"><strong>\ud83d\udd0e Petit rappel : qu&rsquo;est-ce qu&rsquo;un un mod\u00e8le de machine learning ?<\/strong><br>Un mod\u00e8le de machine learning (= apprentissage automatique), c\u2019est un programme qui apprend \u00e0 rep\u00e9rer des r\u00e9gularit\u00e9s dans des donn\u00e9es pass\u00e9es pour pr\u00e9dire, classer ou segmenter de nouveaux cas. C\u2019est utile\u2026 si ces r\u00e9gularit\u00e9s existent vraiment, et si elles aident \u00e0 prendre des d\u00e9cisions.<br><br>Mais attention, le mod\u00e8le ne comprend pas ce qu\u2019il fait ! Il ne fait que rep\u00e9rer des r\u00e9gularit\u00e9s math\u00e9matiques dans les donn\u00e9es. S\u2019il a \u00e9t\u00e9 mal aliment\u00e9, ou si les donn\u00e9es sont biais\u00e9es, les r\u00e9sultats seront trompeurs.<br><br>Trois e<em>xemples<\/em> :<br><em>  &#8211; Pr\u00e9dire un \u00e9v\u00e9nement (classification ou r\u00e9gression) : va-t-il y avoir un risque de chute ?<br>  &#8211; Regrouper automatiquement des cas similaires (clustering) : y a-t-il des profils types de parcours post-op\u00e9ratoire ?<br>  &#8211; D\u00e9tecter des comportements inhabituels (anomaly detection) : ce signal de capteur sort-il de l\u2019ordinaire ?<\/em><\/p>\n\n\n\n<h2 class=\"wp-block-heading\"><strong>D\u00e9cryptage \u2013 Le boom du machine learning en sant\u00e9 (2010\u20132025)<\/strong><\/h2>\n\n\n\n<p>Pour commencer, jetons un petit regard en arri\u00e8re pour voir ce que nous dit l&rsquo;\u00e9volution du machine learning en sant\u00e9 du point de vue des publications. Le constat est clair : l\u2019int\u00e9r\u00eat scientifique pour le machine learning, tous sujets confondus en sant\u00e9, conna\u00eet une <strong>croissance spectaculaire<\/strong>, confirm\u00e9e par plusieurs revues de litt\u00e9rature r\u00e9centes.<\/p>\n\n\n\n<h3 class=\"wp-block-heading\"><strong>Quelques chiffres cl\u00e9s :<\/strong><\/h3>\n\n\n\n<ul class=\"wp-block-list\">\n<li>Une simple recherche sur le site PubMed (<a href=\"https:\/\/pubmed.ncbi.nlm.nih.gov\/\">https:\/\/pubmed.ncbi.nlm.nih.gov\/<\/a>), \u201cMachine Learning + Healthcare\u201d OR \u201cArtificial intelligence + Healthcare\u201d, donne <strong>35 105 r\u00e9ponses<\/strong> (au 1er ao\u00fbt 2025).<br><\/li>\n<\/ul>\n\n\n\n<figure data-wp-context=\"{&quot;imageId&quot;:&quot;69e2259d3513b&quot;}\" data-wp-interactive=\"core\/image\" data-wp-key=\"69e2259d3513b\" class=\"wp-block-image is-style-default wp-lightbox-container\"><img decoding=\"async\" data-wp-class--hide=\"state.isContentHidden\" data-wp-class--show=\"state.isContentVisible\" data-wp-init=\"callbacks.setButtonStyles\" data-wp-on--click=\"actions.showLightbox\" data-wp-on--load=\"callbacks.setButtonStyles\" data-wp-on-window--resize=\"callbacks.setButtonStyles\" src=\"https:\/\/lh7-rt.googleusercontent.com\/docsz\/AD_4nXdyhWX-NrD4uNq0tiktJLcRhp0IlosrE1-D_bDEUJyVsa0xTDsQPI8GoiMTVGE6BTnnxIXMfjcpcxSI3yCCqcJQw5mioYEWoD92IXZJTNjx0brx97cPiX0mpaXOhJoqb-QMjcaxew?key=Ylxwb57Ne-i0ybw5GAX8ug\" alt=\"\"\/><button\n\t\t\tclass=\"lightbox-trigger\"\n\t\t\ttype=\"button\"\n\t\t\taria-haspopup=\"dialog\"\n\t\t\taria-label=\"Agrandir\"\n\t\t\tdata-wp-init=\"callbacks.initTriggerButton\"\n\t\t\tdata-wp-on--click=\"actions.showLightbox\"\n\t\t\tdata-wp-style--right=\"state.imageButtonRight\"\n\t\t\tdata-wp-style--top=\"state.imageButtonTop\"\n\t\t>\n\t\t\t<svg xmlns=\"http:\/\/www.w3.org\/2000\/svg\" width=\"12\" height=\"12\" fill=\"none\" viewBox=\"0 0 12 12\">\n\t\t\t\t<path fill=\"#fff\" d=\"M2 0a2 2 0 0 0-2 2v2h1.5V2a.5.5 0 0 1 .5-.5h2V0H2Zm2 10.5H2a.5.5 0 0 1-.5-.5V8H0v2a2 2 0 0 0 2 2h2v-1.5ZM8 12v-1.5h2a.5.5 0 0 0 .5-.5V8H12v2a2 2 0 0 1-2 2H8Zm2-12a2 2 0 0 1 2 2v2h-1.5V2a.5.5 0 0 0-.5-.5H8V0h2Z\" \/>\n\t\t\t<\/svg>\n\t\t<\/button><figcaption class=\"wp-element-caption\">Graphique cr\u00e9\u00e9 par l&rsquo;auteur<\/figcaption><\/figure>\n\n\n\n<ul class=\"wp-block-list\">\n<li>Une \u00e9tude publi\u00e9e dans <em>Frontiers in Medicine<\/em> a analys\u00e9 <strong>22 \u202f950 articles publi\u00e9s entre 1993 et 2023<\/strong>. Elle montre une acc\u00e9l\u00e9ration tr\u00e8s rapide apr\u00e8s 2010, avec un pic marqu\u00e9 \u00e0 partir de 2019 (<mark style=\"background-color:rgba(0, 0, 0, 0)\" class=\"has-inline-color has-vivid-cyan-blue-color\">Y. Xie et al. \u00ab Evolution of artificial intelligence in healthcare: a 30-year bibliometric study \u00bb, <em>Front. Med.<\/em>, vol. 11, janv. 2025, doi:<a href=\"https:\/\/doi.org\/10.3389\/fmed.2024.1505692\"> 10.3389\/fmed.2024.1505692<\/a><\/mark><strong>)<\/strong><br><\/li>\n\n\n\n<li>Une autre \u00e9tude recense toutes les publications contenant les mots-cl\u00e9s \u201cmachine learning\u201d et \u201chealthcare\u201d dans <strong>Scopus<\/strong> de 2000 \u00e0 2024.&nbsp; Le volume passe de quelques dizaines \u00e0 <strong>plus de 3 000 articles par an<\/strong> (<mark style=\"background-color:rgba(0, 0, 0, 0)\" class=\"has-inline-color has-vivid-cyan-blue-color\">A. Dalky et al., \u00ab Global Research Trends, Hotspots, Impacts, and Emergence of Artificial Intelligence and Machine Learning in Health and Medicine: A 25-Year Bibliometric Analysis \u00bb, Healthcare, vol. 13, n\u1d52 8, Art. n\u1d52 8, janv. 2025, doi: 10.3390\/healthcare13080892<\/mark>).<\/li>\n<\/ul>\n\n\n\n<h3 class=\"wp-block-heading\"><strong>Interpr\u00e9tation de cette explosion<\/strong><\/h3>\n\n\n\n<p>Nous proposons de r\u00e9sumer ces chiffres en 3 phases :<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>La phase<strong> exploratoire<\/strong> (jusqu\u2019en 2012) : le machine learning est une curiosit\u00e9 technique.<\/li>\n\n\n\n<li>La phase <strong>d\u2019adoption<\/strong> (2015\u20132019) : multiplication par 10 \u00e0 20 des publications annuelles.<\/li>\n\n\n\n<li>La phase<strong> d\u2019emballement<\/strong> (post-2020) : plusieurs milliers d\u2019articles par an, dans toutes les sp\u00e9cialit\u00e9s m\u00e9dicales (et pas que l\u2019imagerie !).<\/li>\n<\/ul>\n\n\n\n<p>Mais ce volume impressionnant pose question. Une grande partie de ces publications sont des d\u00e9monstrations techniques ou m\u00e9thodologiques. Elles ne d\u00e9bouchent pas toujours sur des usages cliniques ou organisationnels.<\/p>\n\n\n\n<p>Pourquoi cela pose probl\u00e8me ?<br>&gt; Parce que mod\u00e9liser n\u2019a de sens que si cela aide \u00e0 d\u00e9cider, \u00e0 orienter, \u00e0 prioriser.<br>&gt; Parce qu\u2019un bon mod\u00e8le ne vaut que s\u2019il est compr\u00e9hensible, interpr\u00e9table et utilis\u00e9.<\/p>\n\n\n\n<h2 class=\"wp-block-heading\"><strong>Quand le machine learning est utile : retours d\u2019exp\u00e9rience<\/strong><\/h2>\n\n\n\n<p>Il serait injuste de critiquer le machine learning sans reconna\u00eetre ses apports concrets \u2014 \u00e0 condition qu\u2019il soit bien utilis\u00e9. Dans plusieurs projets, nous avons vu des approches de classification, de clustering ou de r\u00e9gression produire des r\u00e9sultats r\u00e9ellement utiles pour l\u2019action. D\u2019autant que la barri\u00e8re \u00e0 l\u2019entr\u00e9e pour utiliser ces m\u00e9thodes est devenue tr\u00e8s basse. \u00c7a n\u2019est plus l&rsquo;apanage des seuls experts techniques. Voici 4 retours d&rsquo;exp\u00e9riences qui d&rsquo;usages r\u00e9ussis.<\/p>\n\n\n\n<div class=\"wp-block-group is-nowrap is-layout-flex wp-container-core-group-is-layout-ad2f72ca wp-block-group-is-layout-flex\">\n<p class=\"has-text-align-left\"><strong>REX 1. Pr\u00e9vention cibl\u00e9e dans les territoires<\/strong><br>Dans un programme r\u00e9gional de pr\u00e9vention des maladies cardiovasculaires, les crit\u00e8res de ciblage classiques (\u00e2ge, ant\u00e9c\u00e9dents m\u00e9dicaux) laissaient de c\u00f4t\u00e9 certains profils \u00e0 risque. Un mod\u00e8le de classification bas\u00e9 sur des donn\u00e9es m\u00e9dico-sociales et de consommation de soins a permis d\u2019identifier des patients \u00e0 haut risque non rep\u00e9r\u00e9s par les outils traditionnels. Gr\u00e2ce \u00e0 cela, les interventions de pr\u00e9vention (appels infirmiers, entretiens de motivation, courrier d\u2019information) sont mieux cibl\u00e9es, tout en restant \u00e9quitables. Ici, le mod\u00e8le ne remplace pas l\u2019expertise humaine : il l\u2019oriente.<\/p>\n<\/div>\n\n\n<div class=\"wp-block-image\">\n<figure class=\"aligncenter size-full is-resized\"><img loading=\"lazy\" decoding=\"async\" width=\"726\" height=\"726\" src=\"https:\/\/dali.science\/wp-content\/uploads\/2025\/08\/illustration-REX1.png\" alt=\"\" class=\"wp-image-215\" style=\"width:196px;height:auto\" srcset=\"https:\/\/dali.science\/wp-content\/uploads\/2025\/08\/illustration-REX1.png 726w, https:\/\/dali.science\/wp-content\/uploads\/2025\/08\/illustration-REX1-300x300.png 300w, https:\/\/dali.science\/wp-content\/uploads\/2025\/08\/illustration-REX1-150x150.png 150w, https:\/\/dali.science\/wp-content\/uploads\/2025\/08\/illustration-REX1-100x100.png 100w\" sizes=\"auto, (max-width: 726px) 100vw, 726px\" \/><figcaption class=\"wp-element-caption\"><sub>G\u00e9n\u00e9r\u00e9 par IA par l&rsquo;auteur<\/sub><\/figcaption><\/figure>\n<\/div>\n\n\n<p><\/p>\n\n\n\n<div class=\"wp-block-group is-nowrap is-layout-flex wp-container-core-group-is-layout-ad2f72ca wp-block-group-is-layout-flex\">\n<p><strong>REX 2. Optimisation du suivi post-op\u00e9ratoire<\/strong><br>Dans un centre hospitalier, un projet a analys\u00e9 les parcours de patients op\u00e9r\u00e9s pour des chirurgies digestives. En combinant des donn\u00e9es cliniques avec du process mining, l\u2019\u00e9quipe a utilis\u00e9 un algorithme de clustering pour identifier des groupes de patients avec des trajectoires post-op\u00e9ratoires atypiques. L\u2019un des groupes pr\u00e9sentait un taux de r\u00e9hospitalisation \u00e9lev\u00e9, li\u00e9 \u00e0 un d\u00e9faut de contact infirmier \u00e0 J+3. Ce signal, r\u00e9v\u00e9l\u00e9 par le mod\u00e8le, a conduit \u00e0 une refonte du protocole de suivi. Une fois de plus, l\u2019algorithme n\u2019\u00e9tait qu\u2019un r\u00e9v\u00e9lateur\u202f: la vraie d\u00e9cision s\u2019est prise ensuite.<\/p>\n<\/div>\n\n\n<div class=\"wp-block-image\">\n<figure class=\"aligncenter size-full is-resized\"><img loading=\"lazy\" decoding=\"async\" width=\"651\" height=\"648\" src=\"https:\/\/dali.science\/wp-content\/uploads\/2025\/08\/illustration-REX2.png\" alt=\"\" class=\"wp-image-214\" style=\"width:197px;height:auto\" srcset=\"https:\/\/dali.science\/wp-content\/uploads\/2025\/08\/illustration-REX2.png 651w, https:\/\/dali.science\/wp-content\/uploads\/2025\/08\/illustration-REX2-300x300.png 300w, https:\/\/dali.science\/wp-content\/uploads\/2025\/08\/illustration-REX2-150x150.png 150w, https:\/\/dali.science\/wp-content\/uploads\/2025\/08\/illustration-REX2-100x100.png 100w\" sizes=\"auto, (max-width: 651px) 100vw, 651px\" \/><figcaption class=\"wp-element-caption\"><sub>G\u00e9n\u00e9r\u00e9 par IA par l&rsquo;auteur<\/sub><\/figcaption><\/figure>\n<\/div>\n\n\n<p><\/p>\n\n\n\n<div class=\"wp-block-group is-nowrap is-layout-flex wp-container-core-group-is-layout-ad2f72ca wp-block-group-is-layout-flex\">\n<p><strong>REX 3. Dispositifs m\u00e9dicaux : d\u00e9tection pr\u00e9coce d\u2019usure ou d\u2019incidents<br><\/strong>Un fabricant de dispositifs m\u00e9dicaux connect\u00e9s a d\u00e9ploy\u00e9 un algorithme de d\u00e9tection d\u2019anomalies sur les signaux capt\u00e9s par ses capteurs. L\u2019objectif\u202f: d\u00e9tecter des signes pr\u00e9coces de d\u00e9faillance (mat\u00e9rielle ou physiologique) avant que l\u2019\u00e9v\u00e9nement ne survienne. Une fois le syst\u00e8me en place, plusieurs cas ont \u00e9t\u00e9 \u00e9vit\u00e9s. L\u2019impact r\u00e9el ? Moins d\u2019hospitalisations \u00e9vitables, un SAV plus r\u00e9actif, et surtout une vigilance renforc\u00e9e dans l\u2019usage du dispositif.<\/p>\n<\/div>\n\n\n<div class=\"wp-block-image\">\n<figure class=\"aligncenter size-full is-resized\"><img loading=\"lazy\" decoding=\"async\" width=\"794\" height=\"726\" src=\"https:\/\/dali.science\/wp-content\/uploads\/2025\/08\/illustration-REX3.png\" alt=\"\" class=\"wp-image-217\" style=\"width:193px;height:auto\" srcset=\"https:\/\/dali.science\/wp-content\/uploads\/2025\/08\/illustration-REX3.png 794w, https:\/\/dali.science\/wp-content\/uploads\/2025\/08\/illustration-REX3-300x274.png 300w, https:\/\/dali.science\/wp-content\/uploads\/2025\/08\/illustration-REX3-768x702.png 768w\" sizes=\"auto, (max-width: 767px) 89vw, (max-width: 1000px) 54vw, (max-width: 1071px) 543px, 580px\" \/><figcaption class=\"wp-element-caption\"><sub>G\u00e9n\u00e9r\u00e9 par IA par l&rsquo;auteur<\/sub><\/figcaption><\/figure>\n<\/div>\n\n\n<p><\/p>\n\n\n\n<div class=\"wp-block-group is-nowrap is-layout-flex wp-container-core-group-is-layout-ad2f72ca wp-block-group-is-layout-flex\">\n<p><strong>REX 4. Essais cliniques : s\u00e9lection plus fine des participants<br><\/strong>Dans un projet de recherche clinique sur une th\u00e9rapie innovante, les chercheurs soup\u00e7onnaient que certains sous-groupes de patients r\u00e9agissaient diff\u00e9remment au traitement. En amont du protocole, une analyse exploratoire via clustering non supervis\u00e9 a permis d\u2019identifier des profils diff\u00e9renci\u00e9s, bas\u00e9s sur des marqueurs biologiques et des scores fonctionnels. Ces profils ont ensuite \u00e9t\u00e9 utilis\u00e9s pour affiner les crit\u00e8res d\u2019inclusion, dans le cadre d\u2019une sous-\u00e9tude. R\u00e9sultat : une meilleure puissance statistique, une interpr\u00e9tation enrichie des r\u00e9sultats, et une hypoth\u00e8se plus cibl\u00e9e pour une future phase III.<\/p>\n<\/div>\n\n\n<div class=\"wp-block-image\">\n<figure class=\"aligncenter size-full is-resized\"><img loading=\"lazy\" decoding=\"async\" width=\"793\" height=\"793\" src=\"https:\/\/dali.science\/wp-content\/uploads\/2025\/08\/illustration-REX4__.jpg\" alt=\"\" class=\"wp-image-224\" style=\"width:169px;height:auto\" srcset=\"https:\/\/dali.science\/wp-content\/uploads\/2025\/08\/illustration-REX4__.jpg 793w, https:\/\/dali.science\/wp-content\/uploads\/2025\/08\/illustration-REX4__-300x300.jpg 300w, https:\/\/dali.science\/wp-content\/uploads\/2025\/08\/illustration-REX4__-150x150.jpg 150w, https:\/\/dali.science\/wp-content\/uploads\/2025\/08\/illustration-REX4__-768x768.jpg 768w, https:\/\/dali.science\/wp-content\/uploads\/2025\/08\/illustration-REX4__-100x100.jpg 100w\" sizes=\"auto, (max-width: 767px) 89vw, (max-width: 1000px) 54vw, (max-width: 1071px) 543px, 580px\" \/><figcaption class=\"wp-element-caption\"><sub>G\u00e9n\u00e9r\u00e9 par IA par l&rsquo;auteur<\/sub><\/figcaption><\/figure>\n<\/div>\n\n\n<p><\/p>\n\n\n\n<h3 class=\"wp-block-heading\"><strong>Une constante : ce n\u2019est pas l\u2019algorithme qui agit, c\u2019est l\u2019\u00e9cosyst\u00e8me autour<\/strong><\/h3>\n\n\n\n<p>Ces exemples ne font pas appel \u00e0 la plus haute complexit\u00e9 technique qui soit. Parfois, il s\u2019agissait simplement d\u2019un arbre de d\u00e9cision, d\u2019un random forest ou d\u2019un clustering hi\u00e9rarchique. Ce qui fait la diff\u00e9rence, c\u2019est que :<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>les mod\u00e8les ont \u00e9t\u00e9 construits en partant d\u2019un besoin m\u00e9tier concret,<\/li>\n\n\n\n<li>les r\u00e9sultats ont \u00e9t\u00e9 int\u00e9gr\u00e9s dans un processus de d\u00e9cision ou d\u2019organisation,<\/li>\n\n\n\n<li>et surtout, les utilisateurs finaux ont \u00e9t\u00e9 impliqu\u00e9s d\u00e8s le d\u00e9but.<\/li>\n<\/ul>\n\n\n\n<h2 class=\"wp-block-heading\"><strong>Quel point de vue adopter sur le sujet ?<\/strong><\/h2>\n\n\n\n<p>Pour nous, faire du machine learning, ce n\u2019est pas une posture technologique. Ce doit \u00eatre une d\u00e9marche rigoureuse, construite avec des experts du domaine concern\u00e9.<\/p>\n\n\n\n<p>Les questions cl\u00e9s \u00e0 se poser avant de lancer un projet :<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>Quelle d\u00e9cision le mod\u00e8le est-il cens\u00e9 \u00e9clairer ?<\/li>\n\n\n\n<li>Quels indicateurs seraient vraiment utiles \u00e0 produire ?<\/li>\n\n\n\n<li>Qui est le public cible du mod\u00e8le ? Est-il form\u00e9 pour l\u2019utiliser ?<\/li>\n\n\n\n<li>A-t-on besoin d\u2019un mod\u00e8le statistique complexe, ou d\u2019un tableau crois\u00e9 bien construit ?<\/li>\n<\/ul>\n\n\n\n<h2 class=\"wp-block-heading\"><strong>Trois erreurs fr\u00e9quentes \u00e0 \u00e9viter<\/strong><\/h2>\n\n\n\n<p><strong>\u274cFaire du machine learning pour faire \u201cmoderne\u201d <br><\/strong>Une IA qui pr\u00e9dit tout, sans qu\u2019on sache \u00e0 quoi \u00e7a sert\u2026 n\u2019a aucun impact. Le besoin doit pr\u00e9c\u00e9der l\u2019outil.<\/p>\n\n\n\n<p><strong>\u274cChoisir la m\u00e9thode avant de d\u00e9finir le probl\u00e8me<br><\/strong>Ce n\u2019est pas parce qu\u2019un mod\u00e8le XGBoost ou un r\u00e9seau de neurones marche ailleurs qu\u2019il est adapt\u00e9 ici. <em>(\u201c\u00a0\u00bbQuand on n\u2019a qu\u2019un marteau, tout finit par ressembler \u00e0 un clou.\u00a0\u00bb A. Maslow<\/em>)<br><br><strong>\u274cOublier l\u2019appropriation<br><\/strong>Un mod\u00e8le performant mais incompris ne sera jamais utilis\u00e9. La p\u00e9dagogie et la transparence sont des leviers cl\u00e9s. #Explicabilit\u00e9-de-l\u2019IA<\/p>\n\n\n\n<h2 class=\"wp-block-heading\"><strong>En conclusion<\/strong><\/h2>\n\n\n\n<p>Apr\u00e8s plus d\u2019une d\u00e9cennie de travail sur des sujets de data science en sant\u00e9, je reste convaincu du potentiel du machine learning \u2014 mais pas au mythe de la bo\u00eete noire magique. Nous croyons \u00e0 la <strong>co-construction<\/strong>, \u00e0 l\u2019<strong>utilit\u00e9 terrain<\/strong>, et \u00e0 une science des donn\u00e9es <strong>au service des d\u00e9cisions<\/strong>.<\/p>\n\n\n\n<blockquote class=\"wp-block-quote is-layout-flow wp-block-quote-is-layout-flow\">\n<p>\u201cUn bon projet IA ne commence pas par un algorithme, il commence par une bonne question.\u201d<\/p>\n\n\n\n<p>Retrouvez nos autres articles de blog<br><a href=\"https:\/\/dali.science\/index.php\/2025\/08\/01\/la-simulation-de-flux-au-service-des-soignants\/\">\ud83d\udc49 La simulation de flux au service des soignants<br><\/a><a href=\"https:\/\/dali.science\/index.php\/2025\/07\/15\/modeliser-un-parcours-de-soin-cest-plus-que-dessiner-un-diagramme-de-flux\/\">\ud83d\udc49Mod\u00e9liser un parcours de soin, c\u2019est plus que dessiner un diagramme de flux<br><\/a><a href=\"https:\/\/dali.science\/index.php\/2025\/07\/01\/ameliorer-organisation-hopital-modeles-mathematiques\/\">\ud83d\udc49Am\u00e9liorer l\u2019organisation des h\u00f4pitaux\u2026 avec des mod\u00e8les math\u00e9matiques<br><\/a><\/p>\n<\/blockquote>\n","protected":false},"excerpt":{"rendered":"<p>\u00ab On pourrait faire du machine learning sur les donn\u00e9es de sant\u00e9\u2026 \u00bb. La proposition semble prometteuse, avec un c\u00f4t\u00e9 moderne, voire incontournable &#8230;<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[12],"tags":[14,15,13,16],"class_list":["post-202","post","type-post","status-publish","format-standard","hentry","category-machine-learning-ia","tag-data-science","tag-donnees-de-sante","tag-machine-learning","tag-retour-dexperience"],"yoast_head":"<!-- This site is optimized with the Yoast SEO plugin v27.4 - https:\/\/yoast.com\/product\/yoast-seo-wordpress\/ -->\n<title>Faut-il (vraiment) faire du machine learning sur les donn\u00e9es de sant\u00e9 ?<\/title>\n<meta name=\"description\" content=\"Machine learning et intelligence artificielle en sant\u00e9 : entre engouement et lucidit\u00e9. Cas d\u2019usage concrets, erreurs fr\u00e9quentes, et conseils pour des projets utiles et partag\u00e9s. faire du machine learning sur donn\u00e9es de sant\u00e9 &amp; boom du machine learning &amp; IA g\u00e9n\u00e9rative\" \/>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/dali.science\/index.php\/2025\/08\/21\/data-science-sante-machine-learning\/\" \/>\n<meta property=\"og:locale\" content=\"fr_FR\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"Faut-il (vraiment) faire du machine learning sur les donn\u00e9es de sant\u00e9 ?\" \/>\n<meta property=\"og:description\" content=\"Machine learning et intelligence artificielle en sant\u00e9 : entre engouement et lucidit\u00e9. Cas d\u2019usage concrets, erreurs fr\u00e9quentes, et conseils pour des projets utiles et partag\u00e9s. faire du machine learning sur donn\u00e9es de sant\u00e9 &amp; boom du machine learning &amp; IA g\u00e9n\u00e9rative\" \/>\n<meta property=\"og:url\" content=\"https:\/\/dali.science\/index.php\/2025\/08\/21\/data-science-sante-machine-learning\/\" \/>\n<meta property=\"article:published_time\" content=\"2025-08-21T07:04:50+00:00\" \/>\n<meta property=\"article:modified_time\" content=\"2025-08-21T10:44:19+00:00\" \/>\n<meta property=\"og:image\" content=\"https:\/\/lh7-rt.googleusercontent.com\/docsz\/AD_4nXdyhWX-NrD4uNq0tiktJLcRhp0IlosrE1-D_bDEUJyVsa0xTDsQPI8GoiMTVGE6BTnnxIXMfjcpcxSI3yCCqcJQw5mioYEWoD92IXZJTNjx0brx97cPiX0mpaXOhJoqb-QMjcaxew?key=Ylxwb57Ne-i0ybw5GAX8ug\" \/>\n<meta name=\"author\" content=\"Martin PRODEL\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:label1\" content=\"\u00c9crit par\" \/>\n\t<meta name=\"twitter:data1\" content=\"Martin PRODEL\" \/>\n\t<meta name=\"twitter:label2\" content=\"Dur\u00e9e de lecture estim\u00e9e\" \/>\n\t<meta name=\"twitter:data2\" content=\"9 minutes\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\\\/\\\/schema.org\",\"@graph\":[{\"@type\":\"Article\",\"@id\":\"https:\\\/\\\/dali.science\\\/index.php\\\/2025\\\/08\\\/21\\\/data-science-sante-machine-learning\\\/#article\",\"isPartOf\":{\"@id\":\"https:\\\/\\\/dali.science\\\/index.php\\\/2025\\\/08\\\/21\\\/data-science-sante-machine-learning\\\/\"},\"author\":{\"name\":\"Martin PRODEL\",\"@id\":\"https:\\\/\\\/dali.science\\\/#\\\/schema\\\/person\\\/370a84159837b710cf2d94f62c32cc9f\"},\"headline\":\"Faut-il (vraiment) faire du machine learning sur les donn\u00e9es de sant\u00e9 ?\",\"datePublished\":\"2025-08-21T07:04:50+00:00\",\"dateModified\":\"2025-08-21T10:44:19+00:00\",\"mainEntityOfPage\":{\"@id\":\"https:\\\/\\\/dali.science\\\/index.php\\\/2025\\\/08\\\/21\\\/data-science-sante-machine-learning\\\/\"},\"wordCount\":1722,\"commentCount\":0,\"publisher\":{\"@id\":\"https:\\\/\\\/dali.science\\\/#organization\"},\"image\":{\"@id\":\"https:\\\/\\\/dali.science\\\/index.php\\\/2025\\\/08\\\/21\\\/data-science-sante-machine-learning\\\/#primaryimage\"},\"thumbnailUrl\":\"https:\\\/\\\/lh7-rt.googleusercontent.com\\\/docsz\\\/AD_4nXdyhWX-NrD4uNq0tiktJLcRhp0IlosrE1-D_bDEUJyVsa0xTDsQPI8GoiMTVGE6BTnnxIXMfjcpcxSI3yCCqcJQw5mioYEWoD92IXZJTNjx0brx97cPiX0mpaXOhJoqb-QMjcaxew?key=Ylxwb57Ne-i0ybw5GAX8ug\",\"keywords\":[\"data science\",\"donn\u00e9es de sant\u00e9\",\"machine learning\",\"retour d'exp\u00e9rience\"],\"articleSection\":[\"machine learning \\\/ IA\"],\"inLanguage\":\"fr-FR\",\"potentialAction\":[{\"@type\":\"CommentAction\",\"name\":\"Comment\",\"target\":[\"https:\\\/\\\/dali.science\\\/index.php\\\/2025\\\/08\\\/21\\\/data-science-sante-machine-learning\\\/#respond\"]}]},{\"@type\":\"WebPage\",\"@id\":\"https:\\\/\\\/dali.science\\\/index.php\\\/2025\\\/08\\\/21\\\/data-science-sante-machine-learning\\\/\",\"url\":\"https:\\\/\\\/dali.science\\\/index.php\\\/2025\\\/08\\\/21\\\/data-science-sante-machine-learning\\\/\",\"name\":\"Faut-il (vraiment) faire du machine learning sur les donn\u00e9es de sant\u00e9 ?\",\"isPartOf\":{\"@id\":\"https:\\\/\\\/dali.science\\\/#website\"},\"primaryImageOfPage\":{\"@id\":\"https:\\\/\\\/dali.science\\\/index.php\\\/2025\\\/08\\\/21\\\/data-science-sante-machine-learning\\\/#primaryimage\"},\"image\":{\"@id\":\"https:\\\/\\\/dali.science\\\/index.php\\\/2025\\\/08\\\/21\\\/data-science-sante-machine-learning\\\/#primaryimage\"},\"thumbnailUrl\":\"https:\\\/\\\/lh7-rt.googleusercontent.com\\\/docsz\\\/AD_4nXdyhWX-NrD4uNq0tiktJLcRhp0IlosrE1-D_bDEUJyVsa0xTDsQPI8GoiMTVGE6BTnnxIXMfjcpcxSI3yCCqcJQw5mioYEWoD92IXZJTNjx0brx97cPiX0mpaXOhJoqb-QMjcaxew?key=Ylxwb57Ne-i0ybw5GAX8ug\",\"datePublished\":\"2025-08-21T07:04:50+00:00\",\"dateModified\":\"2025-08-21T10:44:19+00:00\",\"description\":\"Machine learning et intelligence artificielle en sant\u00e9 : entre engouement et lucidit\u00e9. Cas d\u2019usage concrets, erreurs fr\u00e9quentes, et conseils pour des projets utiles et partag\u00e9s. faire du machine learning sur donn\u00e9es de sant\u00e9 & boom du machine learning & IA g\u00e9n\u00e9rative\",\"breadcrumb\":{\"@id\":\"https:\\\/\\\/dali.science\\\/index.php\\\/2025\\\/08\\\/21\\\/data-science-sante-machine-learning\\\/#breadcrumb\"},\"inLanguage\":\"fr-FR\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\\\/\\\/dali.science\\\/index.php\\\/2025\\\/08\\\/21\\\/data-science-sante-machine-learning\\\/\"]}]},{\"@type\":\"ImageObject\",\"inLanguage\":\"fr-FR\",\"@id\":\"https:\\\/\\\/dali.science\\\/index.php\\\/2025\\\/08\\\/21\\\/data-science-sante-machine-learning\\\/#primaryimage\",\"url\":\"https:\\\/\\\/lh7-rt.googleusercontent.com\\\/docsz\\\/AD_4nXdyhWX-NrD4uNq0tiktJLcRhp0IlosrE1-D_bDEUJyVsa0xTDsQPI8GoiMTVGE6BTnnxIXMfjcpcxSI3yCCqcJQw5mioYEWoD92IXZJTNjx0brx97cPiX0mpaXOhJoqb-QMjcaxew?key=Ylxwb57Ne-i0ybw5GAX8ug\",\"contentUrl\":\"https:\\\/\\\/lh7-rt.googleusercontent.com\\\/docsz\\\/AD_4nXdyhWX-NrD4uNq0tiktJLcRhp0IlosrE1-D_bDEUJyVsa0xTDsQPI8GoiMTVGE6BTnnxIXMfjcpcxSI3yCCqcJQw5mioYEWoD92IXZJTNjx0brx97cPiX0mpaXOhJoqb-QMjcaxew?key=Ylxwb57Ne-i0ybw5GAX8ug\"},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\\\/\\\/dali.science\\\/index.php\\\/2025\\\/08\\\/21\\\/data-science-sante-machine-learning\\\/#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"Accueil\",\"item\":\"https:\\\/\\\/dali.science\\\/\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"Faut-il (vraiment) faire du machine learning sur les donn\u00e9es de sant\u00e9 ?\"}]},{\"@type\":\"WebSite\",\"@id\":\"https:\\\/\\\/dali.science\\\/#website\",\"url\":\"https:\\\/\\\/dali.science\\\/\",\"name\":\"dali.science\",\"description\":\"Transformer les donn\u00e9es en d\u00e9cisions \u00e9clair\u00e9es\",\"publisher\":{\"@id\":\"https:\\\/\\\/dali.science\\\/#organization\"},\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\\\/\\\/dali.science\\\/?s={search_term_string}\"},\"query-input\":{\"@type\":\"PropertyValueSpecification\",\"valueRequired\":true,\"valueName\":\"search_term_string\"}}],\"inLanguage\":\"fr-FR\"},{\"@type\":\"Organization\",\"@id\":\"https:\\\/\\\/dali.science\\\/#organization\",\"name\":\"DALI\",\"url\":\"https:\\\/\\\/dali.science\\\/\",\"logo\":{\"@type\":\"ImageObject\",\"inLanguage\":\"fr-FR\",\"@id\":\"https:\\\/\\\/dali.science\\\/#\\\/schema\\\/logo\\\/image\\\/\",\"url\":\"https:\\\/\\\/dali.science\\\/wp-content\\\/uploads\\\/2025\\\/05\\\/logo-avec-slogan.png\",\"contentUrl\":\"https:\\\/\\\/dali.science\\\/wp-content\\\/uploads\\\/2025\\\/05\\\/logo-avec-slogan.png\",\"width\":1299,\"height\":475,\"caption\":\"DALI\"},\"image\":{\"@id\":\"https:\\\/\\\/dali.science\\\/#\\\/schema\\\/logo\\\/image\\\/\"},\"sameAs\":[\"https:\\\/\\\/www.linkedin.com\\\/company\\\/dali-science\"]},{\"@type\":\"Person\",\"@id\":\"https:\\\/\\\/dali.science\\\/#\\\/schema\\\/person\\\/370a84159837b710cf2d94f62c32cc9f\",\"name\":\"Martin PRODEL\",\"image\":{\"@type\":\"ImageObject\",\"inLanguage\":\"fr-FR\",\"@id\":\"https:\\\/\\\/secure.gravatar.com\\\/avatar\\\/262a602c5fc065dd1dd94152b8a24b282b70ef33af194b5b89e93b757741d4a4?s=96&d=mm&r=g\",\"url\":\"https:\\\/\\\/secure.gravatar.com\\\/avatar\\\/262a602c5fc065dd1dd94152b8a24b282b70ef33af194b5b89e93b757741d4a4?s=96&d=mm&r=g\",\"contentUrl\":\"https:\\\/\\\/secure.gravatar.com\\\/avatar\\\/262a602c5fc065dd1dd94152b8a24b282b70ef33af194b5b89e93b757741d4a4?s=96&d=mm&r=g\",\"caption\":\"Martin PRODEL\"},\"sameAs\":[\"http:\\\/\\\/www.dali.science\"],\"url\":\"https:\\\/\\\/dali.science\\\/index.php\\\/author\\\/daliwordpress\\\/\"}]}<\/script>\n<!-- \/ Yoast SEO plugin. -->","yoast_head_json":{"title":"Faut-il (vraiment) faire du machine learning sur les donn\u00e9es de sant\u00e9 ?","description":"Machine learning et intelligence artificielle en sant\u00e9 : entre engouement et lucidit\u00e9. Cas d\u2019usage concrets, erreurs fr\u00e9quentes, et conseils pour des projets utiles et partag\u00e9s. faire du machine learning sur donn\u00e9es de sant\u00e9 & boom du machine learning & IA g\u00e9n\u00e9rative","robots":{"index":"index","follow":"follow","max-snippet":"max-snippet:-1","max-image-preview":"max-image-preview:large","max-video-preview":"max-video-preview:-1"},"canonical":"https:\/\/dali.science\/index.php\/2025\/08\/21\/data-science-sante-machine-learning\/","og_locale":"fr_FR","og_type":"article","og_title":"Faut-il (vraiment) faire du machine learning sur les donn\u00e9es de sant\u00e9 ?","og_description":"Machine learning et intelligence artificielle en sant\u00e9 : entre engouement et lucidit\u00e9. Cas d\u2019usage concrets, erreurs fr\u00e9quentes, et conseils pour des projets utiles et partag\u00e9s. faire du machine learning sur donn\u00e9es de sant\u00e9 & boom du machine learning & IA g\u00e9n\u00e9rative","og_url":"https:\/\/dali.science\/index.php\/2025\/08\/21\/data-science-sante-machine-learning\/","article_published_time":"2025-08-21T07:04:50+00:00","article_modified_time":"2025-08-21T10:44:19+00:00","og_image":[{"url":"https:\/\/lh7-rt.googleusercontent.com\/docsz\/AD_4nXdyhWX-NrD4uNq0tiktJLcRhp0IlosrE1-D_bDEUJyVsa0xTDsQPI8GoiMTVGE6BTnnxIXMfjcpcxSI3yCCqcJQw5mioYEWoD92IXZJTNjx0brx97cPiX0mpaXOhJoqb-QMjcaxew?key=Ylxwb57Ne-i0ybw5GAX8ug","type":"","width":"","height":""}],"author":"Martin PRODEL","twitter_card":"summary_large_image","twitter_misc":{"\u00c9crit par":"Martin PRODEL","Dur\u00e9e de lecture estim\u00e9e":"9 minutes"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"Article","@id":"https:\/\/dali.science\/index.php\/2025\/08\/21\/data-science-sante-machine-learning\/#article","isPartOf":{"@id":"https:\/\/dali.science\/index.php\/2025\/08\/21\/data-science-sante-machine-learning\/"},"author":{"name":"Martin PRODEL","@id":"https:\/\/dali.science\/#\/schema\/person\/370a84159837b710cf2d94f62c32cc9f"},"headline":"Faut-il (vraiment) faire du machine learning sur les donn\u00e9es de sant\u00e9 ?","datePublished":"2025-08-21T07:04:50+00:00","dateModified":"2025-08-21T10:44:19+00:00","mainEntityOfPage":{"@id":"https:\/\/dali.science\/index.php\/2025\/08\/21\/data-science-sante-machine-learning\/"},"wordCount":1722,"commentCount":0,"publisher":{"@id":"https:\/\/dali.science\/#organization"},"image":{"@id":"https:\/\/dali.science\/index.php\/2025\/08\/21\/data-science-sante-machine-learning\/#primaryimage"},"thumbnailUrl":"https:\/\/lh7-rt.googleusercontent.com\/docsz\/AD_4nXdyhWX-NrD4uNq0tiktJLcRhp0IlosrE1-D_bDEUJyVsa0xTDsQPI8GoiMTVGE6BTnnxIXMfjcpcxSI3yCCqcJQw5mioYEWoD92IXZJTNjx0brx97cPiX0mpaXOhJoqb-QMjcaxew?key=Ylxwb57Ne-i0ybw5GAX8ug","keywords":["data science","donn\u00e9es de sant\u00e9","machine learning","retour d'exp\u00e9rience"],"articleSection":["machine learning \/ IA"],"inLanguage":"fr-FR","potentialAction":[{"@type":"CommentAction","name":"Comment","target":["https:\/\/dali.science\/index.php\/2025\/08\/21\/data-science-sante-machine-learning\/#respond"]}]},{"@type":"WebPage","@id":"https:\/\/dali.science\/index.php\/2025\/08\/21\/data-science-sante-machine-learning\/","url":"https:\/\/dali.science\/index.php\/2025\/08\/21\/data-science-sante-machine-learning\/","name":"Faut-il (vraiment) faire du machine learning sur les donn\u00e9es de sant\u00e9 ?","isPartOf":{"@id":"https:\/\/dali.science\/#website"},"primaryImageOfPage":{"@id":"https:\/\/dali.science\/index.php\/2025\/08\/21\/data-science-sante-machine-learning\/#primaryimage"},"image":{"@id":"https:\/\/dali.science\/index.php\/2025\/08\/21\/data-science-sante-machine-learning\/#primaryimage"},"thumbnailUrl":"https:\/\/lh7-rt.googleusercontent.com\/docsz\/AD_4nXdyhWX-NrD4uNq0tiktJLcRhp0IlosrE1-D_bDEUJyVsa0xTDsQPI8GoiMTVGE6BTnnxIXMfjcpcxSI3yCCqcJQw5mioYEWoD92IXZJTNjx0brx97cPiX0mpaXOhJoqb-QMjcaxew?key=Ylxwb57Ne-i0ybw5GAX8ug","datePublished":"2025-08-21T07:04:50+00:00","dateModified":"2025-08-21T10:44:19+00:00","description":"Machine learning et intelligence artificielle en sant\u00e9 : entre engouement et lucidit\u00e9. Cas d\u2019usage concrets, erreurs fr\u00e9quentes, et conseils pour des projets utiles et partag\u00e9s. faire du machine learning sur donn\u00e9es de sant\u00e9 & boom du machine learning & IA g\u00e9n\u00e9rative","breadcrumb":{"@id":"https:\/\/dali.science\/index.php\/2025\/08\/21\/data-science-sante-machine-learning\/#breadcrumb"},"inLanguage":"fr-FR","potentialAction":[{"@type":"ReadAction","target":["https:\/\/dali.science\/index.php\/2025\/08\/21\/data-science-sante-machine-learning\/"]}]},{"@type":"ImageObject","inLanguage":"fr-FR","@id":"https:\/\/dali.science\/index.php\/2025\/08\/21\/data-science-sante-machine-learning\/#primaryimage","url":"https:\/\/lh7-rt.googleusercontent.com\/docsz\/AD_4nXdyhWX-NrD4uNq0tiktJLcRhp0IlosrE1-D_bDEUJyVsa0xTDsQPI8GoiMTVGE6BTnnxIXMfjcpcxSI3yCCqcJQw5mioYEWoD92IXZJTNjx0brx97cPiX0mpaXOhJoqb-QMjcaxew?key=Ylxwb57Ne-i0ybw5GAX8ug","contentUrl":"https:\/\/lh7-rt.googleusercontent.com\/docsz\/AD_4nXdyhWX-NrD4uNq0tiktJLcRhp0IlosrE1-D_bDEUJyVsa0xTDsQPI8GoiMTVGE6BTnnxIXMfjcpcxSI3yCCqcJQw5mioYEWoD92IXZJTNjx0brx97cPiX0mpaXOhJoqb-QMjcaxew?key=Ylxwb57Ne-i0ybw5GAX8ug"},{"@type":"BreadcrumbList","@id":"https:\/\/dali.science\/index.php\/2025\/08\/21\/data-science-sante-machine-learning\/#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"Accueil","item":"https:\/\/dali.science\/"},{"@type":"ListItem","position":2,"name":"Faut-il (vraiment) faire du machine learning sur les donn\u00e9es de sant\u00e9 ?"}]},{"@type":"WebSite","@id":"https:\/\/dali.science\/#website","url":"https:\/\/dali.science\/","name":"dali.science","description":"Transformer les donn\u00e9es en d\u00e9cisions \u00e9clair\u00e9es","publisher":{"@id":"https:\/\/dali.science\/#organization"},"potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/dali.science\/?s={search_term_string}"},"query-input":{"@type":"PropertyValueSpecification","valueRequired":true,"valueName":"search_term_string"}}],"inLanguage":"fr-FR"},{"@type":"Organization","@id":"https:\/\/dali.science\/#organization","name":"DALI","url":"https:\/\/dali.science\/","logo":{"@type":"ImageObject","inLanguage":"fr-FR","@id":"https:\/\/dali.science\/#\/schema\/logo\/image\/","url":"https:\/\/dali.science\/wp-content\/uploads\/2025\/05\/logo-avec-slogan.png","contentUrl":"https:\/\/dali.science\/wp-content\/uploads\/2025\/05\/logo-avec-slogan.png","width":1299,"height":475,"caption":"DALI"},"image":{"@id":"https:\/\/dali.science\/#\/schema\/logo\/image\/"},"sameAs":["https:\/\/www.linkedin.com\/company\/dali-science"]},{"@type":"Person","@id":"https:\/\/dali.science\/#\/schema\/person\/370a84159837b710cf2d94f62c32cc9f","name":"Martin PRODEL","image":{"@type":"ImageObject","inLanguage":"fr-FR","@id":"https:\/\/secure.gravatar.com\/avatar\/262a602c5fc065dd1dd94152b8a24b282b70ef33af194b5b89e93b757741d4a4?s=96&d=mm&r=g","url":"https:\/\/secure.gravatar.com\/avatar\/262a602c5fc065dd1dd94152b8a24b282b70ef33af194b5b89e93b757741d4a4?s=96&d=mm&r=g","contentUrl":"https:\/\/secure.gravatar.com\/avatar\/262a602c5fc065dd1dd94152b8a24b282b70ef33af194b5b89e93b757741d4a4?s=96&d=mm&r=g","caption":"Martin PRODEL"},"sameAs":["http:\/\/www.dali.science"],"url":"https:\/\/dali.science\/index.php\/author\/daliwordpress\/"}]}},"_links":{"self":[{"href":"https:\/\/dali.science\/index.php\/wp-json\/wp\/v2\/posts\/202","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/dali.science\/index.php\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/dali.science\/index.php\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/dali.science\/index.php\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/dali.science\/index.php\/wp-json\/wp\/v2\/comments?post=202"}],"version-history":[{"count":20,"href":"https:\/\/dali.science\/index.php\/wp-json\/wp\/v2\/posts\/202\/revisions"}],"predecessor-version":[{"id":286,"href":"https:\/\/dali.science\/index.php\/wp-json\/wp\/v2\/posts\/202\/revisions\/286"}],"wp:attachment":[{"href":"https:\/\/dali.science\/index.php\/wp-json\/wp\/v2\/media?parent=202"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/dali.science\/index.php\/wp-json\/wp\/v2\/categories?post=202"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/dali.science\/index.php\/wp-json\/wp\/v2\/tags?post=202"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}