[Reblog] Will Getting More Granular Help Doctors Make Better Decisions?

From the 8 February 2015 post at The Health Care Blog

Will Getting More Granular Help Doctors Make Better Decisions?

Excerpt (longish post)

But, there are many things that data will never do well.  For certain things, physician heuristics may lead to better decisions than any predictive model.

Heuristics are shortcuts, based on experience and training that allow doctors to solve problems quickly.  They are pattern maps that physicians are trained to recognize. But, heuristics have a reputation for leading to imperfect answers: Wikipedia notes that heuristics lead to solutions that “(are) not guaranteed to be optimal, but good enough for a given set of goals…. (they) ease the cognitive load of making a decision.”  Humans use them because we simply can’t process information in sequential binary fashion the way computers do.

It would be a mistake to call heuristics a sad substitute for big data.  Some cognitive scientists have made the argument, and I think they’re right, that heuristics aren’t simply a shortcut for coming to good-enough answers. For the right kinds of problems, heuristically generated answers are often better than the those generated by computers.

How can this be?

I often think of the following cartoon in Randall Munroe’s superb recent book, What If? Serious Scientific Answers to Absurd Hypothetical Questions.  In trying to compare human and computer thinking, he rightly notes that each excels at different things.  In this cartoon, for example, humans can quickly determine what they thought happened.  Most people can tell you that the kid knocked over the vase and the cat is checking it out, without going through millions of alternate scenarios.  Monroe notes that most computers would struggle to quickly come to the same conclusion.

So, from the perspective of an emergency doctor, here are the three leading problems with the applied use of complex analytics in the clinical setting:

• 1. The garbage in, garbage out problem.  In short, humans regularly obfuscate their medical stories and misattribute causality. You need humans to guide the patient narrative and ignore red herrings.

• 2. If we want to be able to diagnose, screen and manage an ER full of runny-nosed kids with fevers, we simply can’t afford the time it takes for computers to sequentially process millions of data points. The challenge is at one simple and nuanced: allowing 99% of uncomplicated colds to go home while catching the one case of meningitis. It’s not something that a computer does well: it’s a question of balancing sensitivity (finding all true cases of meningitis among a sea of colds) and specificity (excluding meningitis correctly) and doctors seem to do better than computers when hundreds of cases need to be seen a day.

• 3. There is a problem with excess information, where too much data actually opacifies the answer you’re looking for. Statisticians call this “overfitting” the data. What they mean is that as you add more and more data points to an equation or regression model, the variability of random error around each point gets factored in as well, creating “noise”. The more variables, the more noise.

The paradox is that ignoring information often leads to simpler and ultimately better decisions.

…

Related articles

• It takes more than great algorithms to make personalization work
• Sanjay Bakshi: Daniel Kahneman vs Gary Klein

[Journal Article] The Emergent Discipline of Health Web Science -with related links and articles

Larger image –>http://www.flickr.com/photos/40726922@N07/4702688723

Came across this article through an online professional health community.  It describes how the Internet is changing approaches to healthcare issues.  Current evidence shows Web sites can empower professional and lay alike through informational Web pages, social media, health record annotations and linkages for exploration and analysis. However, these applications can be built on to better serve the health care related needs of all.  The Web can be better” engineered for health research, clinical research, and clinical practice. In addition, it is desirable to support consumers who utilize the Web for gathering information about health and well-being and to elucidate approaches to providing social support to both patients and caregivers. Finally, there is the motivation to improve both the effectiveness and efficiency of health care.” The paper goes on to outline channelling further efforts in these areas.

• Social networks
• Patient Engagement Through Citizen Science and Crowdsourcing
• Sensors, Smart Technology and Expert Patients
• “Big Data”, Semantic, and Other Integration Technologies
• Rapid, Automated, Contextualized Knowledge Discovery and Application

From the full text of the article

Abstract

The transformative power of the Internet on all aspects of daily life, including health care, has been widely recognized both in the scientific literature and in public discourse. Viewed through the various lenses of diverse academic disciplines, these transformations reveal opportunities realized, the promise of future advances, and even potential problems created by the penetration of the World Wide Web for both individuals and for society at large. Discussions about the clinical and health research implications of the widespread adoption of information technologies, including the Internet, have been subsumed under the disciplinary label of Medicine 2.0. More recently, however, multi-disciplinary research has emerged that is focused on the achievement and promise of the Web itself, as it relates to healthcare issues. In this paper, we explore and interrogate the contributions of the burgeoning field of Web Science in relation to health maintenance, health care, and health policy. From this, we introduce Health Web Science as a subdiscipline of Web Science, distinct from but overlapping with Medicine 2.0. This paper builds on the presentations and subsequent interdisciplinary dialogue that developed among Web-oriented investigators present at the 2012 Medicine 2.0 Conference in Boston, Massachusetts.

Read the entire article here

Related links

• Health Web Science Lab – “A Hippocratic Revolution in Medicine”

The Health WebScience Lab is a multi-disciplinary research initiative between Moray College UHI, NHS Grampian, HIE OpenFinder and Sitekit Solutions Ltd based in the Highlands of Scotland committed to improving health locally, nationally and internationally.

This initiative will lead, connect and collaborate on research in the emerging discipline of WebScience and Healthcare to create communities which take responsibility for their own wellbeing and self-care. This will be achieved through the application of information and other communication technologies via the internet across a whole range of functions that affect health care thereby stimulating novel research between health care professionals, the community at large and industry.

• Health Web Science Twitter account

studies ” the effects of the interaction of healthcare with the web, and of the web with healthcare” and how one can be effectively harnessed to change the other

• 3rd International Conference on Web Science (Program with many links to presentations)

Related articles

• Participatory Medicine 2.0 (projecthealthdesign.typepad.com)
• HealthCamp Boston 2012: Brainstorming the Future of Health Care (healthblawg.typepad.com)
• Semantic Web (pantypeblog.wordpress.com)
• Cooperation in Health: Mapping Collaborative Networks on the Web (plosone.org)
• Text Analytics and Semantic Processing Fuel New Web Paradigm (arnoldit.com)
• What is WebScience? (mymindbursts.com)
• Medicine 2.0’13: 6th World Congress on Social Media, Mobile Apps and Web 2.0 in Health, Medicine, and Biomedical Research (walterfarah.net)

Better medicine, brought to you by big data through new types of data analysis

 

A good overview of how improved data analysis and presentation is improving health care delivery.

I especially liked the slideshare presentation found below in Related Articles.
The 42 slides in Big data – a brief overview outlines what big data is, its sources and processes, how it is analyzed, current “players”,examples, market analysis, future, and opportunities.

From the 15 July 2012 blog post at Gigaom

Slowly but surely, health care is becoming a killer app for big data. Whether it’s Hadoop, machine learning, natural-language processing or some other technique, folks in the worlds of medicine and hospital administration understand that new types of data analysis are the key to helping them take their fields to the next level.

Here are some of the interesting use cases we’ve written about over the past year or so, and a few others I’ve just come across recently. If you have a cool one — or a suggestion for a new use of big data within the healthcare space — share it in the comments:

Genomics. This is the epitomic case for big data and health care. Genome sequencing isgetting cheaper by the day and produces mountains of data. Companies such asDNAnexus, Bina Technologies, Appistry and NextBio want to make analyzing that data to discover cures for diseases faster, easier and cheaper than ever using lots cutting-edge algorithms and lots of cloud computing cores.

BI[definition of business intelligence] for doctors. Doctors and staff at Seattle Children’s Hospital are using Tableau to analyze and visualize terabytes of data dispersed across the institution’s servers and databases. Not only does visualizing the data help reduce medical errors and help the hospital plan trials but, as of this time last year, its focus on data had saved the hospital $3 million on supply chain costs….

..Semantic search. Imagine you’re a doctor trying to learn about a new patient or figure out who among your patients might benenfit from a new technique. But patient records have been scattered throughout departments, vary in format and, perhaps worst of all, all use the ontologies of the department that created the record. A startup called Apixio is trying to fix this by centralizing records in the cloud and applying semantic analysis to uncover everything doctors need, regardless who wrote it…

..Getting ahead of disease. It’s always good if you figure out how to diagnose diseases early without expensive tests, and that’s just what Seton Healthcare was able to dothanks to its big data efforts…

and more!

Related articles

• Better medicine, brought to you by big data [GigaOM] (gigaom.com)
• Intel and NextBio seek Big Data upgrades in genomics (fiercebiotechit.com)
• Big Changes Are Ahead For The Health Care Industry, Courtesy Of Big Data (fastcompany.com)

• Big data – a brief overview (slideshare.net) [a slide presentation, 42 slides]

• Oracle adopts the popular R language for the enterprise and big data. (oracle.com)
• Presentation: Scalability Challenges in Big Data Science (architects.dzone.com)
• Salesforce intros Radian6 Insights for social big data (zdnet.com)
• Big Data Modeling – Part I – Defining “Big Data” and “Data Modeling” (infocus.emc.com)
• NextBio and Intel Announce Collaboration to Optimize Use of Hadoop Stack And Move Forward With Big Data Technologies in Genomics (ducknetweb.blogspot.com)
• A Beautiful Friendship: Big Data and Social Media (blogs.sap.com)
• Stanford rides Big Data wave in medical research (fiercebiotechit.com)
• Big Data? Big Deal! (clean-clouds.com)