Aller au contenu principal

AI Clinical Decision Support

Category: Agent
Node id: ai-cds

Standards: HIPAA

Provides AI-powered clinical decision support including drug interactions, differential diagnosis, treatment guidelines, and more

What it does​

  • Evidence-based clinical decision support
  • Drug-drug interaction analysis
  • Differential diagnosis generation ranked by probability
  • Treatment guideline recommendations (AHA, ACC, NCCN, ADA)
  • Preventive care screening per USPSTF guidelines
  • Sepsis screening with SIRS, qSOFA, and NEWS-2 scoring
  • Antibiotic stewardship recommendations

Use cases​

  • Generating differential diagnoses from patient presentations
  • Checking medication lists for drug interactions and contraindications
  • Recommending evidence-based treatment plans per clinical guidelines
  • Identifying overdue preventive care screenings
  • Early sepsis detection and intervention recommendations

Runtime​

  • Performance tier: variable

Input ports​

NameTypeRequiredDescription
patient_datajsonNoPatient data as FHIR or structured JSON
clinical_contexttextNoClinical context narrative
querytextNoSpecific clinical question

Output ports​

NameTypeRequiredDescription
recommendationstextNoAI-generated clinical recommendations
cds_typetextNoType of CDS performed
modeltextNoModel used
disclaimertextNoClinical disclaimer
http_statusnumberNoHTTP status code from LLM

Settings​

Basic​

SettingNameTypeRequiredDefaultAllowed valuesDescription
LLM Endpointllm_endpointstringYesBase URL of the LLM API
ModelmodelstringNogpt-4oModel name to use
CDS Typecds_typeenumNogeneralgeneral, drug_interaction, diagnosis_differential, treatment_guideline, preventive_care, sepsis_screening, antibiotic_stewardshipType of clinical decision support

Auth​

SettingNameTypeRequiredDefaultAllowed valuesDescription
API Keyapi_keysecretNoAPI key for the LLM provider

Errors​

CodeDescription
ENDPOINT_REQUIREDllm_endpoint is not configured
INPUT_REQUIREDEither patient_data or clinical_context must be provided
CDS_REQUEST_FAILEDLLM API call failed during clinical decision support
HTTP_ERRORLLM endpoint returned an HTTP error status

Runtime notes​

  • Output is flat (not nested under result) — recommendations, cds_type, model, disclaimer, and http_status are top-level keys
  • The disclaimer field is always present and must not be removed when wiring to downstream display nodes
  • At least one of patient_data or clinical_context must be provided; both can be combined for richer analysis
  • cds_type in the output reflects the configured type, useful for routing in multi-CDS pipelines

See Build any workflow for how to put nodes together and Ports, types and data flow for how port types connect.

Quelque chose n'est pas clair ? Envoyez vos commentaires sur la documentation par courriel.