Med-PaLM: An AI model that integrates multimodal biomedical data
The Fifth Elephant For members

Med-PaLM: An AI model that integrates multimodal biomedical data

The Fifth Elephant paper reading meet-up - May 2024

Tickets

Loading…

About the Paper

Medicine is inherently multimodal and clinicians interpret data from a wide range of modalities when providing care, including clinical notes, laboratory tests, vital signs and observations, medical images, and genomics. Since issues like clinician burnout and lack of access to quality healthcare were rising, the emergence of foundation models turned out to be a turning point in addressing and solving these global problems.

Med-PaLM M is a large multimodal generative model that interprets biomedical data spanning clinical language, medical imaging, genomics, and more performing competently on a diverse array of tasks — all with the same set of model weights.

It is built by fine-tuning and aligning PaLM-E — an embodied multimodal language model developed by the researchers at Google AI using MultiMedBench, a newly curated open source biomedical benchmark. MultiMedBench contains 14 diverse tasks such as medical question answering, mammography and dermatology image interpretation, radiology report generation and summarization, and genomic variant calling.

Link to the paper: https://arxiv.org/pdf/2307.14334.pdf

Key Takeaways:

  • How LLMs help scale up world-class healthcare on a global scale.
  • Understand the importance of AI in clinical workflows and how it exponentiates efficiency and utility with its multimodality.
  • Demonstrate the importance of biomedical fine-tuning and alignment.
  • Learn about zero-shot generalization to novel medical concepts and tasks, positive transfer learning across tasks, and emergent zero-shot medical reasoning.
  • Build intuition for developing large-scale generalist biomedical AI with language as a common grounding across tasks which shows the possibility of combinatorial generalization and positive task transfer.

About the Speaker

Harini Anand is a CSE Undergrad with a keen interest in Computational Cognition. She has formerly worked on building cognitive tools for reducing the onset of Dementia as a student entrepreneur.

RSVP

This paper discussion’s venue is TBD. In-person attendance is free. The Fifth Elephant members can join remotely to watch the live stream.

About The Fifth Elephant monthly paper discussions

The monthly discussions are organized to understand popular papers in Generative AI, DL, and ML domains. Papers are curated to benefit the community. The paper discussion is organized on the first Friday of each month, from 5:30 PM — 7:00 PM.

How you can contribute

  1. Suggest a paper at https://hasgeek.com/fifthelephant/call-for-papers/sub
  2. Moderate/discuss a paper someone else is proposing.
  3. Pick up a membership to support The Fifth Elephant’s activities.
  4. Spread the word among colleagues and friends. Join The Fifth Elephant Telegram group or WhatsApp group.

Contact

For inquiries, leave a comment or call The Fifth Elephant at +91-7676332020.

Hosted by

The Fifth Elephant - known as one of the best data science and Machine Learning conference in Asia - has transitioned into a year-round forum for conversations about data and ML engineering; data science in production; data security and privacy practices. more

About the Paper

Medicine is inherently multimodal and clinicians interpret data from a wide range of modalities when providing care, including clinical notes, laboratory tests, vital signs and observations, medical images, and genomics. Since issues like clinician burnout and lack of access to quality healthcare were rising, the emergence of foundation models turned out to be a turning point in addressing and solving these global problems.

Med-PaLM M is a large multimodal generative model that interprets biomedical data spanning clinical language, medical imaging, genomics, and more performing competently on a diverse array of tasks — all with the same set of model weights.

It is built by fine-tuning and aligning PaLM-E — an embodied multimodal language model developed by the researchers at Google AI using MultiMedBench, a newly curated open source biomedical benchmark. MultiMedBench contains 14 diverse tasks such as medical question answering, mammography and dermatology image interpretation, radiology report generation and summarization, and genomic variant calling.

Link to the paper: https://arxiv.org/pdf/2307.14334.pdf

Key Takeaways:

  • How LLMs help scale up world-class healthcare on a global scale.
  • Understand the importance of AI in clinical workflows and how it exponentiates efficiency and utility with its multimodality.
  • Demonstrate the importance of biomedical fine-tuning and alignment.
  • Learn about zero-shot generalization to novel medical concepts and tasks, positive transfer learning across tasks, and emergent zero-shot medical reasoning.
  • Build intuition for developing large-scale generalist biomedical AI with language as a common grounding across tasks which shows the possibility of combinatorial generalization and positive task transfer.

About the Speaker

Harini Anand is a CSE Undergrad with a keen interest in Computational Cognition. She has formerly worked on building cognitive tools for reducing the onset of Dementia as a student entrepreneur.

RSVP

This paper discussion’s venue is TBD. In-person attendance is free. The Fifth Elephant members can join remotely to watch the live stream.

About The Fifth Elephant monthly paper discussions

The monthly discussions are organized to understand popular papers in Generative AI, DL, and ML domains. Papers are curated to benefit the community. The paper discussion is organized on the first Friday of each month, from 5:30 PM — 7:00 PM.

How you can contribute

  1. Suggest a paper at https://hasgeek.com/fifthelephant/call-for-papers/sub
  2. Moderate/discuss a paper someone else is proposing.
  3. Pick up a membership to support The Fifth Elephant’s activities.
  4. Spread the word among colleagues and friends. Join The Fifth Elephant Telegram group or WhatsApp group.

Contact

For inquiries, leave a comment or call The Fifth Elephant at +91-7676332020.

Hosted by

The Fifth Elephant - known as one of the best data science and Machine Learning conference in Asia - has transitioned into a year-round forum for conversations about data and ML engineering; data science in production; data security and privacy practices. more