Unlocking the Secrets of Biological Data with LLMs

Tarun Mamidi

(with help of GenAI)

LLM (Large Language model)

https://www.bualabs.com/archives/4402/what-is-large-language-model-llm/

Thirunavukarasu, A.J., Ting, D.S.J., Elangovan, K. et al. Large language models in medicine. Nat Med 29, 1930–1940 (2023).

The future landscape of large language models in medicine

Clusmann, J., Kolbinger, F.R., Muti, H.S. et al. The future landscape of large language models in medicine. Commun Med 3, 141 (2023)

Example-1: Gene Set Summarization using Large Language Models

https://arxiv.org/pdf/2305.13338.pdf

Example-2: Understanding protein language

Madani, A., et al. Large language models generate functional protein sequences across diverse families. Nat Biotechnol 41, 1099–1106 (2023)

Example-3: Variant impact prediction

https://github.com/facebookresearch/esm

Example-4: Can LLMs be used to predict protein structure?

How can I use LLMs in my research?

Rapid fire questions:

How many

• used chatGPT in general?
• asked it about your research topic?
• are satisfied with the answer?
• tried multiple prompts to get the right answer?
• used plugins to achieve your goal?
• used other LLMs?
• heard the term “Hallucination”?

https://arxiv.org/pdf/2309.01219.pdf

Fine tuning

Prompt Engineering

(with examples)

Chat with PDFs

https://blog.nextideatech.com/chat-with-documents-using-langchain-gpt-4-python/

Chat with data

Autonomous multiple agents

https://github.com/OpenBMB/ChatDev

MetaGPT

https://github.com/geekan/MetaGPT

Potential for LLMs in medicine

Moor, M., Banerjee, O., Abad, Z.S.H. et al. Foundation models for generalist medical artificial intelligence. Nature 616, 259–265 (2023).

Ethical Considerations in Using LLMs in Medicine

Privacy and Data Security:

• Privacy concerns when using LLMs to analyze patient data.
• Importance of robust data security measures to protect sensitive medical information.

Bias and Fairness:

• Risk of bias in LLMs, which can lead to unequal treatment or diagnosis.
• Need for fairness audits and bias mitigation strategies in model development.

Informed Consent:

• Importance of informed consent when collecting and using patient data.
• How LLMs can help in improving the clarity of consent forms and patient communication.

Transparency and Accountability:

• Challenge of understanding LLM decision-making processes, which can be seen as "black boxes."
• Importance of model transparency and mechanisms for accountability in healthcare applications.

Regulatory Compliance:

• Need to comply with existing healthcare regulations and standards.
• Role of regulatory bodies in overseeing the use of LLMs in medical practice and research.

Review

https://nipchan.neocities.org/pages/about.html

Opportunities are endless …

Thank you!