homeGeek CultureWebstoreeCards!Forums!Joy of Tech!AY2K!webcam

The Geek Culture Forums


Post New Topic  New Poll  Post A Reply
my profile | directory login | | search | faq | forum home
  next oldest topic   next newest topic
» The Geek Culture Forums   » Other Geeky Stuff   » Ask a Geek!   » Getting started with big data and machine learning

 - UBBFriend: Email this page to someone!    
Author Topic: Getting started with big data and machine learning
Stibbons
SuperBlabberMouth!
Member # 2515

Member Rated:
5
Icon 1 posted January 13, 2016 09:16      Profile for Stibbons   Author's Homepage     Send New Private Message       Edit/Delete Post   Reply With Quote 
Hey! Been doing a lot of lurking and not much posting recently so thought I'd pitch in with a question. Big Data and machine learning seems to be pretty sexy nowadays, and despite efforts to the contrary my research seems to be heading that way too. So I need to actually learn this stuff. The theory seems pretty comprehensible, but I can't find any decent courses online into actually doing any big data work. If you know of anything - especially using R - let me know please!
Posts: 1141 | From: UK | Registered: Dec 2003  |  IP: Logged
TheMoMan
BlabberMouth, a Blabber Odyssey
Member # 1659

Member Rated:
4
Icon 1 posted January 16, 2016 10:21      Profile for TheMoMan         Edit/Delete Post   Reply With Quote 
Stibbons, I have some questions about big data and tracking out-breaks. With the advent of faster data could smaller pockets of infections be tracked with better precision?

--------------------
Those who would give up essential liberty to purchase a little temporary safety deserve neither liberty nor safety.


Benjamin Franklin,

Posts: 5835 | From: Just South of the Huron National Forest, in the water shed of the Rifle River | Registered: Sep 2002  |  IP: Logged
Stibbons
SuperBlabberMouth!
Member # 2515

Member Rated:
5
Icon 1 posted January 27, 2016 12:11      Profile for Stibbons   Author's Homepage     Send New Private Message       Edit/Delete Post   Reply With Quote 
Hey MoMan, the problem isn't really faster data, it's getting access to the data in itself.

My work is all in the intensive care unit - we have more data coming off our patients than we know what to do with, but it's only available to us because we're in an ICU, and need it to guide clinical management. Yes, there are refinements we can make to track spread of e.g. drug resistance microbes in this environment quicker, but the core problem of getting the data isn't there.

Contrast this with trying to track an outbreak in the community. How do we get information on wellness of patients? It relies firstly on them presenting to a healthcare provider, which they may not until it's too late for their info to be useful. And even if they do, how do we get that data from the provider to those doing the data analysis? Given the fear of Big Data that the public has extends into their health records, it's notoriously difficult to access information which may be useful (see the Care.Data issues we had in the UK). This is one of the reasons some illnesses have a legal responsibility on the diagnosing clinician to report, because otherwise we have no right to that patient's data and can't track them.

Posts: 1141 | From: UK | Registered: Dec 2003  |  IP: Logged
TheMoMan
BlabberMouth, a Blabber Odyssey
Member # 1659

Member Rated:
4
Icon 1 posted January 27, 2016 13:48      Profile for TheMoMan         Edit/Delete Post   Reply With Quote 
Stibbons, I remember looking at Health Pro. magazines in the late sixties showing clusters of Cancers, that were located around heavy manufacturing areas. I knew of these areas and still went to work for those companies because of the wages that they paid. I got my money and Pension and left, only to be stricken with Prostate Cancer. Causation or effect dubious.

--------------------
Those who would give up essential liberty to purchase a little temporary safety deserve neither liberty nor safety.


Benjamin Franklin,

Posts: 5835 | From: Just South of the Huron National Forest, in the water shed of the Rifle River | Registered: Sep 2002  |  IP: Logged
dragonman97

SuperFan!
Member # 780

Member Rated:
4
Icon 1 posted January 27, 2016 20:32      Profile for dragonman97   Author's Homepage     Send New Private Message       Edit/Delete Post   Reply With Quote 
Stibbons,

Do you have an academic affiliation, or is this strictly a clinical facility?

The rules on patient privacy can make analysis beyond individual care tricky unless you have really good protocols for anonymization. Even so, it's probably best to have an expert opinion on how to pursue that. I certainly think it's cool to look for these patterns and make more meaningful use of the data, but I'm acutely aware of the privacy issues. If a hospital's data gets out...it's bad. As a friend has said - if your credit card info. is leaked, you can change it; if your medical history is leaked, you can't change it. (And screwy as it may seem, it could be valuable to some particularly nefarious characters.)

--------------------
There are three things you can be sure of in life: Death, taxes, and reading about fake illnesses online...

Posts: 9331 | From: Westchester County, New York | Registered: May 2001  |  IP: Logged
Stibbons
SuperBlabberMouth!
Member # 2515

Member Rated:
5
Icon 1 posted January 29, 2016 03:41      Profile for Stibbons   Author's Homepage     Send New Private Message       Edit/Delete Post   Reply With Quote 
Dman, bit of both. I'm using freely available anonymised US data currently to build models, then applying them to real-time data coming from my own patients in critical care. I'm luckily not struggling to acquire the data, I'm just lost in a wasteland of partial understanding of methodology!
Posts: 1141 | From: UK | Registered: Dec 2003  |  IP: Logged
quantumfluff
BlabberMouth, a Blabber Odyssey
Member # 450

Member Rated:
5
Icon 1 posted January 31, 2016 15:57      Profile for quantumfluff     Send New Private Message       Edit/Delete Post   Reply With Quote 
Don't mix R and machine learning. Try TenserFlow. https://www.tensorflow.org

I would also start by looking at any available history about Google Flu Trends and how they ended up being a able to see outbreaks before the CDC.

Posts: 2901 | From: 5 to 15 meters above sea level | Registered: Jun 2000  |  IP: Logged
Rednivek

Member # 1148

Icon 1 posted March 16, 2016 00:11      Profile for Rednivek     Send New Private Message       Edit/Delete Post   Reply With Quote 
For a really good intro to Machine Learning, check out the free course from Caltech. (156). Sometimes its interactive, but you can also check it out yourself on Youtube:

https://www.youtube.com/playlist?list=PLD63A284B7615313A

Posts: 714 | From: Detroit/Windsor | Registered: Feb 2002  |  IP: Logged


All times are Eastern Time  
Post New Topic  New Poll  Post A Reply Close Topic    Move Topic    Delete Topic next oldest topic   next newest topic
 - Printer-friendly view of this topic
Hop To:

Contact Us | Geek Culture Home Page

© 2015 Geek Culture

Powered by Infopop Corporation
UBB.classicTM 6.4.0



homeGeek CultureWebstoreeCards!Forums!Joy of Tech!AY2K!webcam