Device type for Artificial Intelligence - TTL Representation

AI Transparency on FHIR, published by HL7 International / Electronic Health Records. This guide is not an authorized publication; it is the continuous build for version 1.0.0-ballot built by the FHIR (HL7® FHIR® Standard) CI Build. This version is based on the current content of https://github.com/HL7/aitransparency-ig/ and changes regularly. See the Directory of published versions
: Device type for Artificial Intelligence

Page standards status: Trial-use
Maturity Level: 2
Raw ttl | Download

@prefix fhir: <http://hl7.org/fhir/> .
@prefix owl: <http://www.w3.org/2002/07/owl#> .
@prefix rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#> .
@prefix rdfs: <http://www.w3.org/2000/01/rdf-schema#> .
@prefix xsd: <http://www.w3.org/2001/XMLSchema#> .

# - resource -------------------------------------------------------------------

 a fhir:CodeSystem ;
  fhir:nodeRole fhir:treeRoot ;
  fhir:id [ fhir:v "AIdeviceTypeCS"] ; # 
  fhir:meta [
    ( fhir:security [
fhir:system [
fhir:v "http://terminology.hl7.org/CodeSystem/v3-ObservationValue"^^xsd:anyURI ;
fhir:l <http://terminology.hl7.org/CodeSystem/v3-ObservationValue>       ] ;
fhir:code [ fhir:v "AIAST" ] ;
fhir:display [ fhir:v "Artificial Intelligence asserted" ]     ] )
  ] ; # 
  fhir:language [ fhir:v "en"] ; # 
  fhir:text [
fhir:status [ fhir:v "generated" ] ;
fhir:div [ fhir:v "<div xmlns=\"http://www.w3.org/1999/xhtml\"><p class=\"res-header-id\"><b>Generated Narrative: CodeSystem AIdeviceTypeCS</b></p><a name=\"AIdeviceTypeCS\"> </a><a name=\"hcAIdeviceTypeCS\"> </a><div style=\"display: inline-block; background-color: #d9e0e7; padding: 6px; margin: 4px; border: 1px solid #8da1b4; border-radius: 5px; line-height: 60%\"><p style=\"margin-bottom: 0px\"/><p style=\"margin-bottom: 0px\">Security Label: Artificial Intelligence asserted (Details: ObservationValue code AIAST = 'Artificial Intelligence asserted')</p></div><p><b>Properties</b></p><p><b>This code system defines the following properties for its concepts</b></p><table class=\"grid\"><tr><td><b>Name</b></td><td><b>Code</b></td><td><b>URI</b></td><td><b>Type</b></td></tr><tr><td>Not Selectable</td><td>abstract</td><td>http://hl7.org/fhir/concept-properties#notSelectable</td><td>boolean</td></tr></table><p><b>Concepts</b></p><p>This case-sensitive code system <code>http://hl7.org/fhir/uv/aitransparency/CodeSystem/AIdeviceTypeCS</code> defines the following codes in a Grouped By hierarchy:</p><table class=\"codes\"><tr><td><b>Lvl</b></td><td style=\"white-space:nowrap\"><b>Code</b></td><td><b>Display</b></td><td><b>Definition</b></td><td><b>Not Selectable</b></td></tr><tr><td>1</td><td style=\"white-space:nowrap\">AI-By-Scenario<a name=\"AIdeviceTypeCS-AI-By-Scenario\"> </a></td><td>Classification by Application Scenario</td><td>This category classifies AI systems based on their application scenarios in the medical field.</td><td>true</td></tr><tr><td>2</td><td style=\"white-space:nowrap\">  Intelligent-Diagnosis-and-Treatment<a name=\"AIdeviceTypeCS-Intelligent-Diagnosis-and-Treatment\"> </a></td><td>Intelligent Diagnosis and Treatment</td><td>By analyzing massive volumes of medical data, these AI systems assist doctors in making more accurate diagnostic and treatment decisions.</td><td/></tr><tr><td>2</td><td style=\"white-space:nowrap\">  Medical-Image-Analysis<a name=\"AIdeviceTypeCS-Medical-Image-Analysis\"> </a></td><td>Medical Image Analysis</td><td>Leveraging deep learning technologies, these AI tools automatically identify lesion areas in medical images.</td><td/></tr><tr><td>2</td><td style=\"white-space:nowrap\">  Personalized-Treatment<a name=\"AIdeviceTypeCS-Personalized-Treatment\"> </a></td><td>Personalized Treatment</td><td>These AI systems create precise patient profiles to formulate personalized treatment plans.</td><td/></tr><tr><td>2</td><td style=\"white-space:nowrap\">  Drug-Discovery-and-Development<a name=\"AIdeviceTypeCS-Drug-Discovery-and-Development\"> </a></td><td>Drug Discovery and Development</td><td>AI in this category accelerates the screening of candidate drugs and optimizes the design of clinical trials.</td><td/></tr><tr><td>2</td><td style=\"white-space:nowrap\">  Medical-Quality-Control<a name=\"AIdeviceTypeCS-Medical-Quality-Control\"> </a></td><td>Medical Quality Control</td><td>These AI tools are used to generate standardized medical document templates and detect defects in medical documents and images.</td><td/></tr><tr><td>2</td><td style=\"white-space:nowrap\">  Patient-Services<a name=\"AIdeviceTypeCS-Patient-Services\"> </a></td><td>Patient Services</td><td>AI systems here provide patients with services such as intelligent medical guidance, symptom self-assessment, and medical consultation.</td><td/></tr><tr><td>1</td><td style=\"white-space:nowrap\">AI-By-DataType<a name=\"AIdeviceTypeCS-AI-By-DataType\"> </a></td><td>Classification by Processed Data Type</td><td>This category classifies AI systems based on the types of medical data they primarily process.</td><td>true</td></tr><tr><td>2</td><td style=\"white-space:nowrap\">  AI-for-Medical-Imaging-Data<a name=\"AIdeviceTypeCS-AI-for-Medical-Imaging-Data\"> </a></td><td>AI for Medical Imaging Data</td><td>It mainly processes medical imaging data such as X-rays, MRIs, and CT scans.</td><td/></tr><tr><td>2</td><td style=\"white-space:nowrap\">  AI-for-Physiological-Signal-Data<a name=\"AIdeviceTypeCS-AI-for-Physiological-Signal-Data\"> </a></td><td>AI for Physiological Signal Data</td><td>This type of AI deals with physiological signal data like electrocardiograms (ECG) and electroencephalograms (EEG).</td><td/></tr><tr><td>2</td><td style=\"white-space:nowrap\">  AI-for-Medical-Text-Data<a name=\"AIdeviceTypeCS-AI-for-Medical-Text-Data\"> </a></td><td>AI for Medical Text Data</td><td>It processes text data such as electronic health records (EHRs) and medical abstracts.</td><td/></tr><tr><td>1</td><td style=\"white-space:nowrap\">AI-By-Model<a name=\"AIdeviceTypeCS-AI-By-Model\"> </a></td><td>Classification by Technical Model</td><td>This category classifies AI systems based on the underlying technical models they employ.</td><td>true</td></tr><tr><td>2</td><td style=\"white-space:nowrap\">  Machine-Learning-Models<a name=\"AIdeviceTypeCS-Machine-Learning-Models\"> </a></td><td>Machine Learning Models</td><td>They include supervised learning models (e.g., Support Vector Machines (SVM), Random Forests (RF)), which can be used for disease classification and risk prediction; unsupervised learning models (e.g., K-means clustering), which can discover hidden characteristics of patient subgroups; and reinforcement learning models, which can be applied in dynamic treatment plan management.</td><td/></tr><tr><td>2</td><td style=\"white-space:nowrap\">  Deep-Learning-Models<a name=\"AIdeviceTypeCS-Deep-Learning-Models\"> </a></td><td>Deep Learning Models</td><td>Examples include Convolutional Neural Networks (CNNs), which perform excellently in medical image analysis; Recurrent Neural Networks (RNNs) and their variant LSTMs, which are suitable for processing time-series physiological signal data; Generative Adversarial Networks (GANs), which can be used to synthesize training data and alleviate the scarcity of medical data; and Transformer models, which are widely used in multiple tasks such as medical imaging, text analysis, and physiological signal prediction.</td><td/></tr><tr><td>2</td><td style=\"white-space:nowrap\">  Large-Language-Models<a name=\"AIdeviceTypeCS-Large-Language-Models\"> </a></td><td>Large Language Models</td><td>These models, such as GPT-4 and PaLM, are trained on massive text datasets and can perform various natural language processing tasks, including medical text understanding, generation, and question answering.</td><td/></tr><tr><td>2</td><td style=\"white-space:nowrap\">  Hybrid-Models<a name=\"AIdeviceTypeCS-Hybrid-Models\"> </a></td><td>Hybrid Models</td><td>These models combine multiple AI techniques to leverage their respective strengths. For instance, combining CNNs and RNNs can effectively process medical image sequences; integrating machine learning and deep learning models can enhance disease prediction accuracy; and combining rule-based systems with machine learning can improve interpretability and reliability in clinical decision support.</td><td/></tr><tr><td>1</td><td style=\"white-space:nowrap\">Artificial-Intelligence<a name=\"AIdeviceTypeCS-Artificial-Intelligence\"> </a></td><td>All kinds of Artificial Intelligence</td><td>Any type of Artificial Intelligence system, undifferentiated.</td><td/></tr></table></div>"^^rdf:XMLLiteral ]
  ] ; # 
  fhir:extension ( [
fhir:url [
fhir:v "http://hl7.org/fhir/StructureDefinition/structuredefinition-wg"^^xsd:anyURI ;
fhir:l <http://hl7.org/fhir/StructureDefinition/structuredefinition-wg>     ] ;
fhir:value [
a fhir:Code ;
fhir:v "ehr"     ]
  ] [
fhir:url [
fhir:v "http://hl7.org/fhir/StructureDefinition/structuredefinition-fmm"^^xsd:anyURI ;
fhir:l <http://hl7.org/fhir/StructureDefinition/structuredefinition-fmm>     ] ;
fhir:value [
a fhir:Integer ;
fhir:v 2 ;
      ( fhir:extension [
fhir:url [
fhir:v "http://hl7.org/fhir/StructureDefinition/structuredefinition-conformance-derivedFrom"^^xsd:anyURI ;
fhir:l <http://hl7.org/fhir/StructureDefinition/structuredefinition-conformance-derivedFrom>         ] ;
fhir:value [
a fhir:Canonical ;
fhir:v "http://hl7.org/fhir/uv/aitransparency/ImplementationGuide/hl7.fhir.uv.aitransparency"^^xsd:anyURI ;
fhir:l <http://hl7.org/fhir/uv/aitransparency/ImplementationGuide/hl7.fhir.uv.aitransparency>         ]       ] )     ]
  ] [
fhir:url [
fhir:v "http://hl7.org/fhir/StructureDefinition/structuredefinition-standards-status"^^xsd:anyURI ;
fhir:l <http://hl7.org/fhir/StructureDefinition/structuredefinition-standards-status>     ] ;
fhir:value [
a fhir:Code ;
fhir:v "trial-use" ;
      ( fhir:extension [
fhir:url [
fhir:v "http://hl7.org/fhir/StructureDefinition/structuredefinition-conformance-derivedFrom"^^xsd:anyURI ;
fhir:l <http://hl7.org/fhir/StructureDefinition/structuredefinition-conformance-derivedFrom>         ] ;
fhir:value [
a fhir:Canonical ;
fhir:v "http://hl7.org/fhir/uv/aitransparency/ImplementationGuide/hl7.fhir.uv.aitransparency"^^xsd:anyURI ;
fhir:l <http://hl7.org/fhir/uv/aitransparency/ImplementationGuide/hl7.fhir.uv.aitransparency>         ]       ] )     ]
  ] ) ; # 
  fhir:url [
fhir:v "http://hl7.org/fhir/uv/aitransparency/CodeSystem/AIdeviceTypeCS"^^xsd:anyURI ;
fhir:l <http://hl7.org/fhir/uv/aitransparency/CodeSystem/AIdeviceTypeCS>
  ] ; # 
  fhir:version [ fhir:v "1.0.0-ballot"] ; # 
  fhir:name [ fhir:v "AIdeviceTypeCS"] ; # 
  fhir:title [ fhir:v "Device type for Artificial Intelligence"] ; # 
  fhir:status [ fhir:v "active"] ; # 
  fhir:experimental [ fhir:v false] ; # 
  fhir:date [ fhir:v "2026-04-21T21:00:23+00:00"^^xsd:dateTime] ; # 
  fhir:publisher [ fhir:v "HL7 International / Electronic Health Records"] ; # 
  fhir:contact ( [
fhir:name [ fhir:v "HL7 International / Electronic Health Records" ] ;
    ( fhir:telecom [
fhir:system [ fhir:v "url" ] ;
fhir:value [ fhir:v "http://www.hl7.org/Special/committees/ehr" ]     ] [
fhir:system [ fhir:v "email" ] ;
fhir:value [ fhir:v "ehr@lists.hl7.org" ]     ] )
  ] ) ; # 
  fhir:description [ fhir:v "This CodeSystem contains codes for the Device.type that indicate that the Device is an AI. The codes here were created by AI."] ; # 
  fhir:jurisdiction ( [
    ( fhir:coding [
fhir:system [
fhir:v "http://unstats.un.org/unsd/methods/m49/m49.htm"^^xsd:anyURI ;
fhir:l <http://unstats.un.org/unsd/methods/m49/m49.htm>       ] ;
fhir:code [ fhir:v "001" ]     ] )
  ] ) ; # 
  fhir:caseSensitive [ fhir:v true] ; # 
  fhir:hierarchyMeaning [ fhir:v "grouped-by"] ; # 
  fhir:content [ fhir:v "complete"] ; # 
  fhir:count [ fhir:v "17"^^xsd:nonNegativeInteger] ; # 
  fhir:property ( [
fhir:code [ fhir:v "abstract" ] ;
fhir:uri [
fhir:v "http://hl7.org/fhir/concept-properties#notSelectable"^^xsd:anyURI ;
fhir:l <http://hl7.org/fhir/concept-properties#notSelectable>     ] ;
fhir:type [ fhir:v "boolean" ]
  ] ) ; # 
  fhir:concept ( [
fhir:code [ fhir:v "AI-By-Scenario" ] ;
fhir:display [ fhir:v "Classification by Application Scenario" ] ;
fhir:definition [ fhir:v "This category classifies AI systems based on their application scenarios in the medical field." ] ;
    ( fhir:property [
fhir:code [ fhir:v "abstract" ] ;
fhir:value [
a fhir:Boolean ;
fhir:v true       ]     ] ) ;
    ( fhir:concept [
fhir:code [ fhir:v "Intelligent-Diagnosis-and-Treatment" ] ;
fhir:display [ fhir:v "Intelligent Diagnosis and Treatment" ] ;
fhir:definition [ fhir:v "By analyzing massive volumes of medical data, these AI systems assist doctors in making more accurate diagnostic and treatment decisions." ]     ] [
fhir:code [ fhir:v "Medical-Image-Analysis" ] ;
fhir:display [ fhir:v "Medical Image Analysis" ] ;
fhir:definition [ fhir:v "Leveraging deep learning technologies, these AI tools automatically identify lesion areas in medical images." ]     ] [
fhir:code [ fhir:v "Personalized-Treatment" ] ;
fhir:display [ fhir:v "Personalized Treatment" ] ;
fhir:definition [ fhir:v "These AI systems create precise patient profiles to formulate personalized treatment plans." ]     ] [
fhir:code [ fhir:v "Drug-Discovery-and-Development" ] ;
fhir:display [ fhir:v "Drug Discovery and Development" ] ;
fhir:definition [ fhir:v "AI in this category accelerates the screening of candidate drugs and optimizes the design of clinical trials." ]     ] [
fhir:code [ fhir:v "Medical-Quality-Control" ] ;
fhir:display [ fhir:v "Medical Quality Control" ] ;
fhir:definition [ fhir:v "These AI tools are used to generate standardized medical document templates and detect defects in medical documents and images." ]     ] [
fhir:code [ fhir:v "Patient-Services" ] ;
fhir:display [ fhir:v "Patient Services" ] ;
fhir:definition [ fhir:v "AI systems here provide patients with services such as intelligent medical guidance, symptom self-assessment, and medical consultation." ]     ] )
  ] [
fhir:code [ fhir:v "AI-By-DataType" ] ;
fhir:display [ fhir:v "Classification by Processed Data Type" ] ;
fhir:definition [ fhir:v "This category classifies AI systems based on the types of medical data they primarily process." ] ;
    ( fhir:property [
fhir:code [ fhir:v "abstract" ] ;
fhir:value [
a fhir:Boolean ;
fhir:v true       ]     ] ) ;
    ( fhir:concept [
fhir:code [ fhir:v "AI-for-Medical-Imaging-Data" ] ;
fhir:display [ fhir:v "AI for Medical Imaging Data" ] ;
fhir:definition [ fhir:v "It mainly processes medical imaging data such as X-rays, MRIs, and CT scans." ]     ] [
fhir:code [ fhir:v "AI-for-Physiological-Signal-Data" ] ;
fhir:display [ fhir:v "AI for Physiological Signal Data" ] ;
fhir:definition [ fhir:v "This type of AI deals with physiological signal data like electrocardiograms (ECG) and electroencephalograms (EEG)." ]     ] [
fhir:code [ fhir:v "AI-for-Medical-Text-Data" ] ;
fhir:display [ fhir:v "AI for Medical Text Data" ] ;
fhir:definition [ fhir:v "It processes text data such as electronic health records (EHRs) and medical abstracts." ]     ] )
  ] [
fhir:code [ fhir:v "AI-By-Model" ] ;
fhir:display [ fhir:v "Classification by Technical Model" ] ;
fhir:definition [ fhir:v "This category classifies AI systems based on the underlying technical models they employ." ] ;
    ( fhir:property [
fhir:code [ fhir:v "abstract" ] ;
fhir:value [
a fhir:Boolean ;
fhir:v true       ]     ] ) ;
    ( fhir:concept [
fhir:code [ fhir:v "Machine-Learning-Models" ] ;
fhir:display [ fhir:v "Machine Learning Models" ] ;
fhir:definition [ fhir:v "They include supervised learning models (e.g., Support Vector Machines (SVM), Random Forests (RF)), which can be used for disease classification and risk prediction; unsupervised learning models (e.g., K-means clustering), which can discover hidden characteristics of patient subgroups; and reinforcement learning models, which can be applied in dynamic treatment plan management." ]     ] [
fhir:code [ fhir:v "Deep-Learning-Models" ] ;
fhir:display [ fhir:v "Deep Learning Models" ] ;
fhir:definition [ fhir:v "Examples include Convolutional Neural Networks (CNNs), which perform excellently in medical image analysis; Recurrent Neural Networks (RNNs) and their variant LSTMs, which are suitable for processing time-series physiological signal data; Generative Adversarial Networks (GANs), which can be used to synthesize training data and alleviate the scarcity of medical data; and Transformer models, which are widely used in multiple tasks such as medical imaging, text analysis, and physiological signal prediction." ]     ] [
fhir:code [ fhir:v "Large-Language-Models" ] ;
fhir:display [ fhir:v "Large Language Models" ] ;
fhir:definition [ fhir:v "These models, such as GPT-4 and PaLM, are trained on massive text datasets and can perform various natural language processing tasks, including medical text understanding, generation, and question answering." ]     ] [
fhir:code [ fhir:v "Hybrid-Models" ] ;
fhir:display [ fhir:v "Hybrid Models" ] ;
fhir:definition [ fhir:v "These models combine multiple AI techniques to leverage their respective strengths. For instance, combining CNNs and RNNs can effectively process medical image sequences; integrating machine learning and deep learning models can enhance disease prediction accuracy; and combining rule-based systems with machine learning can improve interpretability and reliability in clinical decision support." ]     ] )
  ] [
fhir:code [ fhir:v "Artificial-Intelligence" ] ;
fhir:display [ fhir:v "All kinds of Artificial Intelligence" ] ;
fhir:definition [ fhir:v "Any type of Artificial Intelligence system, undifferentiated." ]
  ] ) . #