Device type for Artificial Intelligence - AI Transparency on FHIR v1.0.0-current

AI Transparency on FHIR
1.0.0-current - STU1 Ballot International flag

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-current 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

CodeSystem: Device type for Artificial Intelligence

Official URL: http://hl7.org/fhir/uv/aitransparency/CodeSystem/AIdeviceTypeCS	Version: 1.0.0-current	Highlight text changes?
Standards status: Trial-use	Maturity Level: 2	Computable Name: AIdeviceTypeCS
Other Identifiers: OID:2.16.840.1.113883.4.642.40.79.16.1

This CodeSystem contains codes for the Device.type that indicate that the Device is an AI. The codes here were created by AI.

This Code system is referenced in the definition of the following value sets:

AIdeviceTypeVS

Security Label: Artificial Intelligence asserted (Details: ObservationValue code AIAST = 'Artificial Intelligence asserted')

Properties

This code system defines the following properties for its concepts

Name

Code

URI

Type

Not Selectable

abstract

http://hl7.org/fhir/concept-properties#notSelectable

boolean

Concepts

This case-sensitive code system http://hl7.org/fhir/uv/aitransparency/CodeSystem/AIdeviceTypeCS defines the following codes in a Grouped By hierarchy:

Lvl	Code	Display	Definition	Not Selectable
1	Artificial-Intelligence	All kinds of Artificial Intelligence	Any type of Artificial Intelligence system, undifferentiated.
2	AI-By-Model	Classification by Technical Model	This category classifies AI systems based on the underlying technical models they employ.	true
3	Rule-Based-Systems	Rule-Based Systems	Systems that apply explicitly authored rules, logic, knowledge representations, or inference procedures to input data.
3	Machine-Learning-Models	Machine Learning Models	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.
4	Deep-Learning-Models	Deep Learning Models	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.
5	Foundation-Models	Foundation Models	Deep learning models trained on broad data at scale that can be adapted to a range of downstream tasks.	true
6	Large-Language-Models	Large Language Models	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.
6	Vision-Foundation-Models	Vision Foundation Models	Foundation models trained primarily to process and represent visual data for adaptation to vision tasks.
6	Vision-Language-Models	Vision-Language Models	Foundation models that jointly process visual and language data.
3	Hybrid-Models	Hybrid Models	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.
2	AI-By-Behavior	Classification by Output Behavior	Groups AI systems by whether identical inputs and execution conditions are expected to produce identical outputs.	true
3	Deterministic-AI	Deterministic AI	An AI system that produces the same output for the same input when its model, configuration, state, and execution environment are unchanged.
3	Non-Deterministic-AI	Non-deterministic AI	An AI system whose output may vary for the same input and execution conditions, for example because it uses probabilistic sampling or stochastic processing.
2	AI-By-Scenario	Classification by Application Scenario	This category classifies AI systems based on their application scenarios in the medical field.	true
3	Intelligent-Diagnosis-and-Treatment	Intelligent Diagnosis and Treatment	By analyzing massive volumes of medical data, these AI systems assist doctors in making more accurate diagnostic and treatment decisions.
3	Medical-Image-Analysis	Medical Image Analysis	Leveraging deep learning technologies, these AI tools automatically identify lesion areas in medical images.
3	Personalized-Treatment	Personalized Treatment	These AI systems create precise patient profiles to formulate personalized treatment plans.
3	Clinical-Monitoring-and-Early-Warning	Clinical Monitoring and Early Warning	Monitors longitudinal or real-time patient data to identify deterioration, adverse events, or other conditions requiring attention.
3	Clinical-Documentation-and-Workflow	Clinical Documentation and Workflow	Creates, summarizes, extracts, codes, routes, or quality-checks clinical and administrative information to support healthcare workflows.
3	Drug-Discovery-and-Development	Drug Discovery and Development	AI in this category accelerates the screening of candidate drugs and optimizes the design of clinical trials.
3	Medical-Quality-Control	Medical Quality Control	These AI tools are used to generate standardized medical document templates and detect defects in medical documents and images.
3	Patient-Services	Patient Services	AI systems here provide patients with services such as intelligent medical guidance, symptom self-assessment, and medical consultation.
3	Population-Health-and-Operations	Population Health and Healthcare Operations	Supports population stratification, public health, resource planning, scheduling, logistics, or other healthcare operational activities.
2	AI-By-DataType	Classification by Processed Data Type	This category classifies AI systems based on the types of medical data they primarily process.	true
3	AI-for-Medical-Imaging-Data	AI for Medical Imaging Data	It mainly processes medical imaging data such as X-rays, MRIs, and CT scans.
3	AI-for-Physiological-Signal-Data	AI for Physiological Signal Data	This type of AI deals with physiological signal data like electrocardiograms (ECG) and electroencephalograms (EEG).
3	AI-for-Medical-Text-Data	AI for Medical Text Data	It processes text data such as electronic health records (EHRs) and medical abstracts.
3	AI-for-Structured-Clinical-Data	Structured Clinical Data	Processes coded, tabular, or relational health data such as diagnoses, medications, laboratory results, observations, claims, or FHIR resources.
3	AI-for-Genomic-and-Omics-Data	Genomic and Omics Data	Processes genomic, transcriptomic, proteomic, metabolomic, or related molecular data.
3	AI-for-Audio-Data	Audio Data	Processes speech, heart sounds, respiratory sounds, or other clinically relevant audio.
3	AI-for-Video-Data	Video Data	Processes temporal image sequences such as endoscopy, ultrasound cine loops, gait recordings, or procedure video.
3	AI-for-Multimodal-Data	Multimodal Data	Jointly processes two or more data modalities, such as images and text or physiological signals and structured clinical data.