Udi Manber – Healthcare as an information business

1. Udi Manber

1.1. NIH, former Google, Yahoo and Amazon Chief Algorithm Officer

2. Google has enormous computing power

2.1. "When you take a search on Google, you send a man to the moon"

2.2. Same computing power as the whole Apollo mission for one search

2.3. There's importance for including context, recent events,...

2.4. Google learns

2.5. Google would be out of business without user data

3. Medicine doesn't/can't learn

3.1. Google Engineers look at a query and optimize it on screen. They get a vast amount of data on a single query

3.2. Doctors don't have access to the vast amount of data

3.3. User (patient) data is not used in medicine

3.4. Move to electronics from paper form

3.5. Now 80% use complex electronics

3.6. Patient history is locked in system so the systems don't learn – one patient at a time

3.7. No appetite for data

4. Standard Response "Patient data is noisy". We don't want to use it.

4.1. Websites are even noisier

4.2. It definitely takes time to learn

4.3. Develop techniques to learn how to learn from your patient data

5. Long Tail

5.1. When you are dealing with large scale information, long tail rules

5.2. There are so many unusual things in medicine

5.3. So many combinations of diseases in medicine

5.4. Not enough data to compare

5.5. Privacy Issues

5.5.1. Solvable

5.6. It can be done!

6. Data scientists are craving for medical data

6.1. Help us get the data