1.16.2016

Data science at Aspire Healthcare

6 comments:

I'm starting as chief data scientist at Aspire Health. I'm excited about it. Let me tell you why.

Aspire provides in-home nursing services to very sick patients, on a shared savings model. Regular in-home checkins from nurses help patients stay healthier, which reduces unnecessary hospitalizations, which creates significant savings for their insurance companies. It’s a rare niche in medicine where everybody wins.

I started working with Aspire last summer, as a contractor. At first, I thought of it as one of several companies with interesting data challenges. ("Medium-large scale data mining to identify preventable hospitalizations.") Over time, I realized that the company is in a unique place, and that joining would give me a unique opportunity to help shape the future of health care by doing what I do best.

Here are the things that convinced me, in no particular order.

Aspire is targeting one of the biggest opportunities in health care today.
10% of Americans account for 68% of health care costs. Evidence shows that as much as 22% of that cost comes from preventable complications—especially unnecessary hospital visits.

By conservative estimates, preventing the complications that send chronically ill people to the hospital is a $40 billion opportunity—arguably the single biggest opportunity in American health care. And it’s done by improving quality of care to patients, not rationing it.

In other words, Aspire is building a data system that will literally save lives, by helping chronically sick patients avoid complications that can send them to the hospital and/or kill them.

The company is in hypergrowth mode.
Aspire has a proven business model and all the right connections for setting up contracts with insurance companies. That means that the company can scale very, very quickly. I love the pace and urgency of high-growth businesses. I love the challenge of solving one problem after another as quickly as possible, while making sure that the pieces cumulate to more than the sum of their parts.

On top of that, the company is led by a fantastic team. I love learning from a team of pros---people who are very good at their jobs, could choose to do almost anything, and have decided to do this, together, because it’s the most compelling thing they could find to work on.

Data is an integral part of Aspire’s business model.
Like a fintech company, Aspire succeeds largely on the basis of its ability to segment and address risk. Algorithms and supporting data infrastructure play a key role in the company's growth and operating efficiency.

I like the idea of building one of the pillars of the company, rather than an R&D lab or a sub-group within analytics or marketing. There’s nothing wrong with those roles. But given the choice, I prefer to stay closer to the action.

Aspire is poised to shape the future of preventative medicine.
Aspire's first major data challenge is to identify patients who have unnecessary hospitalizations in their future, on a 1-to-2-year horizon. We have to do this for many patients, based on complex, messy input data (administrative records, medical notes), over different time scales (years, weeks, days.)

To do the job right, we’re going to need to blend skills and best practices across disciplines that don’t often talk to each other: applied statistics and research design, scalable data architecture, health IT, and the medical practice of palliative care. It’s a steep learning curve that demands a lot of ingenuity.

Aspire's second major data challenge is to coordinate care from doctors, nurses, social workers, and others, to actively prevent complications that can lead to unnecessary hospitalization. In a market still dominated by fee-for-service medicine, this is rare and exciting. We're going to see more of it, though, and Aspire is one of the teams leading the way.



Long story short, as a data guy who loves working on human-scale problems, Aspire is perfect. There's a lot I need to learn about health care and medicine, and I'm looking forward to learning, growing, and building in this new role.

More to come!



Read More

8.03.2015

Foo camp aftermath

1 comment:
I had a great time at Foo Camp last weekend. It’s an awesome melting pot people and ideas. Extremely high innovation density.

Here are my notes (“things to dig into later”) from the weekend, with no attempt at explanation. These are a mix of nifty resources, topics discussed, and my own thoughts.


Data visualization
Bokeh
Seaborn
Pythonic perambulation 

From the “Cleverness with Slack” session:
Plugins
FutureBot for Slack
Emoji : mechanism for "silence does not imply consent"
Async Standup plugin
Look up the point of contact for a customer
Trello integration
Slackline

Process/culture things to have conversations about:
Scrollback culture: decide which rooms must be read in full.
When to use notifications
Adoption: Engineering teams vs sales teams
People who are uncomfortable broadcasting
Hiring: Slack means people need to communicate well in writing.
Onboarding process
Metadiscussion among teachers/power users as opposed to the main community

Feature requests
Ephemeral access to rooms
First class action links
Faster syncing
Valid DM pairs (for Chinese-wall compliance)

Examples of interesting Slack communities:
Home buying with girlfriend
YC founders
Sandwich network
Julia committers


Slack as antidote to “Trip to HQ phenonmenon"
Deep learning Bots on Slack
Culture transmission — slack brings this into the open.

Precision medicine
figma
"21st century research platform"
Health data as a publicly provided commodity (kinda like census)
BlueButton
Right to access data: BS around HIPPA
Million veterans project

Wouldn’t it be nice if...
wider EMR adoption
NIH data stream
clarify patients’ rights
Accelerating access to data

Google baseline study
value-based payer models

WTF economy
“Tech is becoming the villain in our economic narrative"
But compared to what?
Shift work at McDonalds or Walmart is pretty lousy, too.

What’s working in the new economy?
Voluntary, augmentation, better service

Driving income inequality…?
The wealthy underspend as a percentage of income.
Need better metrics for CPI / GDP / etc.

What work feels like

End state for new economy companies?
(Evil) monopoly?
Regulated monopoly?
Oligopoly?

10 principles for new economy, taking domestic workers as an example.
...

How to keep open source communities healthy
Paid v fully paid
Dedicated v casual
BDFL
Community grows iff casual participants can successfully commit code.
Culture, technology, process barriers

When to break up big modules?

Quantified place
What makes a space a place?
Interactive architecture
Chat rooms for furniture
Google interactive spaces project - pubs network for devices
Spacebrew
Hololens
Architectural Psychology
Experienceing Architectures
Christopher Alexander
Francis Duffy
Comfort in buildings
Sick building syndrome

Lunch
wirecutter.com - really great gadget reviews
thesweethome.com - really great DIY reviews
pseudonimity
Coral Platform @ Mozilla - new, collaborative tools for journalists

Data and design: What can they teach each other about creativity?
Data as window into experience and behavior
Shared mental models require interaction and experience
Numbers don’t move people, stories do.
3 questions for graphs:
What should viewers learn? do? feel?
Good data often elicits conversation, instead of “driving decisions."
Care = 1 / time.
Desire is the engine of story
Designers and data scientists often intimidate each other. This is bad. We need more collaborative teams and common language.
The best data products elicit behavior and data that improve themselves.
More than traditional software, data products bring (recorded, analyzable) data closer to human experience
Therefore, data from data products tends to be easier to interpret and work with?
Idiots are the best resource you have for user testing.
Outliers generate new questions
Bonus: probable things, not impossible things.
Outliers help define the edges of possibility.

How data hurts people
Undocumented migrants
Police “threat” lists of protestors
Uber’s policing of protests in China
SMS to protestors in Ukraine
Facebook stalkers - location gleaned from the posting of a friend of a friend
Random face identification of strangers in Facebook photos
Medical data
Misdiagnosis in medical data
Miscommunication
Overscreening
Small testing -> massive testing
unnecessary risky surgery
direct costs, recovery time, cost to well-being, lost wages
Data distraction
Overreacting to news from 23 and me
Genetic coercion
Wanted: a term like “uncanny valley” for data products that are kind of creepy.
“The machine knows more about me than expected."
Expectancy violation theory
Family falls apart after discovering half-brother through 23 and me
Ashley Madison breach
CBO breach
Wikileaks
WMDs in Iraq:
circumventing expert analysis
cherrypicking bits of data
process is opaque
Nathaniel …?
War crimes and mass graves
“What appears to be…” — essential linguistic cue
RadioShack auctioning user data
"Wiggle room in the evidence."

Data as radioactive waste
Atlantic article: Lies, Damn lies, and medical science

Projector mapping
Pomotion
Ultra-short throw projectors

TurboSquid - 3d models 
Read More

7.31.2015

Data science means shopping and plucking, not just cooking.

11 comments:
Many people think that data science works like this:


But that’s not the whole picture---not even close.

Unless your data pipeline is quite mature, your data is probably more like this.



Unstructured, uncleaned. Still very messy.

Your whole data set is probably more like this:




It contains some of the key ingredients, but not all of them.

I can’t make cookies out of mustard. And I can’t make them out of just chocolate chips and vanilla, either.



Most of the time, building great data products requires shopping and plucking, not just cooking. You can’t cook a great meal until your fridge is stocked with the right ingredients.

Bottom line: your "secret sauce" isn’t an algorithm. It a combination of data cleaning, processing, and curation—plus a judicious choice of the right algorithms.

(Even if you have the right ingredients, you can’t boil your way to good cookies.)




That means you want to work with data scientists who understand the whole process of shopping, plucking, and cooking good data products. If you hire analysts or machine learning specialists who don't know how to pluck and shop, you're going to either (1) get stuck baking mustard cookies, or (2) put a heavy burden on your engineering team to grab and process new data. (1) is yucky. (2) is very slow.

It also means that you don't want to constrain your data scientists to only use the ingredients you already have in your kitchen. You should expect a good data scientist to improve your options by looking for more ways to bring in more data. ("Hm. No eggs. Before we go any further, we're going to need some eggs." "These cookies are okay, but they'd be much better with a dash of cinnamon.")

Practically speaking, "more ways to bring in data" includes things like
  • additional instrumentation within your app/website
  • mashups with public data sources
  • feedback mechanisms within your app/website (e.g. additional profile fields)
  • hand-curated data sets to clean and normalize large data feeds
  • merging in additional sources of user feedback (e.g. customer support tickets)
  • user surveys or interviews
  • etc.

In conclusion, three cheers for cookies!



PS: I’m not saying you should wait for all the perfect ingredients to begin. Great data science usually involves smart sequencing—rapidly learning which data streams add the most value, and developing the systems to gather and process them effectively. Make sugar cookies for now, and add the chocolate chips as soon as you can get them.

PPS: Peter Norvig says that “more data usually beats better algorithms." I’m not disagreeing. Instead, I’m pointing out that at any given point in the life cycle of a data product, your volume of data is more or less fixed. Great data science is about working within that constraint, creating useful data products with the tools and ingredients that are close to hand, and bootstrapping yourself up to the next level.



PPPS: There’s another layer to this conversation: developing the tools (and culture) to enable rapid exploration and deployment of data products. It’s a bit like making sure your kitchen is equipped with a food processor, not just a microwave. But I think this metaphor is already strained enough, so we’ll save that conversation for another day.

Image credits



Read More