{"id":658248,"date":"2026-07-15T14:43:14","date_gmt":"2026-07-15T12:43:14","guid":{"rendered":"https:\/\/aivancity.ai\/blog\/?p=658248"},"modified":"2026-07-15T16:31:07","modified_gmt":"2026-07-15T14:31:07","slug":"phenoprompt-rag-and-agentic-ai-meets-clinical-phenotyping","status":"publish","type":"post","link":"https:\/\/aivancity.ai\/blog\/phenoprompt-rag-and-agentic-ai-meets-clinical-phenotyping\/","title":{"rendered":"PhenoPrompt: RAG and Agentic AI meets Clinical Phenotyping"},"content":{"rendered":"\n<p class=\"wp-block-paragraph\"><em>How a new open-source framework from aivancity is making clinical phenotyping accessible to any clinician \u2014 just by typing a question.<\/em><\/p>\n\n\n\n<figure class=\"wp-block-image size-full\"><img loading=\"lazy\" decoding=\"async\" width=\"1015\" height=\"477\" src=\"https:\/\/aivancity.ai\/blog\/wp-content\/uploads\/2026\/07\/image-7.png\" alt=\"\" class=\"wp-image-658249\" srcset=\"https:\/\/aivancity.ai\/blog\/wp-content\/uploads\/2026\/07\/image-7.png 1015w, https:\/\/aivancity.ai\/blog\/wp-content\/uploads\/2026\/07\/image-7-300x141.png 300w, https:\/\/aivancity.ai\/blog\/wp-content\/uploads\/2026\/07\/image-7-360x169.png 360w\" sizes=\"auto, (max-width: 1015px) 100vw, 1015px\" \/><\/figure>\n\n\n\n<style>\n.aivan-table-wrap {\n  max-width: 980px;\n  margin: 20px 0;\n  font-family: system-ui, -apple-system, \"Segoe UI\", Roboto, Arial, sans-serif;\n}\n\n.aivan-table {\n  width: 100%;\n  border-collapse: separate;\n  border-spacing: 0;\n  border: 1px solid #0064C6;\n  border-radius: 12px;\n  overflow: hidden;\n  box-shadow: 0 8px 24px rgba(0,0,0,0.06);\n  background: #ffffff;\n}\n\n.aivan-table thead th {\n  background-color: #0064C6;\n  color: #ffffff;\n  padding: 16px 18px;\n  text-align: center;\n  font-size: 15px;\n  font-weight: 700;\n}\n\n.aivan-table tbody td {\n  padding: 16px 18px;\n  vertical-align: top;\n  font-size: 14.5px;\n  line-height: 1.7;\n  color: #1f2937;\n  border-top: 1px solid #d9e6f6;\n  border-right: 1px solid #e5eef9;\n  background: #f9fbff;\n}\n\n.aivan-table tbody td:first-child {\n  width: 220px;\n  font-weight: 700;\n  background: #f2f7ff;\n  text-align: center;\n  vertical-align: middle;\n}\n\n.aivan-table tbody td:last-child {\n  border-right: none;\n}\n\n.aivan-table tbody tr:nth-child(even) td {\n  background: #f9fbff;\n}\n\n.aivan-table tbody tr:nth-child(even) td:first-child {\n  background: #f2f7ff;\n}\n\n.aivan-table a {\n  color: #0064C6;\n  text-decoration: none;\n  word-break: break-word;\n}\n\n.aivan-table a:hover {\n  text-decoration: underline;\n}\n<\/style>\n\n<div class=\"aivan-table-wrap\">\n<table class=\"aivan-table\">\n<thead>\n<tr>\n<th>\u00c9l\u00e9ment<\/th>\n<th>Information<\/th>\n<\/tr>\n<\/thead>\n\n<tbody>\n\n<tr>\n<td>Auteurs<\/td>\n<td>Shafiya Kausar, Anuradha Kar<\/td>\n<\/tr>\n\n<tr>\n<td>Titre de l&rsquo;article<\/td>\n<td><em>PhenoPrompt: A Prompt-Based Clinical Phenotyping Framework Using Entity Extraction, Unsupervised Clustering, and Retrieval-Augmented Generation<\/em><\/td>\n<\/tr>\n\n<tr>\n<td>Conf\u00e9rence<\/td>\n<td>IANLP 2026 (1st International Conference on AI and Natural Language Processing), 29\u201330 juin 2026, Faculty of Sciences Ben M&rsquo;Sick (FSBM), Hassan II University of Casablanca, Maroc.<\/td>\n<\/tr>\n\n<tr>\n<td>Session<\/td>\n<td>Session 8 : AI for Healthcare, Biomedical Signals and Clinical Decision Support<\/td>\n<\/tr>\n\n<tr>\n<td>Code source<\/td>\n<td>\n<a href=\"https:\/\/github.com\/Shafiya0101\/phenoprompt\" target=\"_blank\" rel=\"noopener\">\ngithub.com\/Shafiya0101\/phenoprompt\n<\/a><br>\n(Open source, sans GPU)\n<\/td>\n<\/tr>\n\n<tr>\n<td>Application de d\u00e9monstration<\/td>\n<td>\n<a href=\"https:\/\/phenoprompt-skausar-aivancity.streamlit.app\/\" target=\"_blank\" rel=\"noopener\">\nhttps:\/\/phenoprompt-skausar-aivancity.streamlit.app\/\n<\/a>\n<\/td>\n<\/tr>\n\n<\/tbody>\n<\/table>\n<\/div>\n\n\n\n<figure class=\"wp-block-image size-full\"><img loading=\"lazy\" decoding=\"async\" width=\"1015\" height=\"604\" src=\"https:\/\/aivancity.ai\/blog\/wp-content\/uploads\/2026\/07\/image-8.png\" alt=\"\" class=\"wp-image-658250\" srcset=\"https:\/\/aivancity.ai\/blog\/wp-content\/uploads\/2026\/07\/image-8.png 1015w, https:\/\/aivancity.ai\/blog\/wp-content\/uploads\/2026\/07\/image-8-300x179.png 300w, https:\/\/aivancity.ai\/blog\/wp-content\/uploads\/2026\/07\/image-8-360x214.png 360w\" sizes=\"auto, (max-width: 1015px) 100vw, 1015px\" \/><\/figure>\n\n\n\n<h2 class=\"wp-block-heading has-text-color has-link-color wp-elements-a7ff809b3bd99f5b313583cad9fe5931\" style=\"color:#986e13\">Introduction: Classroom Innovation That Goes Beyond Four Walls and Course Hours<\/h2>\n\n\n\n<p class=\"text-justify wp-block-paragraph\">PhenoPrompt began not in a dedicated research project, but in a classroom during the AI for Health course taught by Prof. Anuradha Kar for 5th year programme grande ecole students at&nbsp; aivancity.&nbsp; As part of a course evaluation process students were challenged to identify real-world problems at the intersection of AI and clinical language processing and propose data-driven solutions. It was in this setting that Shafiya Kausar, a current final year Master of Science student on the programme, developed the initial concept: a pipeline that could extract clinical entities from unstructured medical notes using the Inria Medkit library, cluster patients into phenotypic groups, and make those groups queryable without requiring disease-specific algorithms. Recognising both the significance of the idea and its potential to address a genuine gap in computational phenotyping, Prof. Kar encouraged Shafiya and proposed to convert the pipeline into a promptable tool that will benefit clinicians and medical AI researchers for discovering and characterising patient populations and classifying disease phenotypes. Together, they expanded the course prototype into a full research framework, adding the entity-augmented retrieval layer and crucially \u2014 an <em>agentic AI<\/em> component: a tool-calling agent that normalises colloquial or misspelled clinical queries into clean medical concepts before retrieval, making the system robust to the way clinicians actually speak and type. What started as a student assignment became a conference paper accepted at IANLP 2026, the 1st International Conference on AI and Natural Language Processing, and a live, open-source web application deployable on any institution&rsquo;s clinical corpus. The story of PhenoPrompt is, in that sense, also a story about what a well-posed course project and a timely faculty-student collaboration can produce.<\/p>\n\n\n\n<h2 class=\"wp-block-heading has-text-color has-link-color wp-elements-bfc77fdccd8bb1ca452c97642c81fc99\" style=\"color:#986e13\">The problem: phenotyping is still a one-disease-at-a-time search method<\/h2>\n\n\n\n<p class=\"text-justify wp-block-paragraph\">When a clinical researcher wants to study, say, all patients with <strong>type 2 diabetes and early-stage kidney disease<\/strong> at their hospital, they face a surprisingly hard problem. The information exists \u2014 scattered across thousands of clinical notes, laboratory records, and prescriptions \u2014 but accessing it requires writing a bespoke algorithm: a set of ICD codes to include, lab value thresholds, medication filters, and keyword searches in free text. That algorithm typically takes <strong>weeks to build<\/strong>, requires a clinical informatician and a domain expert working together, and must be validated by manually reading hundreds of patient records. And it works for exactly <strong>one disease<\/strong>. To study a second disease, the whole cycle starts over.<\/p>\n\n\n\n<p class=\"text-justify wp-block-paragraph\">The Phenotype KnowledgeBase (PheKB), maintained by the eMERGE consortium, is the community&rsquo;s attempt to share and reuse these algorithms. After years of collective effort across major research institutions, it lists only <strong>46 validated phenotyping algorithms<\/strong>. For the vast complexity of human disease, this small number is a bottleneck that cannot scale.<\/p>\n\n\n\n<figure class=\"wp-block-pullquote\"><blockquote><p><em>A cardiologist on a ward round cannot pause to navigate a scatter plot or submit a data request. She needs to ask a question and get an answer \u2014 now, in her own clinical language.<\/em><\/p><\/blockquote><\/figure>\n\n\n\n<p class=\"text-justify wp-block-paragraph\">This is the problem PhenoPrompt was built to solve. Presented at <strong>IANLP 2026<\/strong> in Casablanca, Morocco, our paper introduces a framework that transforms unstructured clinical notes into a <strong>universal, queryable phenotype space<\/strong> \u2014 one that any clinician can interrogate in plain English, without writing a single SQL query or ICD code.<\/p>\n\n\n\n<h2 class=\"wp-block-heading has-text-color has-link-color wp-elements-e2df817a519993a378d735ae41e3fe2f\" style=\"color:#986e13\">What is clinical phenotyping?<\/h2>\n\n\n\n<p class=\"text-justify wp-block-paragraph\">A <strong>phenotype<\/strong> is the set of observable characteristics of a patient \u2014 their symptoms, diagnoses, medication responses, and clinical trajectory \u2014 that distinguish them from others. Computational phenotyping is the algorithmic identification of patients who share a particular phenotypic profile from electronic health records (EHRs).<\/p>\n\n\n\n<p class=\"text-justify wp-block-paragraph\">In practice, EHRs contain two very different kinds of data. Structured data \u2014 ICD codes, laboratory values, prescription records \u2014 is easy to query but captures only a fraction of clinical reality. A substantial proportion of what clinicians actually know about a patient lives in <strong>free-text notes<\/strong>: the discharge summary, the consultation report, the nursing note. A patient can have peripheral neuropathy documented in every clinical note and never have it appear in a single ICD code, because it was not the primary reason for admission. Existing phenotyping systems largely miss this.<\/p>\n\n\n\n<h3 class=\"wp-block-heading has-text-color has-link-color wp-elements-1f5868288420cc6c50a21d20667e6673\" style=\"color:#0064c6\">Why it matters \u2014 four downstream uses<\/h3>\n\n\n\n<ul class=\"wp-block-list\">\n<li>Precision medicine: identify patient subgroups with distinct treatment responses<\/li>\n\n\n\n<li>Clinical trial cohort discovery: find patients matching eligibility criteria from notes, not just codes<\/li>\n\n\n\n<li>Disease subtyping: reveal heterogeneity within a single diagnostic category<\/li>\n\n\n\n<li>Novel phenotype discovery: surface patient clusters that carry no existing diagnostic label<\/li>\n<\/ul>\n\n\n\n<figure class=\"wp-block-image size-full\"><img loading=\"lazy\" decoding=\"async\" width=\"1015\" height=\"446\" src=\"https:\/\/aivancity.ai\/blog\/wp-content\/uploads\/2026\/07\/image-9.png\" alt=\"\" class=\"wp-image-658251\" srcset=\"https:\/\/aivancity.ai\/blog\/wp-content\/uploads\/2026\/07\/image-9.png 1015w, https:\/\/aivancity.ai\/blog\/wp-content\/uploads\/2026\/07\/image-9-300x132.png 300w, https:\/\/aivancity.ai\/blog\/wp-content\/uploads\/2026\/07\/image-9-360x158.png 360w\" sizes=\"auto, (max-width: 1015px) 100vw, 1015px\" \/><figcaption class=\"wp-element-caption\">Figure 1. Phenotyping and its uses (source: <a href=\"https:\/\/doi.org\/10.3389\/frai.2022.842306\">https:\/\/doi.org\/10.3389\/frai.2022.842306<\/a> )<\/figcaption><\/figure>\n\n\n\n<h2 class=\"wp-block-heading has-text-color has-link-color wp-elements-1f8135541f0bffd6422ed79533c507b1\" style=\"color:#986e13\">Introducing PhenoPrompt<\/h2>\n\n\n\n<p class=\"text-justify wp-block-paragraph\">PhenoPrompt is a three-stage pipeline that connects open-source clinical NLP, unsupervised machine learning, and retrieval-augmented generation into a single, GPU-free, reproducible framework. The key design principle is simple: <strong>build the phenotype space once \u2014 then let users ask questions about it in plain language, on demand, for any disease.<\/strong><\/p>\n\n\n\n<figure class=\"wp-block-image size-full\"><img loading=\"lazy\" decoding=\"async\" width=\"1015\" height=\"475\" src=\"https:\/\/aivancity.ai\/blog\/wp-content\/uploads\/2026\/07\/image-10.png\" alt=\"\" class=\"wp-image-658252\" srcset=\"https:\/\/aivancity.ai\/blog\/wp-content\/uploads\/2026\/07\/image-10.png 1015w, https:\/\/aivancity.ai\/blog\/wp-content\/uploads\/2026\/07\/image-10-300x140.png 300w, https:\/\/aivancity.ai\/blog\/wp-content\/uploads\/2026\/07\/image-10-360x168.png 360w\" sizes=\"auto, (max-width: 1015px) 100vw, 1015px\" \/><figcaption class=\"wp-element-caption\">Figure2: The PhenoPrompt app interface which can be accessed at :\u00a0<br><a href=\"https:\/\/phenoprompt-skausar-aivancity.streamlit.app\/\"><strong>https:\/\/phenoprompt-skausar-aivancity.streamlit.app\/<\/strong><\/a><\/figcaption><\/figure>\n\n\n\n<h3 class=\"wp-block-heading has-text-color has-link-color wp-elements-e092c75de0e90780cc414d20633a5a3e\" style=\"color:#0064c6\">Stage 1 \u2014 Clinical entity extraction with medkit<\/h3>\n\n\n\n<p class=\"text-justify wp-block-paragraph\">Each clinical note is processed using the medkit NLP pipeline: sentence and clause tokenization, rule-based extraction of clinical entities, negation and uncertainty detection to retain only affirmed findings, and construction of a TF-IDF feature matrix that captures the most discriminative clinical concepts across the corpus.<\/p>\n\n\n\n<figure class=\"wp-block-pullquote text-justify\"><blockquote><p><em>Across the full ~28,500-note corpus, the medkit pipeline \u2014 after correcting negation handling and normalising entity surface forms to canonical concepts \u2014 extracted a vocabulary of 78 distinct affirmed clinical concepts \u2014 disorders, medications, findings, and procedures \u2014 from raw free-text notes.<\/em><\/p><\/blockquote><\/figure>\n\n\n\n<h3 class=\"wp-block-heading has-text-color has-link-color wp-elements-4e40a2f065e75524233316879ca47c55\" style=\"color:#0064c6\">Stage 2 \u2014 Embedding map and phenotypic clustering<\/h3>\n\n\n\n<p class=\"text-justify wp-block-paragraph\">The entity feature matrix is compressed into a dense patient embedding using Latent Semantic Analysis (LSA) (Truncated SVD, 64 dimensions), then projected to two dimensions with UMAP to produce the clinical note embedding map: an interactive scatter plot in which spatial proximity reflects shared phenotypic profiles.<\/p>\n\n\n\n<figure class=\"wp-block-image size-full\"><img loading=\"lazy\" decoding=\"async\" width=\"923\" height=\"525\" src=\"https:\/\/aivancity.ai\/blog\/wp-content\/uploads\/2026\/07\/image-11.png\" alt=\"\" class=\"wp-image-658253\" srcset=\"https:\/\/aivancity.ai\/blog\/wp-content\/uploads\/2026\/07\/image-11.png 923w, https:\/\/aivancity.ai\/blog\/wp-content\/uploads\/2026\/07\/image-11-300x171.png 300w, https:\/\/aivancity.ai\/blog\/wp-content\/uploads\/2026\/07\/image-11-360x205.png 360w\" sizes=\"auto, (max-width: 923px) 100vw, 923px\" \/><figcaption class=\"wp-element-caption\">Figure3: Clustering map of clinical entities<\/figcaption><\/figure>\n\n\n\n<p class=\"text-justify wp-block-paragraph\">Clustering is performed in the full 64-dimensional embedding using the HDBSCAN algorithm, which automatically detects phenotypic clusters and labels outliers as noise. It outperformed K-Means, Agglomerative, and DBSCAN. Each cluster is characterized by its top enriched clinical entities, providing interpretable phenotype profiles rather than binary labels.<\/p>\n\n\n\n<figure class=\"wp-block-image size-full\"><img loading=\"lazy\" decoding=\"async\" width=\"803\" height=\"1196\" src=\"https:\/\/aivancity.ai\/blog\/wp-content\/uploads\/2026\/07\/image-12.png\" alt=\"\" class=\"wp-image-658254\" srcset=\"https:\/\/aivancity.ai\/blog\/wp-content\/uploads\/2026\/07\/image-12.png 803w, https:\/\/aivancity.ai\/blog\/wp-content\/uploads\/2026\/07\/image-12-201x300.png 201w, https:\/\/aivancity.ai\/blog\/wp-content\/uploads\/2026\/07\/image-12-360x536.png 360w\" sizes=\"auto, (max-width: 803px) 100vw, 803px\" \/><figcaption class=\"wp-element-caption\">Figure 4. Clusters identified from 28k clinical notes<\/figcaption><\/figure>\n\n\n\n<p class=\"text-justify wp-block-paragraph\"><em>\u00ab\u00a0Inspecting the phenotype profiles, the discovered clusters correspond to recognisable clinical groups \u2014 for example, anticoagulation, asthma-on-inhalers, diabetes\/metabolic, and anxiety\/depression phenotypes. (In our original 500-note evaluation, an LLM judge rated 5 of 6 clusters as coherent, with example labels such as &lsquo;cancer-related phenotype&rsquo; and &lsquo;stroke-like neurological syndrome&rsquo;.)\u00a0\u00bb<\/em><\/p>\n\n\n\n<h3 class=\"wp-block-heading has-text-color has-link-color wp-elements-4a2db0937c76a84cb5aeb3dc300d6b0b\" style=\"color:#0064c6\">Stage 3 \u2014 The RAG and Agentic RAG based clinical querying<\/h3>\n\n\n\n<figure class=\"wp-block-image size-full\"><img loading=\"lazy\" decoding=\"async\" width=\"1015\" height=\"525\" src=\"https:\/\/aivancity.ai\/blog\/wp-content\/uploads\/2026\/07\/image-13.png\" alt=\"\" class=\"wp-image-658255\" srcset=\"https:\/\/aivancity.ai\/blog\/wp-content\/uploads\/2026\/07\/image-13.png 1015w, https:\/\/aivancity.ai\/blog\/wp-content\/uploads\/2026\/07\/image-13-300x155.png 300w, https:\/\/aivancity.ai\/blog\/wp-content\/uploads\/2026\/07\/image-13-360x186.png 360w\" sizes=\"auto, (max-width: 1015px) 100vw, 1015px\" \/><figcaption class=\"wp-element-caption\">Figure 5.\u00a0 How RAG and Agentic AI are used in the PhenoPrompt pipeline<\/figcaption><\/figure>\n\n\n\n<p class=\"text-justify wp-block-paragraph\">At the heart of PhenoPrompt&rsquo;s query interface is a two-layer approach to grounded clinical question answering. In the first layer, standard <strong>Retrieval-Augmented Generation (RAG)<\/strong> is applied: rather than relying on a language model&rsquo;s parametric knowledge, the system first retrieves the most relevant patient notes from the indexed corpus using a weighted entity-matching score that combines exact concept matches, synonym-expanded matches, and partial token matches. The language model then generates its answer exclusively from this retrieved evidence, citing each source note by its identifier \u2014 so every clinical claim in the response is directly traceable to a real note in the dataset, eliminating hallucination by design. In the second and optional layer, an <strong>agentic RAG<\/strong> mode adds an intelligent pre-processing step before retrieval. A single-step tool-calling agent intercepts the clinician&rsquo;s free-text query, normalises colloquial or misspelled phrasing, corrects abbreviations, and extracts the salient clinical concepts \u2014 so a query like <em>\u00ab\u00a0fetch me diabetic patience with kidney issues\u00a0\u00bb<\/em> is silently cleaned into well-formed medical terms before the retrieval call is issued. This makes the system robust to the natural variation in how clinicians actually type, without requiring them to learn controlled vocabulary. The agent degrades gracefully to standard retrieval if the tool call is unavailable, ensuring the system remains functional across any deployment environment.<\/p>\n\n\n\n<style>\n.aiv4-pill{display:inline-flex;align-items:center;gap:4px;background:rgba(255,255,255,0.12);border:1px solid rgba(255,255,255,0.2);border-radius:20px;padding:4px 12px;font-size:11.5px;color:rgba(255,255,255,0.85);white-space:nowrap;font-family:-apple-system,BlinkMacSystemFont,'Segoe UI',Roboto,sans-serif;}\n.aiv4-pill svg{width:11px;height:11px;flex-shrink:0;stroke:#fff;}\n.aiv4-cta:hover{background:#fff !important;color:#232641 !important;}\n@media(max-width:640px){\n  .aiv4-inner{padding:28px 20px 24px !important;}\n  .aiv4-logo{width:140px !important;}\n  .aiv4-title{font-size:21px !important;}\n  .aiv4-cta{width:100% !important;text-align:center !important;}\n}\n<\/style>\n\n<div style=\"border-radius:16px;overflow:hidden;margin:40px 0;position:relative;font-family:-apple-system,BlinkMacSystemFont,'Segoe UI',Roboto,sans-serif;min-height:280px;\">\n\n  <img decoding=\"async\" src=\"https:\/\/aivancity.ai\/sites\/default\/files\/2025-11\/banniere-ia-managers.webp\" alt=\"\" style=\"position:absolute;inset:0;width:100%;height:100%;object-fit:cover;object-position:center center;display:block;\">\n\n  <div style=\"position:absolute;inset:0;background:linear-gradient(105deg,rgba(35,38,65,0.97) 0%,rgba(35,38,65,0.88) 45%,rgba(35,38,65,0.35) 100%);\"><\/div>\n\n  <div class=\"aiv4-inner\" style=\"position:relative;z-index:1;padding:32px 40px 30px;display:flex;flex-direction:column;gap:18px;max-width:640px;box-sizing:border-box;\">\n\n    <div style=\"display:flex;align-items:center;justify-content:space-between;flex-wrap:wrap;gap:10px;\">\n      <img decoding=\"async\" class=\"aiv4-logo\" src=\"https:\/\/aivancity.ai\/blog\/wp-content\/uploads\/2026\/05\/BLANC-FRANCAIS-COMPLET.png\" alt=\"aivancity\" style=\"width:160px;height:auto;display:block;opacity:0.95;\">\n      <span style=\"display:inline-flex;align-items:center;gap:5px;border-radius:6px;padding:4px 10px;font-size:11px;font-weight:700;letter-spacing:0.05em;text-transform:uppercase;background:rgba(57,134,225,0.2);border:1px solid rgba(57,134,225,0.5);color:#3986e1;\">\n        \u25cf Certification RS6787\n      <\/span>\n    <\/div>\n\n    <div style=\"display:flex;flex-direction:column;gap:8px;\">\n      <p style=\"margin:0;font-size:11.5px;font-weight:600;color:rgba(255,255,255,0.45);text-transform:uppercase;letter-spacing:0.1em;\">Formation pour dirigeants<\/p>\n      <h2 class=\"aiv4-title\" style=\"margin:0;font-size:26px;font-weight:700;color:#fff;line-height:1.2;\">IA &amp; Data Science<br><span style=\"color:#3986e1;\">pour les Managers<\/span><\/h2>\n      <p style=\"margin:0;font-size:13px;color:rgba(255,255,255,0.6);line-height:1.6;max-width:480px;\">Int\u00e9grez l&rsquo;IA dans votre strat\u00e9gie d&rsquo;entreprise. Une approche 360\u00b0 \u2014 Technologie, Business &amp; \u00c9thique \u2014 con\u00e7ue pour les d\u00e9cideurs. Pr\u00e9requis : 5 ans d&rsquo;exp\u00e9rience manag\u00e9riale.<\/p>\n    <\/div>\n\n    <div style=\"display:flex;flex-wrap:wrap;gap:7px;align-items:center;\">\n      <span class=\"aiv4-pill\"><svg viewBox=\"0 0 24 24\" fill=\"none\" stroke-width=\"2\"><circle cx=\"12\" cy=\"12\" r=\"10\"><\/circle><polyline points=\"12 6 12 12 16 14\"><\/polyline><\/svg>3 jours<\/span>\n      <span class=\"aiv4-pill\"><svg viewBox=\"0 0 24 24\" fill=\"none\" stroke-width=\"2\"><polyline points=\"20 6 9 17 4 12\"><\/polyline><\/svg>\u00c9ligible CPF \u2014 1 800 \u20ac HT<\/span>\n      <span class=\"aiv4-pill\"><svg viewBox=\"0 0 24 24\" fill=\"none\" stroke-width=\"2\"><path d=\"M12 2C8.13 2 5 5.13 5 9c0 5.25 7 13 7 13s7-7.75 7-13c0-3.87-3.13-7-7-7z\"><\/path><circle cx=\"12\" cy=\"9\" r=\"2.5\"><\/circle><\/svg>Paris-Villejuif &amp; Nice<\/span>\n    <\/div>\n\n    <a class=\"aiv4-cta\" href=\"https:\/\/aivancity.ai\/formations-professionnel\/ia-et-data-science-pour-les-managers\" target=\"_blank\" rel=\"noopener noreferrer\" style=\"display:inline-block;width:fit-content;background:rgba(255,255,255,0.12) !important;color:#fff !important;font-size:13.5px;font-weight:600;padding:11px 26px;border-radius:8px;text-decoration:none !important;white-space:nowrap;border:1px solid rgba(255,255,255,0.35);\">\n      D\u00e9couvrir la formation \u2192\n    <\/a>\n\n  <\/div>\n<\/div>\n\n\n\n<h3 class=\"wp-block-heading has-text-color has-link-color wp-elements-b6a00a90971d24646eadba62f1cd8b41\" style=\"color:#0064c6\">Finally \u2014 The prompt interface<\/h3>\n\n\n\n<p class=\"text-justify wp-block-paragraph\">This is where PhenoPrompt becomes a clinical decision-support tool rather than a batch analysis pipeline. A clinician types a question \u2014 <strong>\u00ab\u00a0What medications are documented for patients with diabetes and kidney disease?\u00a0\u00bb<\/strong> \u2014 and the system responds with a grounded, cited answer drawn from the actual patient notes.<\/p>\n\n\n\n<p class=\"text-justify wp-block-paragraph\">The retrieval engine processes the query through three complementary mechanisms: exact entity matching, <strong>synonym expansion<\/strong> (so \u00ab\u00a0T2DM\u00a0\u00bb finds \u00ab\u00a0diabetes\u00a0\u00bb, \u00ab\u00a0breathless\u00a0\u00bb finds \u00ab\u00a0shortness of breath\u00a0\u00bb), and partial token matching. Notes are scored by a weighted TF-IDF formula across all matched entities, the top-K notes are retrieved, and an LLM generates an answer <strong>constrained to cite only retrieved evidence<\/strong> \u2014 preventing hallucination by design.<\/p>\n\n\n\n<p class=\"text-justify wp-block-paragraph\">The synonym layer was shown to be essential: without it, queries like \u00ab\u00a0breathless\u00a0\u00bb and \u00ab\u00a0fluid overload\u00a0\u00bb returned <strong>zero results<\/strong>; with it, all four synonym-dependent queries recovered the full result set.<\/p>\n\n\n\n<figure class=\"wp-block-pullquote text-justify\"><blockquote><p><em>The system also offers an agentic retrieval mode in which a single-step tool-calling agent first normalises the query into clean clinical concepts \u2014 correcting spelling and extracting salient conditions \u2014 before issuing the retrieval call. This makes the system robust to colloquial or misspelled questions.<\/em><\/p><\/blockquote><\/figure>\n\n\n\n<h2 class=\"wp-block-heading has-text-color has-link-color wp-elements-92eaa307a078fbe48b2853c6960db82b\" style=\"color:#986e13\">What makes PhenoPrompt novel<\/h2>\n\n\n\n<p class=\"text-justify wp-block-paragraph\">Several phenotyping systems exist. What distinguishes PhenoPrompt is a specific combination of properties that, to our knowledge, no prior system achieves together:<\/p>\n\n\n\n<h3 class=\"wp-block-heading has-text-color has-link-color wp-elements-bfd75e3d415eba35d04a5b0c71dbb7ad\" style=\"color:#0064c6\">Universal scope, not disease-specific<\/h3>\n\n\n\n<p class=\"text-justify wp-block-paragraph\">Every existing phenotyping pipeline targets one disease. PhenoPrompt builds a single phenotype space from which any disease can be queried. The index is built once; the query phenotype is specified at query time, not at build time.<\/p>\n\n\n\n<h3 class=\"wp-block-heading has-text-color has-link-color wp-elements-8d43971f0cfded4b994aca7c1e8183ca\" style=\"color:#0064c6\">Phenotype mix, not binary label<\/h3>\n\n\n\n<p class=\"text-justify wp-block-paragraph\">The output is not a case\/control label but a ranked entity profile per cluster, showing which conditions, medications, and findings are enriched relative to the full population. This preserves and surfaces the heterogeneity that binary outputs discard.<\/p>\n\n\n\n<h3 class=\"wp-block-heading has-text-color has-link-color wp-elements-575bfbb59f48c1aab16e4b49dac46a4a\" style=\"color:#0064c6\">Entity-augmented retrieval<\/h3>\n\n\n\n<p class=\"text-justify wp-block-paragraph\">Rather than dense vector similarity alone, retrieval uses the same medkit entity pipeline that built the index \u2014 matching is structural, not just semantic. This makes matches interpretable: you can see exactly which entity triggered a retrieval.<\/p>\n\n\n\n<h3 class=\"wp-block-heading has-text-color has-link-color wp-elements-54ff50ef47badf1857f219192a264a72\" style=\"color:#0064c6\">Free-text first<\/h3>\n\n\n\n<p class=\"text-justify wp-block-paragraph\">Because entities are extracted from clinical notes rather than billing codes, PhenoPrompt captures conditions that are documented but never coded \u2014 a systematic blind spot of all ICD-based systems.<\/p>\n\n\n\n<h3 class=\"wp-block-heading has-text-color has-link-color wp-elements-9de2133f74798f8b34c1365bdbf43586\" style=\"color:#0064c6\">Clinician-accessible<\/h3>\n\n\n\n<p class=\"text-justify wp-block-paragraph\">The entire system is deployed as a web application requiring no SQL expertise, no data engineering, and no per-disease algorithm authoring. A clinician types a question and reads an answer with cited source notes.<\/p>\n\n\n\n<h2 class=\"wp-block-heading has-text-color has-link-color wp-elements-fd10d0e87817ab98afacb692daa451cd\" style=\"color:#986e13\">The live web application<\/h2>\n\n\n\n<p class=\"text-justify wp-block-paragraph\">PhenoPrompt is not just a research prototype. It is deployed as a publicly accessible Streamlit web application at <a href=\"https:\/\/phenoprompt-skausar-aivancity.streamlit.app\/\">https:\/\/phenoprompt-skausar-aivancity.streamlit.app\/<\/a> , built entirely on CPU-only infrastructure and requiring no paid model subscription to use (an open LLM API key unlocks the generative answer layer; without it, the system returns ranked retrieved notes directly).<\/p>\n\n\n\n<figure class=\"wp-block-image size-full\"><img loading=\"lazy\" decoding=\"async\" width=\"1015\" height=\"608\" src=\"https:\/\/aivancity.ai\/blog\/wp-content\/uploads\/2026\/07\/image-14.png\" alt=\"\" class=\"wp-image-658256\" srcset=\"https:\/\/aivancity.ai\/blog\/wp-content\/uploads\/2026\/07\/image-14.png 1015w, https:\/\/aivancity.ai\/blog\/wp-content\/uploads\/2026\/07\/image-14-300x180.png 300w, https:\/\/aivancity.ai\/blog\/wp-content\/uploads\/2026\/07\/image-14-360x216.png 360w\" sizes=\"auto, (max-width: 1015px) 100vw, 1015px\" \/><figcaption class=\"wp-element-caption\">Figure5. Sample clinical question in natural language being asked to PhenoPrompt and its corresponding response from PhenoPrompt. It can be seen that besides answering the question Phenoprompt also provides the reference to relevant clinical note where it found the information. An extended list of such questions may be found in the GitHub repository.<\/figcaption><\/figure>\n\n\n\n<h3 class=\"wp-block-heading has-text-color has-link-color wp-elements-abde66e22327cfd53e9f026cd7cc8be9\" style=\"color:#0064c6\">What you can do in the app:<\/h3>\n\n\n\n<ul class=\"wp-block-list\">\n<li>Phenotype explorer: view the UMAP embedding map of your corpus, coloured by discovered cluster<\/li>\n\n\n\n<li>Cluster profiles: see per-cluster cards with the top enrichment-scored entities for each phenotype<\/li>\n\n\n\n<li>Top concepts view: inspect the most frequent clinical concepts across the corpus<\/li>\n\n\n\n<li>Map overlay: re-colour the embedding map by discovered cluster or by entity density per note<\/li>\n\n\n\n<li>Notes table: browse every note with its cluster label and phenotype profile<\/li>\n\n\n\n<li>PhenoPrompt tab: type clinical questions in plain English and receive grounded, cited answers in standard or agentic retrieval mode<\/li>\n<\/ul>\n\n\n\n<h2 class=\"wp-block-heading has-text-color has-link-color wp-elements-7326fc7fc4b4bf2d48db6830043a0bae\" style=\"color:#986e13\">Presenting PhenoPrompt at IANLP 2026<\/h2>\n\n\n\n<p class=\"text-justify wp-block-paragraph\">PhenoPrompt was accepted for oral presentation at <strong>IANLP 2026 \u2014 the 1st International Conference on Artificial Intelligence and Natural Language Processing<\/strong>, held 29\u201330 June 2026 at the Faculty of Sciences Ben M&rsquo;Sick (FSBM), Hassan II University of Casablanca, Morocco. The conference brings together researchers and practitioners working at the intersection of AI and NLP, with a particular focus on real-world applications including healthcare, education, and industry.<\/p>\n\n\n\n<p class=\"text-justify wp-block-paragraph\">The paper is presented in <strong>Session 8: AI for Healthcare, Biomedical Signals and Clinical Decision Support<\/strong>. The IANLP conference is notable for its strong multilingual and francophone NLP community, and we look forward to discussing how PhenoPrompt&rsquo;s architecture \u2014 built in part on medkit, which was originally developed for French clinical corpora \u2014 could be extended to Arabic and French EHR data, opening phenotyping capabilities to clinical data warehouses beyond the Anglophone world.<\/p>\n\n\n\n<figure class=\"wp-block-pullquote text-justify\"><blockquote><p><em>IANLP 2026 is the first edition of a new conference series dedicated to AI for NLP, organised by AM2I &amp; LTIM &amp; FSBM at Hassan II University of Casablanca. aivancity is represented both in the paper programme and on the scientific committee.<\/em><\/p><\/blockquote><\/figure>\n\n\n\n<h2 class=\"wp-block-heading has-text-color has-link-color wp-elements-516b135b6210f48026fa53385b4d1dca\" style=\"color:#986e13\">What comes next<\/h2>\n\n\n\n<p class=\"text-justify wp-block-paragraph\">Although PhenoPrompt has progressed from a 500-note proof of concept to a 28,500-note corpus, it is an ongoing body of research and development. Future work focuses on validating it with real-world EHRs and clinician annotations, adopting UMLS-grounded biomedical NER, extending support to multilingual clinical notes, and incorporating longitudinal and hierarchical phenotyping to better model disease progression and patient trajectories.<\/p>\n\n\n\n<h2 class=\"wp-block-heading has-text-color has-link-color wp-elements-0c653b95aa68f2421af94736f007d3a0\" style=\"color:#986e13\">In closing<\/h2>\n\n\n\n<figure class=\"wp-block-pullquote text-justify\"><blockquote><p><em>\u00ab\u00a0PhenoPrompt turns the traditional computational phenotyping workflow from a batch analysis pipeline into an interactive, exploratory interface \u2014 accessible to any clinician, for any disease, from any clinical corpus.\u00a0\u00bb<\/em><\/p><\/blockquote><\/figure>\n\n\n\n<p class=\"text-justify wp-block-paragraph\">The ambition behind PhenoPrompt is straightforward: clinical knowledge lives in clinical notes, and the people who understand that knowledge best \u2014 clinicians \u2014 should be able to access it directly, in their own language, without a technical intermediary. We hope that by releasing the framework as <strong>open-source software<\/strong> and deploying a live demo, we can invite the community \u2014 researchers, clinicians, data scientists \u2014 to explore the idea, test it on their own data, and help build it into something clinically useful.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">The full paper will be available in the IANLP 2026 proceedings by Springer. The code is at <a href=\"https:\/\/github.com\/Shafiya0101\/phenoprompt\">github.com\/Shafiya0101\/phenoprompt<\/a>. The live application is at <a href=\"https:\/\/phenoprompt-skausar-aivancity.streamlit.app\/\">https:\/\/phenoprompt-skausar-aivancity.streamlit.app\/<\/a>&nbsp; . We welcome collaboration, feedback, and questions at <a href=\"mailto:kar@aivancity.ai\">kar@aivancity.ai<\/a> and <a href=\"mailto:shafiya.kausar@aivancity.education\">shafiya.kausar@aivancity.education<\/a>.<\/p>\n\n\n\n<hr class=\"wp-block-separator has-alpha-channel-opacity\"\/>\n\n\n\n<div style=\"\n  border-left: 5px solid #0064C6;\n  background: #f4f8fd;\n  padding: 18px 22px;\n  margin: 25px 0;\n  border-radius: 8px;\n  line-height: 1.7;\n  font-size: 15px;\n  color: #1f2937;\n\">\nShafiya Kausar and Anuradha Kar are researchers at aivancity School of AI &amp; Data for Business &amp; Society, France. This article is based on the paper <em>PhenoPrompt: A Prompt-Based Clinical Phenotyping Framework Using Entity Extraction, Unsupervised Clustering, and Retrieval-Augmented Generation<\/em>, presented at IANLP 2026, Casablanca, Morocco.\n<\/div>\n","protected":false},"excerpt":{"rendered":"<p>How a new open-source framework from aivancity is making clinical phenotyping accessible to any clinician \u2014 just by typing a question. \u00c9l\u00e9ment Information Auteurs Shafiya Kausar, Anuradha Kar Titre de l&rsquo;article PhenoPrompt: A Prompt-Based Clinical&#8230;<\/p>\n","protected":false},"author":7,"featured_media":658257,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"content-type":"","footnotes":""},"categories":[28],"tags":[],"class_list":["post-658248","post","type-post","status-publish","format-standard","has-post-thumbnail","category-articles"],"yoast_head":"<!-- This site is optimized with the Yoast SEO plugin v28.0 - https:\/\/yoast.com\/product\/yoast-seo-wordpress\/ -->\n<title>PhenoPrompt: RAG and Agentic AI meets Clinical Phenotyping - aivancity blog<\/title>\n<meta name=\"description\" content=\"PhenoPrompt combine IA agentique, RAG et NLP clinique pour permettre aux m\u00e9decins d\u2019interroger des dossiers m\u00e9dicaux en langage naturel et d\u2019identifier des ph\u00e9notypes cliniques avec des r\u00e9ponses fiables et tra\u00e7ables.\" \/>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/aivancity.ai\/blog\/phenoprompt-rag-and-agentic-ai-meets-clinical-phenotyping\/\" \/>\n<meta property=\"og:locale\" content=\"fr_FR\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"PhenoPrompt: RAG and Agentic AI meets Clinical Phenotyping - aivancity blog\" \/>\n<meta property=\"og:description\" content=\"PhenoPrompt combine IA agentique, RAG et NLP clinique pour permettre aux m\u00e9decins d\u2019interroger des dossiers m\u00e9dicaux en langage naturel et d\u2019identifier des ph\u00e9notypes cliniques avec des r\u00e9ponses fiables et tra\u00e7ables.\" \/>\n<meta property=\"og:url\" content=\"https:\/\/aivancity.ai\/blog\/phenoprompt-rag-and-agentic-ai-meets-clinical-phenotyping\/\" \/>\n<meta property=\"og:site_name\" content=\"aivancity blog\" \/>\n<meta property=\"article:published_time\" content=\"2026-07-15T12:43:14+00:00\" \/>\n<meta property=\"article:modified_time\" content=\"2026-07-15T14:31:07+00:00\" \/>\n<meta property=\"og:image\" content=\"https:\/\/aivancity.ai\/blog\/wp-content\/uploads\/2026\/07\/photo-article-1.png\" \/>\n\t<meta property=\"og:image:width\" content=\"1125\" \/>\n\t<meta property=\"og:image:height\" content=\"750\" \/>\n\t<meta property=\"og:image:type\" content=\"image\/png\" \/>\n<meta name=\"author\" content=\"aivancity\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:label1\" content=\"\u00c9crit par\" \/>\n\t<meta name=\"twitter:data1\" content=\"aivancity\" \/>\n\t<meta name=\"twitter:label2\" content=\"Dur\u00e9e de lecture estim\u00e9e\" \/>\n\t<meta name=\"twitter:data2\" content=\"14 minutes\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\\\/\\\/schema.org\",\"@graph\":[{\"@type\":\"Article\",\"@id\":\"https:\\\/\\\/aivancity.ai\\\/blog\\\/phenoprompt-rag-and-agentic-ai-meets-clinical-phenotyping\\\/#article\",\"isPartOf\":{\"@id\":\"https:\\\/\\\/aivancity.ai\\\/blog\\\/phenoprompt-rag-and-agentic-ai-meets-clinical-phenotyping\\\/\"},\"author\":{\"name\":\"aivancity\",\"@id\":\"https:\\\/\\\/aivancity.ai\\\/blog\\\/#\\\/schema\\\/person\\\/70f8508e84e45571c5fd172ea40ef3d4\"},\"headline\":\"PhenoPrompt: RAG and Agentic AI meets Clinical Phenotyping\",\"datePublished\":\"2026-07-15T12:43:14+00:00\",\"dateModified\":\"2026-07-15T14:31:07+00:00\",\"mainEntityOfPage\":{\"@id\":\"https:\\\/\\\/aivancity.ai\\\/blog\\\/phenoprompt-rag-and-agentic-ai-meets-clinical-phenotyping\\\/\"},\"wordCount\":2480,\"commentCount\":0,\"image\":{\"@id\":\"https:\\\/\\\/aivancity.ai\\\/blog\\\/phenoprompt-rag-and-agentic-ai-meets-clinical-phenotyping\\\/#primaryimage\"},\"thumbnailUrl\":\"https:\\\/\\\/aivancity.ai\\\/blog\\\/wp-content\\\/uploads\\\/2026\\\/07\\\/photo-article-1.png\",\"articleSection\":[\"Articles\"],\"inLanguage\":\"fr-FR\",\"potentialAction\":[{\"@type\":\"CommentAction\",\"name\":\"Comment\",\"target\":[\"https:\\\/\\\/aivancity.ai\\\/blog\\\/phenoprompt-rag-and-agentic-ai-meets-clinical-phenotyping\\\/#respond\"]}]},{\"@type\":\"WebPage\",\"@id\":\"https:\\\/\\\/aivancity.ai\\\/blog\\\/phenoprompt-rag-and-agentic-ai-meets-clinical-phenotyping\\\/\",\"url\":\"https:\\\/\\\/aivancity.ai\\\/blog\\\/phenoprompt-rag-and-agentic-ai-meets-clinical-phenotyping\\\/\",\"name\":\"PhenoPrompt: RAG and Agentic AI meets Clinical Phenotyping - aivancity blog\",\"isPartOf\":{\"@id\":\"https:\\\/\\\/aivancity.ai\\\/blog\\\/#website\"},\"primaryImageOfPage\":{\"@id\":\"https:\\\/\\\/aivancity.ai\\\/blog\\\/phenoprompt-rag-and-agentic-ai-meets-clinical-phenotyping\\\/#primaryimage\"},\"image\":{\"@id\":\"https:\\\/\\\/aivancity.ai\\\/blog\\\/phenoprompt-rag-and-agentic-ai-meets-clinical-phenotyping\\\/#primaryimage\"},\"thumbnailUrl\":\"https:\\\/\\\/aivancity.ai\\\/blog\\\/wp-content\\\/uploads\\\/2026\\\/07\\\/photo-article-1.png\",\"datePublished\":\"2026-07-15T12:43:14+00:00\",\"dateModified\":\"2026-07-15T14:31:07+00:00\",\"author\":{\"@id\":\"https:\\\/\\\/aivancity.ai\\\/blog\\\/#\\\/schema\\\/person\\\/70f8508e84e45571c5fd172ea40ef3d4\"},\"description\":\"PhenoPrompt combine IA agentique, RAG et NLP clinique pour permettre aux m\u00e9decins d\u2019interroger des dossiers m\u00e9dicaux en langage naturel et d\u2019identifier des ph\u00e9notypes cliniques avec des r\u00e9ponses fiables et tra\u00e7ables.\",\"breadcrumb\":{\"@id\":\"https:\\\/\\\/aivancity.ai\\\/blog\\\/phenoprompt-rag-and-agentic-ai-meets-clinical-phenotyping\\\/#breadcrumb\"},\"inLanguage\":\"fr-FR\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\\\/\\\/aivancity.ai\\\/blog\\\/phenoprompt-rag-and-agentic-ai-meets-clinical-phenotyping\\\/\"]}]},{\"@type\":\"ImageObject\",\"inLanguage\":\"fr-FR\",\"@id\":\"https:\\\/\\\/aivancity.ai\\\/blog\\\/phenoprompt-rag-and-agentic-ai-meets-clinical-phenotyping\\\/#primaryimage\",\"url\":\"https:\\\/\\\/aivancity.ai\\\/blog\\\/wp-content\\\/uploads\\\/2026\\\/07\\\/photo-article-1.png\",\"contentUrl\":\"https:\\\/\\\/aivancity.ai\\\/blog\\\/wp-content\\\/uploads\\\/2026\\\/07\\\/photo-article-1.png\",\"width\":1125,\"height\":750},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\\\/\\\/aivancity.ai\\\/blog\\\/phenoprompt-rag-and-agentic-ai-meets-clinical-phenotyping\\\/#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"Accueil\",\"item\":\"https:\\\/\\\/aivancity.ai\\\/blog\\\/\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"PhenoPrompt: RAG and Agentic AI meets Clinical Phenotyping\"}]},{\"@type\":\"WebSite\",\"@id\":\"https:\\\/\\\/aivancity.ai\\\/blog\\\/#website\",\"url\":\"https:\\\/\\\/aivancity.ai\\\/blog\\\/\",\"name\":\"aivancity blog\",\"description\":\"Advancing education in Artificial Intelligence\",\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\\\/\\\/aivancity.ai\\\/blog\\\/?s={search_term_string}\"},\"query-input\":{\"@type\":\"PropertyValueSpecification\",\"valueRequired\":true,\"valueName\":\"search_term_string\"}}],\"inLanguage\":\"fr-FR\"},{\"@type\":\"Person\",\"@id\":\"https:\\\/\\\/aivancity.ai\\\/blog\\\/#\\\/schema\\\/person\\\/70f8508e84e45571c5fd172ea40ef3d4\",\"name\":\"aivancity\",\"image\":{\"@type\":\"ImageObject\",\"inLanguage\":\"fr-FR\",\"@id\":\"https:\\\/\\\/secure.gravatar.com\\\/avatar\\\/0b60f844cf48367ece3a9988562f25406b914c56b83ccd3df68e4c07737dc27e?s=96&d=mm&r=g\",\"url\":\"https:\\\/\\\/secure.gravatar.com\\\/avatar\\\/0b60f844cf48367ece3a9988562f25406b914c56b83ccd3df68e4c07737dc27e?s=96&d=mm&r=g\",\"contentUrl\":\"https:\\\/\\\/secure.gravatar.com\\\/avatar\\\/0b60f844cf48367ece3a9988562f25406b914c56b83ccd3df68e4c07737dc27e?s=96&d=mm&r=g\",\"caption\":\"aivancity\"},\"url\":\"https:\\\/\\\/aivancity.ai\\\/blog\\\/author\\\/bouazizaivancity-ai\\\/\"}]}<\/script>\n<!-- \/ Yoast SEO plugin. -->","yoast_head_json":{"title":"PhenoPrompt: RAG and Agentic AI meets Clinical Phenotyping - aivancity blog","description":"PhenoPrompt combine IA agentique, RAG et NLP clinique pour permettre aux m\u00e9decins d\u2019interroger des dossiers m\u00e9dicaux en langage naturel et d\u2019identifier des ph\u00e9notypes cliniques avec des r\u00e9ponses fiables et tra\u00e7ables.","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:\/\/aivancity.ai\/blog\/phenoprompt-rag-and-agentic-ai-meets-clinical-phenotyping\/","og_locale":"fr_FR","og_type":"article","og_title":"PhenoPrompt: RAG and Agentic AI meets Clinical Phenotyping - aivancity blog","og_description":"PhenoPrompt combine IA agentique, RAG et NLP clinique pour permettre aux m\u00e9decins d\u2019interroger des dossiers m\u00e9dicaux en langage naturel et d\u2019identifier des ph\u00e9notypes cliniques avec des r\u00e9ponses fiables et tra\u00e7ables.","og_url":"https:\/\/aivancity.ai\/blog\/phenoprompt-rag-and-agentic-ai-meets-clinical-phenotyping\/","og_site_name":"aivancity blog","article_published_time":"2026-07-15T12:43:14+00:00","article_modified_time":"2026-07-15T14:31:07+00:00","og_image":[{"width":1125,"height":750,"url":"https:\/\/aivancity.ai\/blog\/wp-content\/uploads\/2026\/07\/photo-article-1.png","type":"image\/png"}],"author":"aivancity","twitter_card":"summary_large_image","twitter_misc":{"\u00c9crit par":"aivancity","Dur\u00e9e de lecture estim\u00e9e":"14 minutes"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"Article","@id":"https:\/\/aivancity.ai\/blog\/phenoprompt-rag-and-agentic-ai-meets-clinical-phenotyping\/#article","isPartOf":{"@id":"https:\/\/aivancity.ai\/blog\/phenoprompt-rag-and-agentic-ai-meets-clinical-phenotyping\/"},"author":{"name":"aivancity","@id":"https:\/\/aivancity.ai\/blog\/#\/schema\/person\/70f8508e84e45571c5fd172ea40ef3d4"},"headline":"PhenoPrompt: RAG and Agentic AI meets Clinical Phenotyping","datePublished":"2026-07-15T12:43:14+00:00","dateModified":"2026-07-15T14:31:07+00:00","mainEntityOfPage":{"@id":"https:\/\/aivancity.ai\/blog\/phenoprompt-rag-and-agentic-ai-meets-clinical-phenotyping\/"},"wordCount":2480,"commentCount":0,"image":{"@id":"https:\/\/aivancity.ai\/blog\/phenoprompt-rag-and-agentic-ai-meets-clinical-phenotyping\/#primaryimage"},"thumbnailUrl":"https:\/\/aivancity.ai\/blog\/wp-content\/uploads\/2026\/07\/photo-article-1.png","articleSection":["Articles"],"inLanguage":"fr-FR","potentialAction":[{"@type":"CommentAction","name":"Comment","target":["https:\/\/aivancity.ai\/blog\/phenoprompt-rag-and-agentic-ai-meets-clinical-phenotyping\/#respond"]}]},{"@type":"WebPage","@id":"https:\/\/aivancity.ai\/blog\/phenoprompt-rag-and-agentic-ai-meets-clinical-phenotyping\/","url":"https:\/\/aivancity.ai\/blog\/phenoprompt-rag-and-agentic-ai-meets-clinical-phenotyping\/","name":"PhenoPrompt: RAG and Agentic AI meets Clinical Phenotyping - aivancity blog","isPartOf":{"@id":"https:\/\/aivancity.ai\/blog\/#website"},"primaryImageOfPage":{"@id":"https:\/\/aivancity.ai\/blog\/phenoprompt-rag-and-agentic-ai-meets-clinical-phenotyping\/#primaryimage"},"image":{"@id":"https:\/\/aivancity.ai\/blog\/phenoprompt-rag-and-agentic-ai-meets-clinical-phenotyping\/#primaryimage"},"thumbnailUrl":"https:\/\/aivancity.ai\/blog\/wp-content\/uploads\/2026\/07\/photo-article-1.png","datePublished":"2026-07-15T12:43:14+00:00","dateModified":"2026-07-15T14:31:07+00:00","author":{"@id":"https:\/\/aivancity.ai\/blog\/#\/schema\/person\/70f8508e84e45571c5fd172ea40ef3d4"},"description":"PhenoPrompt combine IA agentique, RAG et NLP clinique pour permettre aux m\u00e9decins d\u2019interroger des dossiers m\u00e9dicaux en langage naturel et d\u2019identifier des ph\u00e9notypes cliniques avec des r\u00e9ponses fiables et tra\u00e7ables.","breadcrumb":{"@id":"https:\/\/aivancity.ai\/blog\/phenoprompt-rag-and-agentic-ai-meets-clinical-phenotyping\/#breadcrumb"},"inLanguage":"fr-FR","potentialAction":[{"@type":"ReadAction","target":["https:\/\/aivancity.ai\/blog\/phenoprompt-rag-and-agentic-ai-meets-clinical-phenotyping\/"]}]},{"@type":"ImageObject","inLanguage":"fr-FR","@id":"https:\/\/aivancity.ai\/blog\/phenoprompt-rag-and-agentic-ai-meets-clinical-phenotyping\/#primaryimage","url":"https:\/\/aivancity.ai\/blog\/wp-content\/uploads\/2026\/07\/photo-article-1.png","contentUrl":"https:\/\/aivancity.ai\/blog\/wp-content\/uploads\/2026\/07\/photo-article-1.png","width":1125,"height":750},{"@type":"BreadcrumbList","@id":"https:\/\/aivancity.ai\/blog\/phenoprompt-rag-and-agentic-ai-meets-clinical-phenotyping\/#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"Accueil","item":"https:\/\/aivancity.ai\/blog\/"},{"@type":"ListItem","position":2,"name":"PhenoPrompt: RAG and Agentic AI meets Clinical Phenotyping"}]},{"@type":"WebSite","@id":"https:\/\/aivancity.ai\/blog\/#website","url":"https:\/\/aivancity.ai\/blog\/","name":"aivancity blog","description":"Advancing education in Artificial Intelligence","potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/aivancity.ai\/blog\/?s={search_term_string}"},"query-input":{"@type":"PropertyValueSpecification","valueRequired":true,"valueName":"search_term_string"}}],"inLanguage":"fr-FR"},{"@type":"Person","@id":"https:\/\/aivancity.ai\/blog\/#\/schema\/person\/70f8508e84e45571c5fd172ea40ef3d4","name":"aivancity","image":{"@type":"ImageObject","inLanguage":"fr-FR","@id":"https:\/\/secure.gravatar.com\/avatar\/0b60f844cf48367ece3a9988562f25406b914c56b83ccd3df68e4c07737dc27e?s=96&d=mm&r=g","url":"https:\/\/secure.gravatar.com\/avatar\/0b60f844cf48367ece3a9988562f25406b914c56b83ccd3df68e4c07737dc27e?s=96&d=mm&r=g","contentUrl":"https:\/\/secure.gravatar.com\/avatar\/0b60f844cf48367ece3a9988562f25406b914c56b83ccd3df68e4c07737dc27e?s=96&d=mm&r=g","caption":"aivancity"},"url":"https:\/\/aivancity.ai\/blog\/author\/bouazizaivancity-ai\/"}]}},"amp_enabled":true,"_links":{"self":[{"href":"https:\/\/aivancity.ai\/blog\/wp-json\/wp\/v2\/posts\/658248","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/aivancity.ai\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/aivancity.ai\/blog\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/aivancity.ai\/blog\/wp-json\/wp\/v2\/users\/7"}],"replies":[{"embeddable":true,"href":"https:\/\/aivancity.ai\/blog\/wp-json\/wp\/v2\/comments?post=658248"}],"version-history":[{"count":2,"href":"https:\/\/aivancity.ai\/blog\/wp-json\/wp\/v2\/posts\/658248\/revisions"}],"predecessor-version":[{"id":658259,"href":"https:\/\/aivancity.ai\/blog\/wp-json\/wp\/v2\/posts\/658248\/revisions\/658259"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/aivancity.ai\/blog\/wp-json\/wp\/v2\/media\/658257"}],"wp:attachment":[{"href":"https:\/\/aivancity.ai\/blog\/wp-json\/wp\/v2\/media?parent=658248"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/aivancity.ai\/blog\/wp-json\/wp\/v2\/categories?post=658248"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/aivancity.ai\/blog\/wp-json\/wp\/v2\/tags?post=658248"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}