Search papers, labs, and topics across Lattice.
This paper introduces a conservative stochastic control framework for optimizing treatments from irregularly sampled patient data, modeling patient dynamics as a controlled stochastic differential equation. To prevent the exploitation of model errors and out-of-support controls, they incorporate a signature-based Maximum Mean Discrepancy (MMD) regularizer on path space, penalizing treatment plans that deviate from observed trajectories. Empirical results on benchmark datasets demonstrate enhanced robustness and performance compared to non-conservative approaches.
By penalizing treatment plans that lead to trajectory distributions far from observed patient data, this method provides a more robust approach to treatment optimization than standard model-based methods.
We develop a conservative continuous-time stochastic control framework for treatment optimization from irregularly sampled patient trajectories. The unknown patient dynamics are modeled as a controlled stochastic differential equation with treatment as a continuous-time control. Naive model-based optimization can exploit model errors and propose out-of-support controls, so optimizing the estimated dynamics may not optimize the true dynamics. To limit extrapolation, we add a consistent signature-based MMD regularizer on path space that penalizes treatment plans whose induced trajectory distribution deviates from observed trajectories. The resulting objective minimizes a computable upper bound on the true cost. Experiments on benchmark datasets show improved robustness and performance compared to non-conservative baselines.
