Intensive Care as One Big Sequence Modeling Problem
This work addresses the challenge of enabling generalist models for healthcare, potentially improving efficiency and outcomes for patients and providers, though it is incremental as it builds on existing sequence modeling and foundation model concepts.
The paper tackles the problem of narrow task-specific reinforcement learning in healthcare by proposing a new paradigm of healthcare as sequence modeling, where patient-provider interactions are represented as event streams and tasks like diagnosis and treatment are modeled as future event predictions, and demonstrates this by developing MIMIC-SEQ, a benchmark derived from MIMIC-IV data, and training a baseline model to explore its capabilities.
Reinforcement Learning in Healthcare is typically concerned with narrow self-contained tasks such as sepsis prediction or anesthesia control. However, previous research has demonstrated the potential of generalist models (the prime example being Large Language Models) to outperform task-specific approaches due to their capability for implicit transfer learning. To enable training of foundation models for Healthcare as well as leverage the capabilities of state of the art Transformer architectures, we propose the paradigm of Healthcare as Sequence Modeling, in which interaction between the patient and the healthcare provider is represented as an event stream and tasks like diagnosis and treatment selection are modeled as prediction of future events in the stream. To explore this paradigm experimentally we develop MIMIC-SEQ, a sequence modeling benchmark derived by translating heterogenous clinical records from MIMIC-IV dataset into a uniform event stream format, train a baseline model and explore its capabilities.