Clinical Evaluation Report
Evidence and evaluation of clinical AI performance.
Version: 1.0
Date: 9 August 2026
Issued by: Respocare (Pty) Ltd
Location: Johannesburg, South Africa
1. Executive Summary
Respocare Connect AI is a clinical documentation and information support platform designed to assist healthcare professionals with the management, summarisation, and organisation of patient clinical records.
The platform incorporates artificial intelligence technologies to support clinicians in processing large volumes of medical documentation while maintaining full clinician oversight.
This Clinical Evaluation Report presents the methodology and findings of the Respocare clinical evaluation programme, which was conducted to assess the platform's performance, safety characteristics, and suitability for integration into real clinical workflows.
The evaluation focuses on:
- accuracy of information extraction
- reliability of clinical documentation summarisation
- behaviour under incomplete clinical records
- clinician oversight and control
- safety characteristics of AI outputs
The findings of this evaluation indicate that the Respocare Connect AI system performs effectively as an assistive clinical information processing tool when used under clinician supervision.
The system is designed to support healthcare professionals while preserving the authority of the treating clinician for all clinical decisions.
2. System Overview
Respocare Connect AI is a software platform developed by Respocare (Pty) Ltd to assist clinicians with clinical documentation and information retrieval.
The system provides capabilities including:
- analysis of uploaded clinical documents
- summarisation of patient records
- generation of structured clinical documentation drafts
- retrieval of relevant patient information through conversational interaction
The system operates within a controlled platform environment in which clinicians upload and manage patient records.
Artificial intelligence is used to interpret clinical documentation and generate structured outputs that assist clinicians in navigating complex patient records.
All outputs generated by the system require clinician review before use in clinical documentation.
3. Intended Use
Respocare Connect AI is intended to assist healthcare professionals in the organisation and interpretation of clinical documentation.
The system is designed to support tasks such as:
- summarising patient medical histories
- retrieving relevant clinical information
- generating structured draft clinical documentation
- assisting clinicians in navigating longitudinal patient records
The system is intended to be used by licensed healthcare professionals within clinical environments.
Respocare Connect AI does not perform autonomous diagnosis, treatment planning, or clinical decision-making.
Clinical judgement remains the responsibility of the treating clinician.
4. Evaluation Objectives
The clinical evaluation programme was designed to assess the following aspects of system performance:
Documentation processing capability — Evaluation of the system's ability to interpret and structure clinical documentation.
Information retrieval accuracy — Assessment of how effectively the system can locate relevant information within patient records.
Longitudinal record understanding — Evaluation of the system's ability to maintain context across multiple clinical documents and patient encounters.
Safety behaviour — Assessment of the system's ability to acknowledge uncertainty and avoid unsupported conclusions.
Workflow usability — Evaluation of how naturally the system integrates into real clinical workflows.
5. Evaluation Methodology
The Respocare clinical evaluation programme used a structured methodology designed to simulate realistic clinical workflows.
The evaluation consisted of multiple components.
5.1 Synthetic Clinical Dataset Construction
Structured patient datasets were created to simulate realistic clinical scenarios.
These datasets contained multiple clinical records representing patient encounters over time.
Document types included:
- initial consultation notes
- follow-up consultation documentation
- laboratory test results
- imaging reports
- specialist referral letters
- diagnostic summaries
- treatment updates
The datasets were designed to reflect the fragmented and evolving nature of real clinical records.
5.2 Longitudinal Patient Journey Simulation
Each dataset simulated a patient journey spanning multiple visits and clinical events.
This allowed evaluation of the system's ability to:
- track evolving patient information
- maintain contextual understanding across visits
- interpret clinical changes over time
5.3 Clinical Prompt Testing
The system was tested using prompts designed to replicate real clinician interactions.
Prompt categories included:
- requests for patient summaries
- extraction of laboratory results
- clarification of diagnoses
- identification of clinical trends
- preparation of documentation drafts
These prompts were intentionally simple and reflective of routine clinical tasks.
5.4 Workflow Simulation
The evaluation assessed the complete workflow of the system, including:
- uploading clinical documents
- AI analysis of documents
- generation of summaries and structured information
- clinician review and approval of outputs
This approach ensured that both system performance and usability were evaluated.
6. Human Oversight Model
Respocare Connect AI is designed around a human-in-the-loop architecture.
The evaluation confirmed that:
- clinicians review AI-generated outputs
- clinicians approve documentation before it becomes part of the patient record
- clinical authority remains with the healthcare professional
This design ensures that AI functions as a support tool rather than an autonomous system.
7. Safety Evaluation
Safety behaviour was evaluated by testing how the system responds when clinical information is incomplete or ambiguous.
The system demonstrated the ability to:
- acknowledge uncertainty
- avoid unsupported clinical conclusions
- rely only on available patient record information
These behaviours are important safeguards when AI systems are used in healthcare environments.
8. Performance Findings
Across evaluation scenarios, the system demonstrated strong performance in the following areas:
Clinical information extraction — The system successfully identified and summarised key clinical data within patient records.
Documentation structuring — The system produced structured draft documentation that clinicians could review and refine.
Retrieval of relevant information — The system was able to locate relevant clinical details within patient records in response to user prompts.
Workflow integration — The system supported clinician workflows without introducing disruptive processes.
9. Risk Considerations
As with any AI system, potential risks exist.
These include:
- misinterpretation of incomplete clinical records
- omission of relevant information
- over-reliance on automated summaries
These risks are mitigated through the human-in-the-loop design of the system and clinician verification of outputs.
10. Clinical Responsibility
Respocare Connect AI does not replace the clinical judgement of healthcare professionals.
Clinicians remain responsible for:
- interpreting patient information
- confirming the accuracy of documentation
- making clinical decisions regarding patient care
The platform is designed to assist clinicians in managing information rather than making clinical decisions.
11. Ethical Considerations
The development of Respocare Connect AI follows responsible AI development principles.
These include:
- transparency of system behaviour
- preservation of clinician authority
- protection of patient data
- responsible handling of uncertainty
Respocare believes that AI should augment clinical practice while maintaining strong safeguards for patient safety.
12. Limitations
The clinical evaluation programme was conducted using simulated patient datasets designed to replicate realistic clinical scenarios.
While these simulations provide valuable insights into system behaviour, further evaluation in live clinical environments may provide additional data on system performance under real-world conditions.
Respocare plans to continue evaluating system performance as the platform is adopted in clinical settings.
13. Conclusion
The clinical evaluation programme demonstrates that Respocare Connect AI performs effectively as an assistive clinical documentation and information support platform.
The system supports clinicians in navigating complex patient records while maintaining safeguards that preserve clinical authority and patient safety.
Respocare Connect AI is designed to enhance clinical workflows by reducing administrative burden and improving access to relevant patient information.
The platform is intended to function as a clinician-controlled support tool within healthcare environments.
14. Future Development
Respocare intends to continue refining the Respocare Connect AI platform through ongoing system improvements and clinical feedback.
Future development efforts may include:
- enhanced clinical summarisation capabilities
- expanded document analysis functionality
- further evaluation in real clinical environments
Respocare remains committed to responsible development of AI technologies in healthcare.