As a physician, you're painfully aware that medical knowledge is growing at such a fast pace that it's impossible to keep pace with it all. It can even be difficult to quickly pull up everything you need to make the best decision possible for your patient. All of that can contribute to delayed or incorrect diagnoses, with diagnostic errors costing the U.S. health care system between $250 billion to $500 billion annually. Health2047--the Silicon Valley-based subsidiary of the AMA created to overcome systemic dysfunction in the U.S. health care system--recently launched RecoverX to help combat those challenges. The startup company aims to solve the system-level challenges posed by a body of medical knowledge growing so fast that it's humanly impossible to keep up with.

Medical dialogue generation aims to provide automatic and accurate responses to assist physicians to obtain diagnosis and treatment suggestions in an efficient manner. In medical dialogues two key characteristics are relevant for response generation: patient states (such as symptoms, medication) and physician actions (such as diagnosis, treatments). In medical scenarios large-scale human annotations are usually not available, due to the high costs and privacy requirements. Hence, current approaches to medical dialogue generation typically do not explicitly account for patient states and physician actions, and focus on implicit representation instead. We propose an end-to-end variational reasoning approach to medical dialogue generation. To be able to deal with a limited amount of labeled data, we introduce both patient state and physician action as latent variables with categorical priors for explicit patient state tracking and physician policy learning, respectively. We propose a variational Bayesian generative approach to approximate posterior distributions over patient states and physician actions. We use an efficient stochastic gradient variational Bayes estimator to optimize the derived evidence lower bound, where a 2-stage collapsed inference method is proposed to reduce the bias during model training. A physician policy network composed of an action-classifier and two reasoning detectors is proposed for augmented reasoning ability. We conduct experiments on three datasets collected from medical platforms. Our experimental results show that the proposed method outperforms state-of-the-art baselines in terms of objective and subjective evaluation metrics. Our experiments also indicate that our proposed semi-supervised reasoning method achieves a comparable performance as state-of-the-art fully supervised learning baselines for physician policy learning.

The increased use of electronic health records has made possible the automated extraction of medical policies from patient records to aid in the development of clinical decision support systems. We adapted a boosted Statistical Relational Learning (SRL) framework to learn probabilistic rules from clinical hospital records for the management of physiologic parameters of children with severe cardiac or respiratory failure who were managed with extracorporeal membrane oxygenation. In this preliminary study, the results were promising. In particular, the algorithm returned logic rules for medical actions that are consistent with medical reasoning.