Led by David Sontag, the Clinical Machine Learning Group is interested in advancing machine learning and artificial intelligence, and using these techniques to advance health care.

Broadly, we have two goals:

**Clinical**: To truly make a difference in health care, we need to create algorithms that are useful for solving real clinical problems.**Machine learning**: We need rigorous solutions, which can pave the way for safe deployment of machine learning in high-stakes settings like healthcare.

**7/19/2021**: We have three papers at ICML 2021, with Mike presenting “Regularizing towards Causal Invariance: Linear Models with Proxies”, Hunter presenting “Graph cuts always find a global optimum for Potts models (with a catch)", and Zeshan presenting “Neural Pharmacodynamic State Space Modeling”**5/13/2021**: Monica presented at CHI 2021 on assessing the impact of decision aid on clinicians**3/11/2021**: Watch David present “Using machine learning to guide treatment suggestions” at the European Winter Symposium on Machine Learning Frontiers in Precision Medicine**11/04/2020**: New paper in Science Translational Medicine on learning antibiotic treatment policies**11/04/2020**: We have released AMR-UTI through Physionet, a freely available dataset for studying antibiotic resistance and treatments**08/27/2020**: Mike presented at AISTATS 2020 on characterizing overlap in causal inference**08/08/2020**: Two papers at MLHC 2020: Divya presenting on contextual autocomplete and Monica presenting on clinical entity extraction.

We prove that the alpha-expansion algorithm for MAP inference always returns a globally optimal assignment for Markov Random Fields …

Modeling the time-series of high-dimensional, longitudinal data is important for predicting patient disease progression. However, …

We propose a method for learning linear models whose predictive performance is robust to causal interventions on unobserved variables, …

Healthcare providers are increasingly using machine learning to predict patient outcomes to make meaningful interventions. However, …

Overlap between treatment groups is required for non-parametric estimation of causal effects. If a subgroup of subjects always receives …

Quickly discover relevant content by filtering publications.

Automated decision support can accelerate tedious tasks as users can focus their attention where it is needed most. However, a key …

Key oncology end points are not routinely encoded into electronic medical records (EMRs). We assessed whether natural language …

Reinforcement learning (RL) has the potential to significantly improve clinical decision making. However, treatment policies learned …

Several works have shown that perturbation stable instances of the MAP inference problem in Potts models can be solved exactly using a …