An AI-powered platform has been developed for a healthcare institution to simplify administrative workflows and handle routine staff tasks. The "Assistant" communicates with patients from the first day until full recovery. This AI helper provides reports at every stage of interaction, improving service quality and enhancing the clinic's reputation. The platform is available both as a web application and a mobile app.
.png)
Our client is a multi-specialty medical clinic. They wanted to rid their staff of routine tasks and improve the quality of service for patients who are far from the medical facility but wish to receive consultations.
.jpeg)
The main issue causing the clinic to lose clients is that calls to the reception desk can be missed due to employee negligence or because the employee had to leave their workplace; additionally, the line might be busy.
There is a large amount of paper-based health records that require constant sorting and analysis. Doctors also spend time collecting test results or other medical data from patients.
Considering that the clinic also provides online consultations, the client wants to monitor the quality of such calls without spending a lot of time. They also noted that these calls should not only benefit management but also assist doctors by generating necessary business documents.
Since the clinic plans to expand and competitors have advanced technologically, the company's management turned to BroutonLab to develop cutting-edge patient service solutions.
Together with the client, we decided to create a platform that integrates with other tools, namely:
Additional capabilities of the Assistant:
Several types of information were analyzed for this project:
To improve the Assistant's accuracy, we used open-source data from similar studies such as on GitHub, which allowed us to reduce expenses on data labeling.
For analyzing and labeling resume data, we used commercial software to avoid time-consuming manual processing and additional budgeting.
.jpeg)
We employed powerful models for text data analysis, including systems like LLama and OpenAI Large Embeddings.
By utilizing Deepgram Voice AI technology, we recognized the patient's speech and emotions, allowing us to synthesize natural responses. Retrieval Augmented Generation (RAG) helps quickly find the necessary context from large volumes of information to answer typical questions.
Sentiment Analysis tracks the patient's emotional state, supporting a personalized approach.
We also used TF-IDF (BM25) for data classification, matching symptoms with the appropriate doctor's profile.
For generating reports and documents, the AI uses Robotic Process Automation (RPA).
These approaches provide higher efficiency and service quality than standard solutions.
In developing and optimizing the server side of the project, we used GoLang to deploy neural networks, ensuring high performance and stability.
Communication between different services was implemented using the gRPC protocol, significantly increasing the overall speed of interaction between platform components. This optimized response times and enhanced the efficient processing of requests from patients and doctors, a key aspect of providing quality service.
The client got an integrated AI-powered platform that automates interactions with patients and optimizes workflows. This allowed the clinic to increase patient satisfaction by 30% and reduce processing time by 40%. Thanks to an effective data analysis system, the clinic attracted more new clients, boosting profits by 25%. The project soon caught the attention of a medical holding company seeking to implement similar technologies, expanding horizons for further innovations in healthcare.
