Neuroscience PhD Candidate Ricardo Diaz-Rincon recently presented the paper, “Uncertainty-Aware Prediction of Parkinson’s Disease Medication Needs: A Two-Stage Conformal Prediction Approach,” at the Machine Learning for Health Care 2025 (MLHC) Conference earlier in August. The work has been published in the Proceedings of the 10th Machine Learning for Healthcare Conference, one of the most prestigious venues for AI applications in healthcare.
This paper introduces uncertainty quantification in forecasting of Parkinson’s Disease medication needs, addressing a previously uncharted problem of significant clinical importance. The acceptance at MLHC—a highly competitive conference that showcases cutting-edge research at the intersection of machine learning and healthcare—underscores the innovative nature of this approach to one of the most challenging aspects of Parkinson’s care: optimizing medication management while accounting for the inherent uncertainty in disease progression.
Using electronic health records from inpatient admissions at University of Florida Health spanning from 2011-2021, Diaz-Rincon developed a novel two-stage conformal prediction framework that can anticipate medication needs up to two years in advance with reliable prediction intervals and statistical guarantees. This breakthrough addresses the current clinical reliance on trial-and-error decisions by providing neurologists with both precise predictions and reliable confidence measures.
By providing both precise medication predictions and reliable confidence measures, this breakthrough has the potential to transform clinical decision-making in Parkinson’s care and advance precision medicine for neurodegenerative diseases. The work demonstrates how machine learning, when thoughtfully designed with clinical needs in mind, can move beyond traditional trial-and-error approaches to enable evidence-based treatment decisions that could significantly improve patient outcomes.