New method translates black-box AI models into auditable clinical nomograms

Researchers have developed a new method called PRiSM (Partial Responses in Structured Models) to translate complex, black-box clinical prediction models into understandable nomograms. This technique captures the shape and interaction of effects from the original model, allowing for auditing and selection of variables. When applied to heart transplant recipient data, nomograms derived from various models, including logistic regression, neural networks, random forests, and XGBoost, met noninferiority criteria for discrimination and generally maintained calibration. The PRiSM method is now available as an open-source Python package. AI

IMPACT Enables greater transparency and auditability of AI models in critical healthcare applications.

RANK_REASON The cluster contains an academic paper detailing a new method for translating machine learning models. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• explainable boosting models
• Generalized Additive Models
• logistic regression model
• neural additive models
• Neural Networks
• PRISM
• Python
• random forest
• XGBoost

• paper
• model release

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →