The solution was integrated with the client's existing medical systems, so that the patient file - medical history, chronic conditions, previous investigations, lab results, and past interactions - is aggregated automatically, without the doctor having to switch between applications.
Before the consultation, the doctor receives a structured summary of the patient, together with a search function across the entire file, for quickly identifying a specific element of interest or a relevant prior finding.
During the consultation, the AI agents process the context in real time and provide the doctor with contextual guidance, based on the patient file and on the medical data sources connected to the platform.
At the end of the consultation, the platform automatically drafts a structured clinical report. It remains a draft until the doctor reviews and validates it.
Personal data is anonymized locally, before any call to the AI models, and every processing operation is logged for auditing purposes.
Results
Alteus Medical gives REGINA MARIA doctors a complete picture of the patient from the first seconds of the consultation and removes most of the documentation effort that used to follow it.
During the first 5 months, the platform was in active use across the network: 1,647 doctors, 156,000 medical summaries generated ahead of consultations, 3,500 structured medical reports drafted after visits, and 3,700 medical chat questions answered from the patient file.
The time needed to prepare a consultation is significantly reduced, as the patient file no longer has to be reconstructed manually from several systems.
The medical report is drafted automatically and only validated by the doctor, which reduces the administrative time after each consultation.
The quality of documentation becomes consistent across specialties and locations, supporting a uniform standard of care.
All processing takes place within a governed framework, with local anonymization and full audit trails, in line with the requirements of the healthcare industry.
Technologies
Alteus Medical uses Azure OpenAI models hosted in the EU. It anonymizes patient data locally and applies intent detection and clinical guardrails to each interaction. An end-to-end audit trail records how every request is processed.
For each clinician query, an agent orchestrator divides the work among specialized agents. Using Recursive Language Model (RLM) techniques, they can examine the patient's complete longitudinal record. The orchestrator then combines their findings into one response. DSPy improves prompt behavior against defined evaluation criteria, while MLflow monitors quality and deviations in agent behavior.
Smart model routing selects a suitable model for each task, and Celery workers handle background processing. Caching reuses previous results so the same work does not trigger another LLM call. Usage controls help keep response times and the cost per consultation predictable across the network.
Product note: The implementation described above was tailored to REGINA MARIA's specific systems and scale. In its standard form, Alteus Medical is available as part of the Alteus.ai suite, giving other healthcare providers access to the same clinical AI companion - patient file aggregation, real-time consultation support, and automated report drafting - on a governed, audit-ready foundation. Read more: alteus.ai
RO
EN
