The use of AI tools in healthcare is accelerating, increasing confidence in the results generated and delivered by such tools.
One of the tools already in operation at UVA Health, RAMP, focuses on delivering actionable, verifiable, and explainable machine learning that can be integrated into clinical workflows as a decision support tool to improve patient health outcomes. Gain better insight into trends and facilitate faster delivery of needed care to improve patient outcomes. result.
AI-driven predictive analytics models use complex real-time and historical patient data to provide actionable insights to healthcare professionals and alert healthcare teams when a patient requires urgent attention .
Valentina Baljak is a senior data scientist at UVA Health. She holds a PhD in Information Science and Technology and Applied Machine Learning. UVA Health created and currently uses RAMP.
Baljak and two of her colleagues will discuss AI, RAMP, and more in a session titled “Real-Time Analytics Monitoring Platforms: Doing Practical AI” at HIMSS25 in Las Vegas in March. We spoke with Baljak to understand what she and her colleagues will be talking about in the session and what HIMSS25 attendees can take away from it.
Q. What are the main themes covered in the session and why is it relevant to healthcare and health IT today?
A. With the recent emergence of generative AI models, this topic is gaining even more attention in the healthcare field. This study focuses on real-time clinical decision support tools. Artificial intelligence is not a new term.
At UVA Health, we have been working on developing a real-time prediction system for several years. One of the biggest lessons we learned along the way is that the shape AI should take is the one that best serves the needs of our customers. Clinicians don’t use tools they can’t explain. Building trust in our model and tools required close collaboration every step of the way from day one.
We want to provide a blueprint on how to build systems that work in your environment and raise awareness of the importance of transparency, accountability, and explainability in your models. This is especially important in healthcare settings, where real-time predictions can have a significant impact on patient outcomes.
Q. We will focus heavily on AI. How is it used in healthcare in the context of your session’s focus?
A. A key aspect of RAMP is real-time data collection from EHRs and other data sources. RAMP’s ability to write results back to the patient record in the EHR and alert the medical team in real-time makes it an important tool in clinical practice.
The technology used here is fairly established and all open source. Python provides a solid foundation for ML development, backend connectivity, and data processing. Connections to various data sources are built using FiHR, REST API, and custom HL7. The website is built with Angular.
Our latest major expansion is building new predictive models on top of the largest real-time data stream built in Kafka to collect all vitals and ECG waveforms from bedside monitors.
Q. Participants come to sessions to take away knowledge. What is one outcome they can expect?
A. AI is a fundamental part of modern healthcare and can take many forms depending on your needs. Given the high risks, choosing the right AI approach is important.
If you have the in-house expertise and resources, developing a custom AI system can be a powerful alternative to vendor-provided black-box systems.
Valentina Baljak’s session, “Real-Time Analytics Monitoring Platform: Usable AI in Action,” will be held at HIMSS25 in Las Vegas on Tuesday, March 4th at 12:45 p.m.
