Attention-Based Estimation of the Individual Treatment Benefit Probability under Dose Variation

In three linesDose-AIPTB proposes an attention-based method to estimate the probability that a treatment benefits an individual patient under varying doses. The framework recasts the problem as binary classification of the individual treatment effect sign, using pairwise comparisons and attention-based aggregation. Code available on GitHub.

Summary generated by Claude — human-verified