Progress Update: Hasitha’s AI-Driven Healthcare Research
We’re excited to share an update on Hasitha’s PhD research journey, co-funded by Datarwe and Griffith University. Leveraging her data science background, Hasitha is working to build an AI-driven system that helps clinicians develop optimal, patient-centered treatment plans without relying on the usual “black box” approach.
In addition to her doctoral work, Hasitha is also engaged in collaborative sessions with clinicians from various specialties, aiming to bridge the gap between pure data science and everyday patient care. By participating in bedside observations, conducting simulation-based evaluations, and maintaining open feedback loops with frontline providers, her research is being continuously refined to address real-world clinical needs. This multi-disciplinary approach ensures that the AI tools she develops aren’t just theoretically sound, but also practical, user-friendly, and deeply aligned with the workflows of the healthcare professionals who will ultimately rely on them.
Transforming the “Black Box” into a “Box of Knowledge”
Instead of providing obscure, AI-generated decisions, Hasitha’s research aims to offer transparent, actionable insights—a “box of knowledge.” This approach focuses on:
- Early outcome prediction for TBI patients in ICU settings
- Real-time clinical decision support for resource allocation
- Improving survival outcomes by integrating AI-driven models directly into hospital systems
“The opportunity to work on a problem with such direct clinical impact—where predictive models could improve survival outcomes and optimise resource allocation—motivated me to pursue this research,” Hasitha says.
Deep Learning for ECG Analysis
Hasitha’s work goes beyond TBI-focused models and extends into Electrocardiogram (ECG) signal analysis. By using deep convolutional neural networks (DCNNs), she’s developing methods to detect critical heart abnormalities such as Ventricular Ectopic Beats (VEB) and Supraventricular Ectopic Beats (SVEB). Early detection of these conditions is crucial to prevent life-threatening cardiac arrhythmias.
Three publications detailing these ECG analysis methods are currently in the pipeline, and we’ll share them as soon as they’re released.
