BERT 36L Recon Sepsis Diag is a machine learning model designed for predicting in-hospital mortality related to sepsis using ICD diagnosis text as input. The model is built with a BERT-style Transformer architecture, featuring six encoder branches with six layers each. It is intended for use in scenarios where sepsis diagnosis information is available in ICD-coded text format and mortality prediction is required.
The architecture of the model includes a vocabulary size of 5,175, a hidden size of 128, four attention heads, a feed-forward dimension of 2,048, and supports a maximum sequence length of 30. The total number of parameters in the model is 22,045,058. The model operates in a mode labeled "dig_lab" and is implemented in PyTorch. Loading and inference are performed using the provided BertClassifier class, which allows users to load the model checkpoint and perform predictions on tokenized text inputs.
BERT 36L Recon Sepsis Diag is available on the Hugging Face platform and is suitable for users familiar with PyTorch who need to perform sepsis-related mortality prediction tasks from ICD diagnosis text.
In the Other AI space, BERT 36L Recon Sepsis Diag takes a focused approach. It provides a deep BERT-based model for predicting sepsis mortality from ICD diagnosis text for healthcare research. It is built as an open-source project for healthcare researchers. BERT 36L Recon Sepsis Diag is open source under the Open Source license. It runs on the web and API.
Behind BERT 36L Recon Sepsis Diag is fansen, and the product first shipped in 2023. Key capabilities include sepsis prediction, Deep BERT architecture, and text classification.
Latest indexed changes and source events
fansen/BERT-36L-Recon-Sepsis-Diag verified by the PulseGate indexer
Other apps tracked under the same category.
