Together with Tampere University and University of Jyväskylä, Veracell took on the ambitious project of crafting an automated electronic frailty index (eFI) from patient journals. The challenge? Decrypting the invaluable insights buried within clinician's free-text notes.
Used FinBERT language model to convert rich narratives into structured data.
Transformed free-text into labeled sets for targeted model training.
Collaborated with HUS Acamedic for its seamless Azure ML integration.
Advanced FinBERT language model enabled us to convert rich narratives into actionable structured data, offering fresh perspectives previously hidden in unstructured texts.
By using Azure ML's labeling tools, we effectively harnessed raw patient data, transforming it into labeled sets for targeted model training—our spotlight being incontinence detection, which showed promising results.
We collaborated with HUS Acamedic for its seamless Azure ML integration and data security. Patient data underwent rigorous review and masking, ensuring sensitive details remained confidential while maintaining data's analytical value.
All procedures were meticulously documented in Jupyter Notebooks, ensuring replicability. With our robust groundwork, minor tweaks can further enhance model accuracy and cater to an array of eFI subjects.
"It has been quite beneficial working with Veracell in our NLP project. Veracell engineers helped us get started from the beginning and has been consistently responsive with discussions and ideas for improving our EFI model."
- Jake Lin, PhD, Tampere University
These glimpses are just the tip of the iceberg. Together, let's turn your "what-ifs" into "how-tos," shaping a future we dare to dream.Contact us