What is Big Biomedical Data Science?

A field that develops and applies algorithms, statistical models, and (database / machine learning) systems to manage, visualize, wrangle, summarize, generalize, and control big data to improve health and healthcare.

What is Big Biomedical Data Science?

A field that develops and applies algorithms, statistical models, and (database / machine learning) systems to manage, visualize, wrangle, summarize, generalize, and control big data to improve health and healthcare.

What is Big Biomedical Data Science?

A field that develops and applies algorithms, statistical models, and (database / machine learning) systems to manage, visualize, wrangle, summarize, generalize, and control big data to improve health and healthcare.

data = genomics, imaging, electric health records, etc.

Reason #1: Evolution

• 1 billion years of evolution for human perceptual capacities
• 1 billion receptors at 1 kHz each is 1 terabit per second
• Minimize false negatives: is that a tiger?

Reason #2: We Don't Know What Random Looks Like

Reason #2: We Don't Know What Random Looks Like

Reason #3: Grazing Goat Starves

Reason #4: How many circles per square?

• ratio of volume of ball to cube with diameter 1

Reason #4: How many circles per square?

• ratio of volume of ball to cube with diameter 1
• volume of cube → 1
• volume of ball $\approx \frac{\pi^{n}}{n!}$ → 0
• ball contains no volume in high-dimensions

Reason #4: How many circles per square?

• every additional dimension squares number of corners
• cubes are pointy in high-dimesions

Reason #5: Estimate mean

• sample $x \sim \mathcal{N}_1(\mu,1)$, estimate $\mu$?

Reason #5: Estimate mean

• sample $x \sim \mathcal{N}_1(\mu,1)$, estimate $\mu$?
• sample $[x_1, x_2] \sim \mathcal{N}_2([\mu_1, \mu_2], I)$, estimate $\mu = [\mu_1, \mu_2]$?

Reason #5: Estimate mean

• sample $x \sim \mathcal{N}_1(\mu,1)$, estimate $\mu$?
• sample $[x_1, x_2] \sim \mathcal{N}_2([\mu_1, \mu_2], I)$, estimate $\mu = [\mu_1, \mu_2]$?
• sample $[x_1, x_2, x_3] \sim \mathcal{N}_3(\mu, I)$, estimate $\mu$?

The usual estimator of the mean of a multivariate Gaussian is inadmissable

Stein, 1956

Summary

Don't trust intuition, it sucks at this.

#1: Build Better Human Brains

1. Impose selective pressures
2. wait 1,000 generations
3. have better human brains

#2: Build Knowledge Systems

#2: Build Knowledge Systems

AI Winter #2

#3: Build Learning Systems

AI Winter #3

Approaches that will fail

1. better human brains (evolution)
2. just parametric modeling (knowledge systems)
3. just deep learning (machine learning systems)

Proposed Solution: AI Spring #4

1. Work together: AI + domain experts
2. Build probabilistic generative models using maximum amount of domain knowledge
3. Use those models to guide a search for low-dimensional latent structure

How Many Points to Describe a Line?

How Many Points to Describe a Plane?

Proposed Solution

1. Work together: AI + domain experts
2. Build probabilistic generative models using maximum amount of domain knowledge
3. Use those models to guide a search for low-dimensional latent structure
   • We need $\geq n$ points to describe a line in $n$ dimensions
   • Our $p > n$
   • We need to learn a "line" works best
   • There are $\infty$, so need experts to help guide search

Good News, Bad News

• Good: you can, in theory, already do this
• Bad: no idea of this will work.

A Glimmer of Hope

Conclusions

• big biomedical data science is hard
• our intuitions are bad
• success requires (swallowing pride):
  • convert your knowledge generative models
  • leverage models to guide search for "structure" in learning systems

Acknowledgements

♥, 🦁, 👪, 🌎, 🌌

Four V's of big data

• volume
• velocity
• variety
• veracity

Drosophila Brain Networks

Geodesic Learning Drosophila Brain