Avsnitt 93: Large language models use cases in healthcare

Our guest for this episode is Farzaneh (Kobra) Etminani, Associate Professor in Machine Learning at Halmstad University.

Farzaneh shares with us her research on Machine Learning for Healthcare applications and the recent progress in the field of Machine Learning that she sees promising. In particular, she is very optimistic about large language models such as GPT (Generative Pre-Trained Transformer).

Although large language models may not be enough for an accurate use in their current form, there is a big potential of further fine tuning these models with additional expert data (that is further training the model with additional data that you might have for which you want the language model to perform even better).

Farzaneh also discusses the challenges of data access in terms of a certain language, such as the Swedish language which has a much smaller corpus than the English language. She also acknowledges that fact that the data available in healthcare and medicine is unbalanced, where (thankfully) there are many rare diseases that we don’t have enough data of.