Figure 3
From: Towards a data-driven system for personalized cervical cancer risk stratification

Absolute risk estimated from observed data and model predictions. The \(\eta\) score computed with (9) indicates model performance over female age intervals. The prediction models are matrix factorization (MF), hidden Markov model (HMM), geometric deep learning (GDL) gradient tree boosting (GTB), logistic regression (LR), and random forest (RF).