Improving AI-Based Clinical Decision Support Systems and Their Integration Into Care From the Perspective of Experts: Interview Study Among Different Stakeholders
Notice bibliographique
Résumé
BACKGROUND: Artificial intelligence (AI)-based systems are receiving increasing attention in the health care sector. While the use of AI is well advanced in some medical applications, such as image recognition, it is still in its infancy in others, such as clinical decision support systems (CDSS). Examples of AI-based CDSS can be found in the context of sepsis prediction or antibiotic prescription. Scientific literature indicates that such systems can support physicians in their daily work and lead to improved patient outcomes. Nevertheless, there are various problems and barriers in this context that should be considered. OBJECTIVE: This study aimed to identify opportunities to optimize AI-based CDSS and their integration into care from the perspective of experts. METHODS: Semistructured web-based expert interviews were conducted. Experts representing the perspectives of patients; physicians; caregivers; developers; health insurance representatives; researchers (especially in law and IT); and experts in regulation, market admission and quality management or assurance, and ethics were included. The conversations were recorded and transcribed. Subsequently, a qualitative content analysis was performed. The different approaches to improvement were categorized into groups ("technology," "data," "users," "studies," "law," and "general"). These also served as deductive codes. Inductive codes were determined within an internal project workshop. RESULTS: In total, 13 individual and 2 double interviews were conducted with 17 experts. A total of 227 expert statements were included in the analysis. Suggestions were heterogeneous and concerned improvements: (1) in the systems themselves (eg, implementing comprehensive system training involving [future] users; using a comprehensive and high-quality database; considering usability, transparency, and customizability; preventing automation bias through control mechanisms or intelligent design; conducting studies to demonstrate the benefit of the system), (2) on the user side (eg, training [future] physicians could contribute to a more positive attitude and to greater awareness and questioning decision supports suggested by the system and ensuring that the use of the system does not lead to additional work), and (3) in the environment in which the systems are used (eg, increasing the digitalization of the health care system, especially in hospitals; providing transparent public communication about the benefits and risks of AI; providing research funding; clarifying open legal issues, eg, those related to liability; and standardizing and consolidating various approval processes). CONCLUSIONS: This study offers several possible strategies for improving AI-based CDSS and their integration into health care. These were found in the areas of "technology," "data," "users," "studies," "law," and "general." Systems, users, and the environment should be taken into account to ensure that the systems are used safely, effectively, and sustainably. Further studies should investigate both the effectiveness of strategies to improve AI-based CDSS and their integration into health care and the accuracy of their match to specific problems. INTERNATIONAL REGISTERED REPORT IDENTIFIER (IRRID): RR2-10.2196/62704.
Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.
Comment cette classification a été obtenuedéplier
Prédiction machine sur la base complète
Imitation des enseignantsNi prévalence calibrée, ni vérité terrain. Validation humaine à venir. Le volet Gemma est une étiquette directe du modèle pour chaque travail de la base, lue sur la notice réduite au titre. Le volet Codex est un classifieur appris des 10 348 étiquettes directes de Codex et calibré sur les taux pondérés de l'échantillon; les champs sans appui suffisant ne portent aucun appel Codex. Le mode candidate est l'union des deux volets; le consensus est leur intersection. Ces sorties portent le statut machine_predicted_unvalidated et ne sont pas des étiquettes humaines.
Scores du classifieur distillé par catégorie (deux têtes)
| Catégorie | Codex | Gemma |
|---|---|---|
| Métarecherche | 0,021 | 0,031 |
| Méta-épidémiologie (sens strict) | 0,001 | 0,001 |
| Méta-épidémiologie (sens large) | 0,001 | 0,000 |
| Bibliométrie | 0,002 | 0,001 |
| Études des sciences et des technologies | 0,006 | 0,005 |
| Communication savante | 0,003 | 0,004 |
| Science ouverte | 0,001 | 0,004 |
| Intégrité de la recherche | 0,002 | 0,003 |
| Charge utile insuffisante (le modèle a refusé de juger) | 0,002 | 0,000 |
Scores machine (provisoires)
Les deux têtes enseignantes du modèle étudiant, lues sur ce travail. Un score ordonne la base pour la relecture; il n'affirme jamais une catégorie, et le statut de validation accompagne chaque rangée tel quel.
Scores de référence d'un modèle non mature (critères de maturité non atteints, 7 itérations). Un score ordonne; il n'affirme jamais une catégorie.
score_only:v0-immature-baseline · tel quel depuis la passe de notation : score_only signifie que le nombre peut ordonner les travaux, et qu'aucune étiquette de catégorie n'en découleClassification
machine, non validéePrédiction automatique; un appel candidat d’une seule source (Gemma direct ou Codex distillé), pas un consensus.
Le détail, modèle par modèle et score par score, se trouve en fin de page sous « Comment cette classification a été obtenue ».