Related Experiment Video
Updated: Aug 7, 2026
03:14
Augmenting Large Language Models via Vector Embeddings to Improve Domain-Specific Responsiveness
Published on: December 6, 2024
Acceptance of rules generated by machine learning among medical experts
M J Pazzani1, S Mani, W R Shankle
1Department of Information and Computer Science, University of California, Irvine, USA. pazzani@ics.uci.edu
Methods of Information in Medicine
|January 5, 2002
Summary
Monotonicity constraints improve the accuracy and expert acceptance of machine learning rules for medical data analysis. These constraints ensure learned models are coherent and align with existing medical knowledge, enhancing decision-making.
Area of Science:
- Artificial Intelligence
- Medical Informatics
- Machine Learning
Background:
- Medical database analysis offers potential for improved patient outcomes and cost reduction.
- Machine learning techniques are increasingly used to discover patterns for decision support in healthcare.
Purpose of the Study:
- To evaluate if monotonicity constraints enhance the accuracy and interpretability of machine learning-derived rules.
- To assess expert willingness to adopt rules generated with and without monotonicity constraints.
Main Methods:
- Two diverse datasets (dementia screening, mental retardation risk) were used.
- A rule learning system was applied with and without monotonicity constraints.
- Expert evaluation of rule usability and accuracy assessment were conducted.
Main Results:
- Rules generated with monotonicity constraints demonstrated equal or superior accuracy compared to those without.
- Experts expressed a higher willingness to utilize rules developed with monotonicity constraints.
Conclusions:
- Monotonicity constraints can lead to more credible and acceptable machine learning models for medical applications.
- Ensuring learned models are coherent and consistent with existing medical knowledge is crucial for expert adoption.
