Administrative workload in healthcare keeps rising, and Czech hospitals struggle with limited time and staffing. Medicalc Software is tackling the challenge with technology: this year it is launching an AI-driven assistant that helps physicians and nurses automate routine administrative workflows. How does the new tool work, and what can healthcare facilities expect from it? The company’s CEO, RNDr. Jan Kupka, answers our questions.
What was the primary motivation behind creating the AI assistant? Was it driven more by the needs of doctors or by your own innovative ideas?
Surprisingly, the initial drive for this came from the government. The requirements for data standards in electronic data exchange between healthcare facilities created a need for data to be delivered in a strictly defined, structured format. At some point, there was consideration that a discharge report could break down into several hundred individual items. The idea of doctors filling in these items by hand was so daunting that we began to look for creative solutions through AI. This naturally led us to start using AI in other areas to improve efficiency for healthcare professionals, such as converting audio recordings into structured medical reports (during outpatient visits, rounds, surgeries, or autopsies) and digitising paper documents using OCR technology.
Your AI assistant can transform doctor-patient conversations into structured documentation. How technically demanding was it to create a model that comprehends Czech terminology and practices?
Training the model itself is not particularly technically challenging. The key is effectively putting together the various components to achieve the desired results. You are tackling not just one task but several simultaneously, which involves using multiple models. Accessing high-quality data is crucial in the following phase. We benefit from our hospital information system, Medicalc, which operates in about one hundred healthcare facilities across the Czech Republic. This offers an extensive network of partners for collaboration in developing training data and collecting feedback. Another critical factor for success is direct integration with the HIS and our mobile application for healthcare personnel, mMobile, which keeps the AI assistant readily accessible. This way, healthcare workers are not forced to use unfamiliar tools.
What is the process for integrating the AI assistant into the hospital information system, and how long does it typically take to implement it in everyday operations?
The AI assistant comes pre-integrated with the hospital system out of the box. The task is mainly to configure and tailor it to the specific needs of the hospital. Each facility is unique; for example, they may use slightly different document templates. The particular recording equipment also influences the process. Therefore, we conduct a final calibration of the model for each facility. If all goes well, the preparation work, from selecting departments to launching the pilot phase, takes about a month.
What personally inspires you the most about this project? Is it the technology itself or the potential to transform doctors’ daily routines?
Just as no one decides to become a teacher out of a strong desire to fill out attendance registers (paper-based or digital), people do not choose to become doctors or nurses because they enjoy administrative tasks. We see technology as a means to ease the burden of administration on healthcare workers. In the end, we are doing this for our own benefit, too, as we are all patients who desire an efficient healthcare system where healthcare professionals have the time and energy to care for us.
