The Truveta Language Model

Healthcare organizations generate immense volumes of data, but an estimated 95% of this data goes unused – largely because it is fragmented, inaccessible, and unstructured. Advancements in AI have presented a unique opportunity to transform and clean massive streams of healthcare data to make it available for research, innovation, and patient care. The Truveta Language Model (TLM), a large-language model trained on medical records data, cleans billions of EHR data points provided by 30 member health systems to enable innovative healthcare research. 
  
This whitepaper provides insight into:

• What TLM is, how it works, and what sets it apart from other large language models
• How TLM handles the ingestion and cleaning of healthcare data to ensure high-quality inputs for research
• The TLM training process and the clinical experts involved
• How TLM extracts and normalizes concepts contained within clinical notes