Peter Norvig, Google - Stanford Big Data 2015

TL;DR

Navigating the sea of data in the field of biomedicine requires overcoming cultural, computational, and methodological challenges, while also embracing new approaches and tools.

Transcript

so it's it's clear that I'm the one that knows the least about biomedicine of anybody on this panel and probably anybody in the whole room but I can try to tell you a little bit about what it's like to live in a sea of data and swim in it every day and how to get along with that and and hopefully it'll apply to the type of work that you guys do so ... Read More

Key Insights

• 💨 The shift towards analyzing shared data in biomedicine requires addressing cultural challenges and finding ways to credit those who create, share, and analyze data.
• ✊ Computing power is a critical resource, and researchers need to make strategic decisions about renting or buying and choosing vendors.
• 👋 Best practices in software engineering, including test-driven development, privacy, security, version control, and data management, are crucial in biomedical research.
• 🔊 The high volume of data in biomedicine increases the risks of errors and the need for careful analysis and interpretation.
• 👨‍🔬 Biomedical research has experienced progress in recognizing objects in pictures and generating captions using sophisticated mathematical models and vast amounts of data.
• 🌍 The ability to accurately identify objects and generate captions indicates a deep understanding of both the visual world and the structure of language.
• 🍵 While progress has been made in this field, challenges remain, such as handling novel and unusual data scenarios and improving accuracy in image recognition and captioning.
• 🫵 Achieving promising results in biomedicine requires a combination of sophisticated mathematics, a comprehensive view of the world, and specific knowledge about visual systems.

Questions & Answers

Q: How does the new approach to data in biomedicine differ from traditional scientific research?

The new approach involves collecting and analyzing shared data from multiple sources, shifting from an individual to a collaborative effort.

Q: What are the social challenges in sharing data in the field of biomedicine?

The cultural component of data sharing includes issues of credit, funding, and tenure, which need to be addressed to change the current mindset.

Q: How should researchers manage computing power for large-scale data analysis?

The field is transitioning towards renting computing power rather than owning data centers, and researchers need to consider if they should source from a single vendor or multiple ones based on their needs.

Q: What software engineering best practices are important in managing biomedical data?

Core engineering principles such as maintaining tests, privacy, security, version control, provenance, and robust coding practices should be implemented to ensure efficiency and reliability.

Summary & Key Takeaways

• Traditional scientific research follows a linear process of theory, data collection, comparison, and publication, but a new approach is emerging where shared and analyzed data from multiple sources play a central role.

• Sharing data and changing the cultural mindset around data ownership and credit poses a social challenge in the field of biomedicine.

• Computing power is a crucial resource, and the decision to rent or buy, choose a single vendor or multiple ones, and effectively manage and archive large amounts of data are important considerations.

• Software engineering best practices, such as version control, privacy, security, and provenance, are essential in managing and analyzing biomedical data.

• Data mining presents new challenges, where the sheer volume of possibilities and experiments can lead to a high risk of errors.