Papers by Jean-Baptiste Lamy
HAL (Le Centre pour la Communication Scientifique Directe), Aug 21, 2019
Medical terminologies are the basis of interoperability in medicine. They allow connecting the va... more Medical terminologies are the basis of interoperability in medicine. They allow connecting the various systems and data, and they facilitate search in databases. An example is the MedDRA terminology, which is used in particular for coding drug adverse events. However, these terminologies are often complex and they involve a huge number of terms. Consequently, it is difficult to browse them or to find the desired terms. Traditional approaches consist of lexical search, with the problems of synonymy and polysemy, or tree-based navigation, but the user often gets "lost" in the tree. Here, we propose a new approach for browsing medical terminologies: the use of pictograms and icons, for formulating the query in complement of a textual search box, and for displaying the search results. We applied this approach to the MedDRA terminology. We present both the methods and search algorithms and the resulting browsing interface, as well as the qualitative opinions of two pharmacovigilance experts.
Journal of the American Medical Informatics Association, May 11, 2019
Introduction: Clinical decision support systems (CDSS) implementing clinical practice guidelines ... more Introduction: Clinical decision support systems (CDSS) implementing clinical practice guidelines (CPGs) have 2 main limitations: they target only patients for whom CPGs provide explicit recommendations, and their rationale may be difficult to understand. These 2 limitations result in poor CDSS adoption. We designed Anti-bioHelp V R as a CDSS for antibiotic treatment. It displays the recommended and nonrecommended antibiotics, together with their properties, weighted by degree of importance as outlined in the CPGs. The aim of this study was to determine whether AntibioHelp V R could increase the confidence of general practitioners (GPs) in CPG recommendations and help them to extrapolate guidelines to patients for whom CPGs provide no explicit recommendations. Materials and Methods: We carried out a 2-stage crossover study in which GPs responded to clinical cases using CPG recommendations either alone or with explanations displayed through AntibioHelp V R . We compared error rates, confidence levels, and response times. Results: We included 64 GPs. When no explicit recommendation existed for a particular situation, AntibioHelp V R significantly decreased the error rate (À41%, P value ¼ 6x10 À13 ), and significantly increased GP confidence (þ8%, P value ¼ .02). This CDSS was considered to be usable by GPs (SUS score ¼ 64), despite a longer interaction time (þ9-22 seconds). By contrast, AntibioHelp V R had no significant effect if there was an explicit recommendation. Discussion/Conclusion: The visualization of weighted antibiotic properties helps GPs to extrapolate recommendations to patients for whom CPGs provide no explicit recommendations. It also increases GP confidence in their prescriptions for these patients. Further evaluations are required to determine the impact of AntibioHelp V R on antibiotic prescriptions in real clinical practice.
TSI, Dec 25, 2018
Cet article des Editions Lavoisier est disponible en acces libre et gratuit sur tsi.revuesonline.com
HAL (Le Centre pour la Communication Scientifique Directe), Jun 16, 2020
Several drug databases exist, but sometimes differ in their content. Here, we propose a taxonomy ... more Several drug databases exist, but sometimes differ in their content. Here, we propose a taxonomy of the variability we observed, and we present a tool for investigating the variability through the visual comparison of the properties of a given drug as represented in several sources or databases.
HAL (Le Centre pour la Communication Scientifique Directe), 2020
Dans le cancer du sein, l'intelligence artificielle peut aider les médecins à effectuer le diagno... more Dans le cancer du sein, l'intelligence artificielle peut aider les médecins à effectuer le diagnostic et à prescrire le bon traitement. Cependant, la plupart des méthodes récentes (comme l'apprentissage profond) sont des "boîtes noires" qui ne permettent pas d'expliquer les prédictions de machine. Au contraire, les médecins ont besoin de comprendre les recommandations des systèmes d'aide à la décision afin d'y adhérer. Nous proposons une approche visuelle de raisonnement à partir de cas, permettant une visualisation à la fois quantitative et qualitative de la similarité entre les cas. Cette approche a été testée sur 3 jeux de données publics pour le diagnostic et des données réelles pour la thérapie, et présentée à 11 médecins. Cet article est un résumé de: Jean-
2019 Fifth International Conference on Advances in Biomedical Engineering (ICABME)
Breast cancer therapy is particularly complex. Casebased reasoning (CBR) is an approach that can ... more Breast cancer therapy is particularly complex. Casebased reasoning (CBR) is an approach that can support clinicians when prescribing a therapy, and that is able to explain its recommendation to the clinicians. In a previous work, we proposed a visual CBR approach for helping clinicians to choose a treatment between four main categories (surgery, chemotherapy,...). However, these are broad categories and clinicians need more details about the treatment, e.g. several surgeries exist such as lumpectomy. Here, we extend our visual CBR approach for fully supporting the therapy for breast cancer, using a hierarchical approach: first, decide the category, then decide the exact treatment, etc.
Studies in health technology and informatics, 2021
Polypharmacy in elderly is a public health problem with both clinical (increase of adverse drug e... more Polypharmacy in elderly is a public health problem with both clinical (increase of adverse drug events) and economic issues. One solution is medication review, a structured assessment of patients' drug orders by the pharmacist for optimizing the therapy. However, this task is tedious, cognitively complex and error-prone, and only a few clinical decision support systems have been proposed for supporting it. Existing systems are either rule-based systems implementing guidelines, or documentary systems presenting drug knowledge. In this paper, we present the ABiMed research project, and, through literature reviews and brainstorming, we identified five candidate innovations for a decision support system for medication review: patient data transfer from GP to pharmacists, use of semantic technologies, association of rule-based and documentary approaches, use of machine learning, and a two-way discussion between pharmacist and GP after the medication review.
Dans le cancer du sein, l'intelligence artificielle peut aider les médecins à effectuer le di... more Dans le cancer du sein, l'intelligence artificielle peut aider les médecins à effectuer le diagnostic et à prescrire le bon traitement. Cependant, la plupart des méthodes récentes (comme l'apprentissage profond) sont des "boîtes noires" qui ne permettent pas d'expliquer les prédictions de machine. Au contraire, les médecins ont besoin de comprendre les recommandations des systèmes d'aide à la décision afin d'y adhérer. Nous proposons une approche visuelle de raisonnement à partir de cas, permettant une visualisation à la fois quantitative et qualitative de la similarité entre les cas. Cette approche a été testée sur 3 jeux de données publics pour le diagnostic et des données réelles pour la thérapie, et présentée à 11 médecins. Cet article est un résumé de: Jean-Baptiste Lamy, Boomadevi Sekar, Gilles Guezennec, Jacques Bouaud, Brigitte Séroussi. Explainable artificial intelligence for breast cancer: A visual case-based reasoning approach. Artificial Int...
Je remercie Marie-Christine Jaulent pour avoir essayé d'organiser visuellement le laboratoire qu'... more Je remercie Marie-Christine Jaulent pour avoir essayé d'organiser visuellement le laboratoire qu'elle dirige. Je remercie Lina Soualmia et Thierry Lecroq pour leur soutien et leur invitation à Rouen. Je remercie Brigitte Séroussi et Jacques Bouaud pour avoir pensé à moi même en Espagne. Je remercie Karima Sedki et Fadi Badra pour les longues discussions passées à discuter nos discussions. Je remercie Rosy Tsopra pour avoir été une collègue de bureau (presque ;-) supportable. Je remercie Gaoussou Camara pour m'avoir permis d'enseigner Python au pays des pythons. Je remercie Appoh Kouame pour m'avoir contraint, bien malgré lui, de raccourcir mon manuscrit. Je remercie Sylvie Despres pour ne se formaliser qu'informatiquement. Je remercie Alain Venot pour m'avoir appris que la rigueur ne s'apprend pas nécessairement avec rigueur. Je remercie Catherine Duclos pour m'avoir inoculé le virus alors qu'elle travaille sur l'antibiothérapie. Je remercie Vincent Rialle pour m'avoir appris que la curiosité est un vilain défaut mais un beau métier. Je remercie Mobin Yasini, Massoud Toussi et Vahid Ebrahiminia pour la poésie iranienne. Je remercie Gilles Venturini, Sandra Bringay et Nhan le Thanh pour avoir accepté de me rapporter. Je remercie Candice pour l'intérêt qu'elle porte à des langages qu'elle ne comprend pas toujours (ou pas encore !). Je remercie Antoine, Patricia et Estelle pour d'autres relectures. Je remercie également tous ceux que j'ai oublié, pour m'avoir appris que l'oubli est nécessaire à l'apprentissage. 1. En pratique médicale, plusieurs mesures de la glycémie sont nécessaires pour réaliser le diagnostic du diabète.
VCM (Visualization of Concept in Medicine) is an iconic language for representing key medical con... more VCM (Visualization of Concept in Medicine) is an iconic language for representing key medical concepts by icons. However, the use of this language with reference terminologies, such as SNOMED CT, will require the mapping of its icons to the terms of these terminologies. Here, we present and evaluate a semi-automatic semantic method for the mapping of SNOMED CT concepts to VCM icons. Both SNOMED CT and VCM are compositional in nature; SNOMED CT is expressed in description logic and VCM semantics are formalized in an OWL ontology. The proposed method involves the manual mapping of a limited number of underlying concepts from the VCM ontology, followed by automatic generation of the rest of the mapping. We applied this method to the clinical findings of the SNOMED CT CORE subset, and 100 randomly-selected mappings were evaluated by three experts. The results obtained were promising, with 82 of the SNOMED CT concepts correctly linked to VCM icons according to the experts. Most of the er...
Lecture Notes of the Institute for Computer Sciences, Social Informatics and Telecommunications Engineering, 2017
The modernization of African traditional medicine (TM) using IT faces to illiteracy of most of th... more The modernization of African traditional medicine (TM) using IT faces to illiteracy of most of the domain stakeholders. In order to assist traditional medicine practitioner (TMP) in theirs activities, we have propose an icon-based system to visually use plants and recipe in the drug preparation process. Therefore, traditional physicians can easily combine icons for medical prescription. For that, ontoMEDTRAD is an ontology including formal description for knowledge related to iconic representation of plants and recipes. Structurally, ontoMEDTRAD includes two modules: ontoConcept_term and ontoIcone denoting respectively the terms and the icons of concepts in this domain. Thus, avoiding any semantic issues, TMP can be free from language barriers, textual writing and reading in their work of healer. More specifically, this work aims to model plants and recipes in TM and propose compositional iconic language for plants and sketches for recipes.
Lecture Notes in Business Information Processing, 2017
When new drugs come onto the market, physicians have to decide whether they will consider the new... more When new drugs come onto the market, physicians have to decide whether they will consider the new drug for their future prescriptions. However, there is no absolute "right" decision: it depends on the physician's opinion, practice and patient base. Here, we propose a visual approach for supporting this decision using iconic, interactive and graphical presentation techniques for facilitating the comparison of a new drug with already existent drugs. By comparing the drug properties, the physician is aided in his decision task. We designed a prototype containing the properties of 4 new drugs and 22 "comparator" drugs. We presented the resulting system to a group of physicians. Preliminary evaluation results showed that this approach allowed physicians to make a decision when they were lacking information about the new drug, and to change their mind if they were overconfident in the new drug.
2017 21st International Conference Information Visualisation (IV), 2017
Figure 1. Weighted rainbow boxes showing the antibiotics available for treating cystitis in adult... more Figure 1. Weighted rainbow boxes showing the antibiotics available for treating cystitis in adults with risk of complication, and the disadvantages of each drug. The 10 available antibiotic drugs are displayed in columns, and the 6 disadvantages they may present are represented by rectangular red boxes. Each box covers the columns corresponding to the antibiotics having the disadvantage, and antibiotics were ordered so as to limited "holes" in the boxes. Here, a single hole is present (in the "ceftriaxone" column). The height of a box is proportional to the importance of the corresponding disadvantage, allowing the computation of a score for each drug, by visually summing the height of the boxes stacked in the drug's column. Here, lower score means better antibiotics, thus nitrofurantoin is the best treatment.
Studies in health technology and informatics, 2020
Several drug databases exist, but sometimes differ in their content. Here, we propose a taxonomy ... more Several drug databases exist, but sometimes differ in their content. Here, we propose a taxonomy of the variability we observed, and we present a tool for investigating the variability through the visual comparison of the properties of a given drug as represented in several sources or databases.
Summary Aim: To develop and evaluate the impact of an elec- tronic Follow-up Module (FM), based o... more Summary Aim: To develop and evaluate the impact of an elec- tronic Follow-up Module (FM), based on guidelines. Methods: A group of GPs defined the structure and the functionalities of the FM (reminders for recommended follow-up procedures, dysplay of synthetic data). The FM was implemented in an electronic medical recor d system (EMRS). Design: cluster randomised controlled trial, to compare eO + FM (intervention group) versus eO (control group). Population: GPs, users of the EMRS, and their patients with type 2 diabetes and/or hypertension, aged ≥ 25. Data collection : from the medical records, after anonymisation, and without knowing the randomisation group of the patients. We analyzed the up-to-date status of follow-up for eac h procedure in each group, before and after the inter- vention. The impact was measured by the absolute difference of the evolution between the groups, adjusted for age, gender, geographic origin and socio- professional group. Results: Fifty GPs included 27...
Résumé : L’apprentissage des préférences est un problème de recherche qui a reçu beaucoup d’atten... more Résumé : L’apprentissage des préférences est un problème de recherche qui a reçu beaucoup d’attention en intelligence artificielle ces dernières années. Il s’agit d’apprendre un modèle de préférences à partir de préférences observées. Ce modèle peut ensuite être utilisé pour obtenir une meilleure compréhension des préférences, et/ou pour effectuer des prédictions sur de nouvelles instances du problème. Les préférences sont généralement apprises à partir de données, par exemple sous la forme d’une matrice « instances × attributs ». Cependant, lorsque le domaine d’application est complexe, il peut être intéressant d’effectuer l’apprentissage à partir d’une ontologie formelle. Mais cela est plus difficile car (1) tous les attributs impactant les préférences n’ont pas nécessairement le même domaine, et (2) certains attributs peuvent correspondre à des propriétés n-aires (réifiées dans l’ontologie). Dans cet article, nous proposons une méthode pour l’apprentissage d’un modèle de préféren...
Journal of Visual Languages & Computing, 2017
Overlapping set visualization is a well-known problem in information visualization. This problem ... more Overlapping set visualization is a well-known problem in information visualization. This problem considers elements and sets containing all or part of the elements, a given element possibly belonging to more than one set. A typical example is the properties of the 20 amino-acids. A more complex application is the visual comparison of the contraindications or the adverse effects of several similar drugs. The knowledge involved is voluminous, each drug has many contraindications and adverse effects, some of them are shared with other drugs. Another real-life application is the visualization of gene annotation, each gene product being annotated with several annotation terms indicating the associated biological processes, molecular functions and cellular components. In this paper, we present rainbow boxes, a novel technique for visualizing overlapping sets, and its application to the presentation of the properties of amino-acids, the comparison of drug properties, and the visualization of gene annotation. This technique requires solving a combinatorial optimization problem; we propose a specific heuristic and we evaluate and compare it to general optimization algorithms. We also describe a user study comparing rainbow boxes to tables and showing that the former allowed physicians to find information significantly faster. Finally, we discuss the limits and the perspectives of rainbow boxes.
Journal of Biomedical Informatics, 2017
Objective: When a new drug is marketed, physicians must decide whether they will consider it for ... more Objective: When a new drug is marketed, physicians must decide whether they will consider it for their future practice. However, information about new drugs can be biased or hard to find. In this work, our objective was to study whether visual analytics could be used for comparing drug properties such as contraindications and adverse effects, and whether this visual comparison can help physicians to forge their own well-founded opinions about a new drug. Materials and Methods: First, an ontology for comparative drug information was designed, based on the expectations expressed during focus groups comprised of physicians. Second, a prototype of a visual drug comparator website was developed. It implements several visualization methods: rainbow boxes (a new technique for overlapping set visualization), dynamic tables, bar charts and icons. Third, the website was evaluated by 22 GPs for four new drugs. We recorded the general satisfaction, the physician's decision whether to consider the new drug for future prescription, both before and after consulting the website, and their arguments to justify their choice. Results: The prototype website permits the visual comparison of up to 10 drugs, including efficacy, contraindications, interactions, adverse effects, prices, dosage regimens,... All physicians found that the website allowed them to forge a well-founded opinion on the four new drugs. The physicians changed their decision about using a new drug in their future practice in 29 cases (out of 88) after consulting the website. Discussion and conclusion: Visual analytics is a promising approach for presenting drug information and for comparing drugs. The visual comparison of drug properties allows physicians to forge their opinions on drugs. Since drug properties are available in reference texts, reviewed by public health agencies, it could contribute to the independent of drug information.
2016 20th International Conference Information Visualisation (IV), 2016
Figure 1. Rainbow boxes displaying the 79 contraindications (26 distinct) of 8 drugs for erectile... more Figure 1. Rainbow boxes displaying the 79 contraindications (26 distinct) of 8 drugs for erectile dysfunction. The drugs are shown in columns and the contraindications in rectangular horizontal boxes (possibly with holes) covering the columns corresponding to the drugs sharing the contraindications.
Uploads
Papers by Jean-Baptiste Lamy
In this paper, we present rainbow boxes, a novel technique for visualizing overlapping sets, and its application to the properties of amino-acids and to the comparison of drug properties. We also describe a user study comparing rainbow boxes to tables and showing that the former allowed physicians
disponible pour se les approprier est limité. Dans le domaine médical, ce problème est très important, car les
connaissances sont particulièrement complexes et les erreurs peuvent être dramatiques.
Le format textuel reste la référence pour décrire les connaissances médicales avec précision. Mais les
médecins n’ont guère le temps de se référer aux textes durant leurs consultations. Les systèmes d’aide à la
décision clinique ont souvent été présentés comme une solution pour résoudre le problème de l’explosion des
connaissances médicales. Cependant, la plupart de ces systèmes demandent au médecin un gros travail de
saisie de données pour automatiser le raisonnement médical, ce qui conduit à une mauvaise acceptation par
les utilisateurs. Ces systèmes n’ont donc pas abouti au succès initialement espéré.
Dans ce document, nous proposons une approche différente pour l’aide à la décision : la
visualisation des connaissances. Il s’agit de présenter les connaissances de manière visuelle à l’utilisateur,
afin de faciliter la prise de décision. Nous étudierons cette nouvelle approche selon une démarche multidisci-
plinaire intégrant différentes branches de l’informatique : la représentation des connaissances, la visualisation
d’information et l’aide à la décision, avec un champ d’application médical.
Nous présenterons une démarche en 4 étapes pour la visualisation des connaissances :
1. L’étape de représentation des connaissances consiste à structurer et à formaliser les connaissances
que l’on souhaite visualiser. Nous présenterons une méthodologie pour la modélisation du domaine et
la construction d’ontologies dans un contexte médical, et un outil pour la programmation orientée
ontologie en langage Python.
2. L’étape de visualisation iconique consiste à représenter la nature des connaissances à l’aide d’icônes
et de langages iconiques. Nous décrirons une méthode pour formaliser la sémantique des langages
iconiques, ainsi que les applications rendues possibles par cette formalisation.
3. L’étape de visualisation structurelle consiste à représenter la structure des connaissances à l’aide
de technique de visualisation d’information. Nous présenterons les boîtes arc-en-ciel, une technique
nouvelle que nous avons développée pour visualiser les ensembles non disjoints, qui peut s’appliquer à
la visualisation des relations d’instanciation multiples dans les ontologies.
4. L’étape applicative consiste à intégrer la visualisation d’information au sein d’une application ou
d’un site web, et à évaluer auprès des utilisateurs les performances et l’utilisabilité des outils obtenus.
Ces évaluations permettront de valider notre approche.
Cette démarche sera illustrée au travers de trois projets auxquels j’ai participé en tant que postdoc puis
maître de conférences au laboratoire LIMICS, portant sur l’utilisation d’un langage iconique médical dans
les dossiers patients et les guides de bonnes pratiques cliniques, l’information comparative sur les nouveaux
médicaments, et la conception de recettes iconiques pour des remèdes en médecine traditionnelle d’Afrique
de l’Ouest.
Nous montrerons qu’il est possible de visualiser des connaissances abstraites complexes à l’aide d’icônes
et de techniques de visualisation, mais également que cette approche conduit à de bonnes performances et
une bonne acceptation lorsqu’elle est mise en œuvre pour l’aide à la décision.
Materials and Methods: First, an ontology for comparative drug information was designed, based on the expectations expressed during focus groups comprised of physicians. Second, a prototype of a visual drug comparator website was developed. It implements several visualization methods: rainbow boxes (a new technique for overlapping set visualization), dynamic tables, bar charts and icons. Third, the website was evaluated by 22 GPs for four new drugs. We recorded the general satisfaction, the physician’s decision whether to consider the new drug for future prescription, both before and after consulting the website, and their arguments to justify their choice.
Results: The prototype website permits the visual comparison of up to 10 drugs, including efficacy, contraindications, interactions, adverse effects, prices, dosage regimens,... All physicians found that the website allowed them to forge a well-founded opinion on the four new drugs. The physicians changed their decision about using a new drug in their future practice in 29 cases (out of 88) after consulting the website.
Discussion and conclusion: Visual analytics is a promising approach for presenting drug information and for comparing drugs. The visual comparison of drug properties allows physicians to forge their opinions on drugs. Since drug properties are available in reference texts, reviewed by public health agencies, it could contribute to the independent of drug information.
In this paper, we present rainbow boxes, a novel technique for visualizing overlapping sets, and its application to the presentation of the properties of amino-acids, the comparison of drug properties, and the visualization of gene annotation. This technique requires solving a combinatorial optimization problem; we propose a specific heuristic and we evaluate and compare it to general optimization algorithms. We also describe a user study comparing rainbow boxes to tables and showing that the former allowed physicians to find information significantly faster. Finally, we discuss the limits and the perspectives of rainbow boxes.