Private healthcare
How an AI Assistant Helped Private Physicians Build More Complete Patient Medical Histories
Syntropic Ops developed an AI-assisted medical intake system for a service used by private physicians. Before an appointment, patients could speak or chat with an AI assistant that collected their medical history and processed supporting documents, giving the physician a structured summary, key highlights, the complete interview, and the original recording.
- Voice AI agents
- Text AI agents
- Document processing
- Notification systems
Quick Overview
- Industry
- Private healthcare
- Business area
- Patient intake and clinical information management
- Automation type
- AI-assisted medical interview
- Services
- Workflow design, AI agent configuration, document processing, and summary generation
- Tools
- Voice and text AI agents, document-processing technology, and notification systems
- Key result
- More complete medical histories before appointments and easier access to relevant patient information
The Challenge
During an initial consultation, the physician had to collect the patient's medical history through a live interview.
This process depended heavily on the patient's ability to remember, organize, and communicate relevant medical information within a limited period. In some cases, answers were vague, important details were omitted, or supporting medical documents were not presented in a structured way.
This created several challenges:
- Physicians had to spend a significant part of the consultation reconstructing the patient's history.
- Relevant information sometimes emerged too late in the conversation.
- Patients did not always have enough time or confidence to explain their situation fully.
- Blood test results, medical examinations, and reports from other physicians had to be reviewed separately.
- The quality and depth of the information varied from one patient to another.
The goal was not simply to automate a questionnaire. The real business need was to give physicians a more complete clinical context before the appointment without replacing their professional judgment.
The Previous Process
Before the solution was implemented, the process generally worked as follows:
- The patient attended an initial appointment.
- The physician began the consultation without a previously compiled medical history.
- The patient explained their symptoms, medical background, treatments, and concerns.
- The physician asked additional questions to clarify or complete the information.
- Medical tests and supporting documents were reviewed during or after the consultation.
- The physician organized the information and determined the next steps.
This process could work well, but it depended on the patient's memory, the time available, and the physician's ability to identify missing information quickly.
The Objective
The project aimed to improve the collection of patient information before the consultation.
The main objectives were to:
- Give patients more space to explain their medical history.
- Reduce the time physicians spent asking broad intake questions.
- Allow consultations to focus on the most relevant follow-up questions.
- Incorporate information from medical examinations and reports.
- Present patient history in a clear and structured format.
- Improve the patient experience.
- Support physicians without replacing their evaluation or diagnosis.
The Solution
Syntropic Ops created a system of AI agents designed to conduct medical interviews through voice or text.
The workflow began when a patient scheduled an appointment. After booking, the patient received a link to access a virtual call with the AI assistant.
During the interaction, the agent asked questions about the patient's current medical condition, previous health issues, and other information relevant to the upcoming consultation. The agent's instructions were configured using the information and criteria provided by the physicians.
The conversation was not limited to a rigid questionnaire. The agent used information gathered during the interview to determine which follow-up questions were relevant.
Patients could also provide supporting documents, including:
- Blood test results.
- Medical examination results.
- Reports from other healthcare professionals.
- Other documents related to their medical history.
At the end of the process, the system organized the information and notified the physician.
The physician could review:
- A general overview of the interview.
- The most relevant medical history highlights.
- The complete details of the patient's answers.
- The video recording of the interview.
- Information extracted from the supporting documents.
This allowed the physician to begin the consultation with a broader understanding of the case and focus the conversation on confirming information, exploring important issues, and asking more specific questions.
Automation Workflow
- 01
Trigger
The patient schedules an appointment with the physician.
- 02
Access Link
The system sends the patient a link to complete a virtual interview with the AI assistant.
- 03
Interview
The patient communicates with the agent by voice or text and explains their current condition, background, and medical history.
- 04
Contextual Follow-Up
The agent uses information collected during the conversation to select appropriate follow-up questions and explore relevant areas in greater depth.
- 05
Document Processing
The system receives and processes medical examinations, laboratory results, and reports submitted by the patient.
- 06
Information Organization
The responses and documents are converted into a structured medical history containing a summary and a section with key highlights.
- 07
Notification
The physician receives a notification when the information is ready for review.
- 08
Physician Review
The physician can begin with the overview and, when necessary, access the full interview, recording, and supporting documents.
- 09
Human Oversight
The physician remains responsible for evaluating the information, confirming relevant details, and making any diagnosis.
Technical Challenges
Managing the Agent's Memory
One of the main technical challenges was managing the assistant's memory during long interviews. The agent needed to remember information the patient had already provided, avoid unnecessary repetition, and identify which areas still required further exploration. Poor memory management could create two main problems: repeating questions the patient had already answered, or losing important details mentioned earlier in the conversation. To address this, the instructions and use of conversational context were optimized so the agent could maintain continuity and ask relevant questions based on the criteria defined by the physicians.
Adapting the Interview to Medical Criteria
The assistant needed to interview each patient according to the guidelines provided by the healthcare professionals. The challenge was not simply to upload a list of questions—those guidelines had to be transformed into a conversational flow that could adapt to each patient's answers.
Variable Document Quality
The documents provided by patients did not always have consistent quality. Some examinations were poorly scanned, incomplete, or difficult to read. During the initial stages of the project, random document checks were performed to evaluate processing quality and identify possible issues, which highlighted the importance of maintaining human review whenever the legibility or completeness of a document was uncertain.
Reliability and Human Oversight
The solution was designed as a support tool for physicians, not as an autonomous diagnostic system. Physicians could review both the generated summary and the original information whenever they needed to verify a detail, with access to the interview recording and complete response history for additional traceability. When a document had quality issues, the information required review before it could be used as a reliable clinical reference. This approach maintained a clear separation between information collection and organization, which the system supported, and medical decision-making, which remained the responsibility of the physician.
Results
The results were primarily qualitative and were not measured through formal performance indicators.
More Complete Medical Histories
Patients had a dedicated space to explain their medical background before the appointment. This made it possible to collect information that might not have surfaced during a traditional consultation.
Easier Appointment Preparation
Physicians received a condensed overview before seeing the patient. Instead of starting from zero, they could focus on the most important issues and ask more relevant follow-up questions.
Time Savings in Some Cases
For some patients, having the medical history organized in advance reduced the amount of time required to collect general information. The time savings were not formally measured and varied according to the complexity of each case.
Improved Patient Experience
Patients were able to explain their situation in greater detail and felt that their full medical history had been heard and considered.
Support for the Diagnostic Process
The structured information gave physicians a broader view of each case. Based on the experience observed during the project, this supported the evaluation and diagnostic process in many cases, although no clinical measurements were conducted to quantify the improvement.
Business Impact
The solution separated part of the information-gathering process from the live medical consultation. This created operational benefits across several areas:
Better use of physician time
Physicians could dedicate more of the consultation to analyzing the case, clarifying uncertainties, and exploring relevant issues.
Greater consistency
Patients went through a structured medical history collection process before their appointment.
More information in one place
Interview data and supporting documents were presented within the same context.
Improved patient experience
Giving patients more time to explain their history contributed to a stronger sense of being heard and receiving thorough attention.
Better preparation
Physicians could review each case before the consultation and decide which issues required additional attention.
Key Learnings
1. AI Should Support, Not Replace, Medical Judgment
The automation was effective for collecting, organizing, and summarizing information. Clinical interpretation and decision-making remained the responsibility of the physician.
2. Agent Memory Is Essential in Long Interviews
Providing instructions is not enough. The system must preserve the right context, avoid repetition, and use previous answers to determine what to ask next.
3. Medical Documents Require Quality Controls
Automated processing depends on files being readable and complete. When there is uncertainty, the workflow should include human verification.
4. A Structured Conversation Can Improve the Patient Experience
Providing a dedicated space before the appointment allowed some patients to explain their situation more fully and feel that their medical history had been properly considered.
Conclusion
This project demonstrates how artificial intelligence can improve a medical workflow without removing the healthcare professional from the decision-making process.
Syntropic Ops designed a system capable of interviewing patients, processing supporting documents, and organizing information before an appointment. The result was a more structured workflow that helped physicians prepare for each case and gave patients more space to communicate their medical history.
The value of the solution was not limited to the use of an AI agent. It came from designing a complete operational process that combined conversation, memory management, document processing, traceability, and human oversight.
Does your team spend too much time collecting information before each appointment? Syntropic Ops designs AI and automation systems that collect, organize, and deliver information reliably while keeping people in control of important decisions.