Relevance-Based Prediction: A Transparent, Patient-SpecificApproach to Predicting Medical Outcomes
- 1The Johns Hopkins University School of Medicine
Abstract
Accurate prediction of individual medical outcomes is essential for optimizing treatment allocation amid rising costs, coverage denials, and limited clinical resources. Traditional predictive models, including linear regression and neural networks, rely on average effects and cannot tailor predictions to the specific circumstances of individual cases. We present relevance-based prediction (RBP), a model-free and task-specific method that predicts outcomes as weighted averages of observed cases, with weights determined by a rigorously defined and theoretically justified measure of relevance. Unlike model-based methods that rely on fixed calibrated parameters, relevance-based prediction revisits the original data for each prediction and customizes both the cases and variables used. We illustrate RBP by applying it to predict opioid treatment outcomes and demonstrate that it provides case-specific insights unavailable from conventional models, including how each prior case informs a prediction, how each variable affects each predictions reliability and value, and how reliable each prediction is before it is made. These individualized insights may prevent misleading average-based decisions and reduce harmful or suboptimal treatment.
Related
- WITHDRAWN: Relevance Based Prediction: A Transparent, Non-Artificial Intelligence, Mathematical Solution to Personalized Opioid Treatment
- From Molecular State to Clinical Outcomes After Spinal Cord Injury: A Scoping Review of Selected Fluid Biomarkers and Rehabilitation Relevance
- Electrically assisted bicycles in adults: a scoping review of health outcomes and relevance to rehabilitation.
- Predicting the Risk of Unplanned Hospital Encounters in Hemodialysis Patients: A Comparison Between Clinicians and a Machine Learning Prediction Algorithm
- Childhood-Oncology Intelligence for Predicting Cancer Emergence, Optimizing Precision Therapies, and Simulating Lifelong Outcomes in Pediatric Malignancies Using Autonomous Foundation Models
BibTeX
@misc{robinson2026relevancebased,
title = {{Relevance-Based Prediction: A Transparent, Patient-SpecificApproach to Predicting Medical Outcomes}},
author = {C. L. Robinson and D. Turkington and L. Lee and M. Kritzman and R. J. Yong},
howpublished = {medRxiv},
year = {2026},
month = sep,
doi = {10.64898/2026.09.28.26364225},
url = {https://www.medrxiv.org/content/10.64898/2026.09.28.26364225v1},
note = {CC BY-NC-ND},
}
RIS
TY - UNPB
TI - Relevance-Based Prediction: A Transparent, Patient-SpecificApproach to Predicting Medical Outcomes
AU - C. L. Robinson
AU - D. Turkington
AU - L. Lee
AU - M. Kritzman
AU - R. J. Yong
PY - 2026
DA - 2026/09/28
PB - medRxiv
DO - 10.64898/2026.09.28.26364225
UR - https://www.medrxiv.org/content/10.64898/2026.09.28.26364225v1
AB - Accurate prediction of individual medical outcomes is essential for optimizing treatment allocation amid rising costs, coverage denials, and limited clinical resources. Traditional predictive models, including linear regression and neural networks, rely on average effects and cannot tailor predictions to the specific circumstances of individual cases. We present relevance-based prediction (RBP), a model-free and task-specific method that predicts outcomes as weighted averages of observed cases, with weights determined by a rigorously defined and theoretically justified measure of relevance. Unlike model-based methods that rely on fixed calibrated parameters, relevance-based prediction revisits the original data for each prediction and customizes both the cases and variables used. We illustrate RBP by applying it to predict opioid treatment outcomes and demonstrate that it provides case-specific insights unavailable from conventional models, including how each prior case informs a prediction, how each variable affects each predictions reliability and value, and how reliable each prediction is before it is made. These individualized insights may prevent misleading average-based decisions and reduce harmful or suboptimal treatment.
ER -