Skip to main content
SofIA’s clinical chat provides an intelligent conversational assistant designed specifically to support medical practice during consultations. It uses SofIA’s cognitive framework to respond to clinical queries with contextualized information.

Main features

Consultation context

The chat has access to the current consultation context:
  • Real-time transcription: Information from the ongoing doctor-patient conversation
  • Patient data: Information provided through the patientData property
  • Current documentation: Notes and findings from the present consultation

Assistant capabilities

Contextualized responses

Provides specific medical information based on the patient’s context and current consultation.

Clinical reasoning

Uses specialized agents to assist in differential diagnosis, coding, and clinical verification.

Consultation integration

Knows the current context: symptoms, examinations, and findings from the ongoing session.

Medical sources

Accesses medical knowledge sources configured in the system.

Basic configuration

Chat enabled by default

Clinical chat is available by default when SofIA SDK is configured with the basic required properties.

Chat-only mode

For cases where only the conversational assistant is required without audio transcription, the isonlychat property can be used:
  • Purpose: Quick consultations without audio processing
  • Simplified interface: Only chat functionality
  • Lower resource consumption: No audio processing

Typical use cases

Differential diagnosis queries

The chat can assist when the professional needs to explore possible diagnoses based on symptoms and findings observed during the consultation.

Treatment verification

Provides information about therapeutic options considering the patient’s specific context, their background, and current medication.

Coding queries

Assists in identifying appropriate medical codes (ICD-10, SNOMED CT, LOINC) based on diagnoses and procedures performed.

Interaction information

Helps identify possible drug interactions or contraindications considering the patient’s current medication.

Specialized agents

Clinical reasoning agent

Processes medical queries using specialized knowledge in clinical domains to provide relevant and contextualized responses.

Coding agent

Specialized in standard medical terminologies, assists in identifying and verifying appropriate medical codes.

Review agent

Validates the clinical coherence of responses and reduces the possibility of incorrect or inconsistent information.

Best practices for use

Effective query formulation

  • Be specific: Include relevant details of the current case
  • Contextualize: Mention pertinent patient background
  • Focus the question: Formulate concrete and targeted queries
  • Consider context: Leverage that the system knows the current consultation

Information verification

  • Independent validation: Verify important recommendations with institutional protocols
  • Medical criteria: Maintain clinical judgment as the final decision
  • Additional sources: Consult other sources when necessary
  • Documentation: Record assistant use when appropriate

Privacy management

  • Minimum necessary information: Only include data relevant to the consultation
  • Confidentiality: Respect patient privacy in interactions
  • Contextual data: Use information already available in patientData

Important limitations

Assistant role

  • Does not replace clinical judgment: It is an informational support tool
  • Does not make diagnoses: Provides information to support decision-making
  • Context dependent: Quality depends on available information
  • Requires validation: Suggestions must be evaluated by the professional

Technical considerations

  • Source availability: Responses depend on configured knowledge sources
  • Context limitations: May not have access to all relevant clinical information
  • Specialization: Better performance in areas with more specific sources
  • Updates: Information may not reflect the most recent developments

Professional responsibilities

  • Final clinical decision: The professional maintains complete responsibility for medical decisions
  • Critical validation: Verify important information with institutional sources
  • Complete context: Consider factors the system may not know
  • Professional ethics: Maintain ethical standards of medical practice

Workflow integration

During consultation

The chat can be used naturally during patient care for quick queries that support real-time decision-making.

Post-consultation documentation

Chat responses can inform clinical documentation, but always under professional supervision and validation.

Case follow-up

The assistant can help in continuity of care by providing context from previous consultations when available.

Quality and safety

Implemented safety measures

  • Query filtering: The system identifies and filters inappropriate queries
  • Confidence levels: Responses include certainty indicators
  • Automatic review: Validation by specialized agents before providing responses
  • Limitation alerts: Notification when available information is insufficient

Quality indicators

  • Contextual relevance: Responses adapt to the specific context of the case
  • Clinical coherence: Verification of medical logic in responses
  • Reliable sources: Use of established medical literature and clinical guidelines
  • Continuous updating: Constant improvement based on feedback and new sources

Next steps