#HIMSS19 Social Media Ambassador six years in a row. Three HIMSS Davies Awards. Designed first undergraduate medical informatics program. EHR CMIO. Premed-Accountancy major (#1 ranked Illinois), Healthcare Systems Engineer (MSIE, Illinois), MS Intelligent Systems (Artificial Intelligence), ABD (All But Dissertation) Computational Linguistics (CMU). Dr. Workflow. King of All Workflow in Healthcare. The Workflow Bear. Owner of JETS! @HealthITDog and Maker of @MrRIMP (Robot-In-My-Pocket), both on Twitter!.
An event is a change in state, in a patient (blood glucose exceeds threshold), a population (percent of population whose average blood glucose exceeds threshold for specified duration); or even change in logistic state (patient entering waiting room, average patient wait time exceeding a pre-specified limit, or perhaps even an external event such as an admission, a discharge, a transfer).
No. Not those kinds of events.
Maximizing EHR productivity (value of information state divided by cost of information state) requires managing how EHR events trigger EHR workflows. Events in the numerator (info value) include those just listed. Events drive workflows. Examples include ordering a test, changing policies controlling test workflow, notifying whoever is next in a patient-specific workflow that a patient is waiting for a test, and policies controlling workflow escalation. Events in the denominator (info cost) include a big-data ocean of time-stamped EHR-initiated, and sensor-detected, events. Methods for extracting patterns and building models from this data are evolving, but process mining is one increasingly popular tool.
The more consistent, yet also transparent and flexible, the EHR workflow, the greater the potential for systematically improving EHR productivity. The best candidates for automating this unique combination of consistency, transparency, and flexibility include modern business process management suites and their cousins, adaptive case management systems. The best candidate for understanding and improving these resulting workflows is to feed insights of processing mining back into workflow design. Doing so will discover and correct non-conforming workflows and throughput bottlenecks, both necessary steps to systematically improve EHR productivity.
And this virtuous cycle, of improving EHR productivity, will not be possible without incorporating eventing systems.
I grabbed the best bits (fair use, mind you, indented and italicized), added some of my own thoughts, and provided links to additional resources.
“Historically, healthcare excellence has been achieved through individual practitioners focusing on being best in their medical discipline. The industry has evolved into a series of independent providers and processes, focusing on intervention. Focusing within each discipline does not provide consideration for the overall patient experience. Today, the healthcare industry is marked by its poor design, high fragmentation, and stunning inefficiency. Care Process Management (CPM) uses best practices for business process management to improve clinical outcomes without changing care processes or displacing the role of health workers.”
Strong words! But they are sentiments I hear frequently from healthcare IT folks too.
“Care Process Management” is essentially a rebranding of “Business Process Management” for healthcare. The basic motivation to rebrand BPM is that physicians don’t like to think of healthcare as a “business.”
I’ve written (and tweeted) a lot about business process management and healthcare. And I’ve thought about rebranding BPM. Several years ago I defined “clinical groupware” as a kind of brand extension of workflow management systems, business process management, and case management into healthcare.
By the way, natural language processing (NLP, another theme on this blog) faces the same dilemma. Clinical NLP vendors speak and write of both clinical language understanding and natural language processing in almost the same breath or sentence. Both BPM and NLP have stronger brands outside of healthcare than within healthcare. Both BPM and NLP have great promise in medicine (See my comments on IBM’s Watson). It will be interesting to watch this terminology evolve over the next several years.
“Care process management (CPM) is the application of business process automation and optimization techniques to clinical care processes in the healthcare environment. CPM uses best practices for BPM to improve clinical outcomes without changing care processes or displacing the role of health workers.”
“[B]est practices for BPM to improve clinical outcomes without changing care processes” is an interesting idea. I wrote about similar in my Automate Your EMR Cow Paths *and* Reengineer Them Too! blog post. The intuition behind this phraseology, which I share, is those information technologies that least disrupt existing clinical workflows will be most easily accepted and implemented. Many electronic health record and health information systems are criticized by users for unusable workflows and no means to improve them. One of business process management’s unique strengths is that its workflows are controlled by explicitly editable (by non-programmers) workflow plans, sometimes called workflow or process definitions.
Healthcare inefficiency and complexity
“[H]ealthcare ranked as the least efficient industry in the world, with more than $2.5 trillion wasted annually”
Holy. Cow. Would you look at healthcare way up there in the upper right! Healthcare is even more inefficient that education and government. By the way, in the lower left, the least inefficient (most efficient) are communication (which I totally get) and leisure/recreation/clothing (clothing I get, but it’d be interesting to go to the original report to find out what it means for leisure and recreation to be so efficient!).
“BPM is the means by which organizations improve their operations by leveraging internal expertise in new and scalable ways. This is achieved by directly engaging business people in the design, definition and creation of enterprise class process applications. BPM excels at providing comprehensive change management of business processes that results in continuous process improvement.”
“CPM is the strategic application of the BPM methodology to clinical care processes in the hospital/healthcare environment. The focus is on the patient and how care is delivered.”
“The traditional application of BPM to the hospital environment has been challenged because many healthcare practitioners do not see healthcare as a ‘business’ but as a ‘calling.’ Healthcare practitioners are not ‘servicing customers’ but ‘caring for patients.’ However, CPM does not change what a provider does to care for a patient. It focuses on process improvements – on coordination and collaboration between care providers.”
Here I’ll have to quibble (hey, I am very enthusiastic about this paper, but this wouldn’t be a blog post by me, if I didn’t quibble). I don’t think the issue is that clinicians don’t like the word “business.” They may, or may not, but that’s not what’s keeping business process management technology from more rapidly diffusing into healthcare. In fact, in Top Ten Reasons EHR-BPM Tech Is Not (Yet) Widely Deployed in Healthcare, I list the restraining forces, and none of them are about the word “business” versus the word “clinical.”
By the way, I otherwise agree with the characterization of BPM/CPM as a technology for improving processes. It’s consistent with my favorite short definition of business process management as essentially being about “process optimization process.”
Core IT functionality
“BPM engine – Care processes execute within a BPM engine. The engine implements the following functions that are critical to CPM:
Notifies care providers when a patient is admitted, when a consult is requested and when the patient is discharged
Creates and maintains an electronic version of the multi-disciplinary activity plan on behalf of the doctors and nurses
Manages assignments, reassignments and coverage of care activities on behalf of care providers
Can integrate with external systems to schedule medical procedures or equipment”
“Care process management offers a roadmap that will help the healthcare industry to transform delivery. Without changing what healthcare workers do best, it addresses the inherent inefficiencies of the care delivery process. Many aspects of care delivery can be improved using business process management to automate the coordination of care delivery and to enable providers to collaborate more effectively.”
While I may quibble about whether to call it care process management or business process management, I couldn’t agree more with the rest of the paper. I’m delighted to see BPM vendors beginning to make their best cases for use of BPM in healthcare. There’s remarkably alignment between disadvantages of current process-unaware EHRs and HIT systems and advantages of process-aware information systems: workflow management systems, BPM suites, and innovative case management systems.
If you’d like to read more about combining business process management and healthcare information technology, here are some of my past posts on the topic.
If you’d like to join a conversation about these topics I hope you’ll follow me on Twitter. We’re a growing band of BPM and health IT researchers and professionals learning from each other. I’ll gladly introduce you around!
AS (like PS, but comes first) Secure, flexible, scalable, context- and process-aware cloud-based backends will be key to secure, flexible, scalable, context- and process-aware front-end mobile apps used by patients and healthcare providers. While AWS re:Invent did not have a healthcare track, defacto cloud-computing “reference implementations”, such as those of Amazon, Microsoft, Google and others, will have profound impact on mobile health and medical app design.
Amazon Web Services is Amazon’s set of 30-plus cloud computing services.
Amazon Web Services fit together like LEGO building blocks.
“Spin up” an instance of a server (virtual hardware), operating systems (Linux or Windows), and an application server (Java, .NET, Ruby, PHP, Python). (Sounds like a DJ!) Upload your code and you are good to go: inexpensive (dollars vs thousands of dollars), scalable (10 millions users once a week for five minutes), agile (Obama campaign moved AWS-based 27-terabyte database, 500K transactions/second, 200 application infrastructure from East to West Coast night before Sandy), and reliable (they didn’t need to).
Catch the not-so-subtle reference to the Stables “That Was Easy!” button?
I once read a science fiction novel in which entire cities (where non-virtual biological humans lived) were actually generated by software. Imagine 3-D printing that doesn’t require a printer or cartridges and can create a a city the size of New York. (Yes, bugs caused human disasters. Look, that’s why it’s called science *fiction*.) City planners casually referred to “spinning up a suburb” in the same casual tone Amazonians speak of spinning up a a server to host some application or other.
A “bookseller” created the 10th wonder of the world, in the words of their CTO. 🙂 It’s a remarkable story, told dramatically well at the link in the tweet below.
The little Twitter bird and #AWSreInvent and #reinvent hashtags were omnipresent. By the way “swf” stands for “Simple Workflow”, Amazon’s workflow engine-in-the-cloud. We get to more about it later. Reinvention was a theme. Not just reinvention of IT, but reinvention of everything that depends on IT, which seems like just about everything these days.
I hope I’ve given you a vivid impression of the (nine)Teen(99) Spirit that pervaded the first Amazon Web Services user conference. Now let’s dig down in its substantive aspects relevant to health information management.
Is AWS Secure Enough for Patient-Identifiable Data?
I’m not a security expert. But I absolutely understand the need for security. As occasional developer, I prefer making wonderful things possible for good people than terrible things difficult for bad people. That said, I absolutely understand the need for security. I even said it twice, here, in the same paragraph. Amazon Web Services speakers were focussed and concerned to make sure we attendees understood and appreciated how important security is to them. For example, everything is encrypted in files and structured and unstructured databases. Transmission too.
Among almost a dozen security certifications and regimes I did catch sight of a logo for the Health Insurance Portability and Accountability Act. I found the following document, and quote, on the Amazon Web Services website.
Andy Jassy, SVP Amazon Web Services and Amazon Infrastructure, spent a lot of his keynote talking about security. Folks who (presumably) know a lot about security than me were tweeting things like “Very impressed with Cloud Security.”
RT @vince_riv: It’s obvious that b4 this year I had the wrong opinion about AWS security. Very impressed with Cloud Security. #reinvent
Hey! If the CTO of the CIA is impressed, I’m impressed. Here’s the article from which the quote apparently came from.
What Does AWS Bring to the EHR and Health IT Party?
Most of the benefits for healthcare of cloud computing are the same benefits for other industries and purposes. However, healthcare has some interesting cloud computing use cases worth noodling. Benefits of Cloud Computing
Replace capital expenses with operational expenses
Lower total cost of ownership
You don’t have to guess how many servers and software licenses to buy
Faster to develop, more easily modified, enables more innovation
Does routine stuff (the “undifferentiated muck”) for you so you can focus on your unique value proposition
Amazon Web Services are a wide and deep defacto reference implementation of a cloud computing ecosystem. One the great things about reference implementations is they lend themselves to “crisp”, rather than “fuzzy” definitions. The first sentence of Getting Started Guide: Analyzing Big Data with AWS defines big data simply and elegantly:
“Big data refers to data sets that are too large to be hosted in traditional relational databases and are inefficient to analyze using nondistributed applications.”
Of course, everything looks like a nail if you have a hammer (Amazon Elastic MapReduce, yes, Amazon has an EMR!). In that sense this is a self-serving definition. But that does not diminish its relative crispiness relative to a multitude of other definitions of big data that do not rely on specific software platforms and tools.
Speaking of self-serving (hey! “self”-“serving” is a pun! AWS has a bunch of servers. It’s a “self-serve” MIS department for infrastructure and platforms) how about this list of predicted increased skills demand: cloud service manager, cloud solution integrator, cloud solution architect, cloud service broker. Of interest to me, since this blog is about EHR workflow and business process management systems, is tweet refers refers to “business designer” and “business process architect” as “BPM”.
Why is the software development cycle some much faster? Cloud’s benefits: cheaper, more predictable, more flexible, scalable, etc. In some ways, cloud computing returns us to the era (I thought was gone) where a couple people in a garage can invent something insanely great and change the world. Innovation is much discussed in healthcare. The right programmer and the right doctor working together in a virtual garage are today’s version of this meritocratic ideal.
By the way, the day before AWS re:Invent I attended a daylong bootcamp.
Building Internet-Scale Web Apps using Elastic Beanstalk and Amazon RDS
“This hands-on course will teach you how to quickly and easily build, deploy, and manage highly available, Internet-scale web applications using Elastic BeanStalk. You will build a sample application that uses Amazon S3 for storage, Amazon CloudFront for edge caching with maximum performance, and uses real-time metrics to Auto Scale your application servers on EC2. You will also learn to use and operate a fully managed relational database using Amazon RDS. You will learn best practices for running high available databases, including seamless database failover, monitoring, data loading, backups/snapshots, horizontal and vertical scaling, tuning, and more.”
It was awesome. In retrospect I chose well. I probably would’t have been to attend the three or four sessions on the same topics the next two days. The lines were that long. If this was a tweet I’d end it with the hashtag #smug.
This is an amazing example of how scalable mobile solutions can be if they are cloud-based. Apparently there is a popular TV show in China that can be viewed on mobile phones. Once a week a push notification causes 10 million users interact with their phones vigorously for a short number of minutes. 10 million users, five minutes a week, the rest of the week: nothing. That’s scalable!
Operating System (provided by IaaS, provided by PaaS)
Virtualization (provided by IaaS, provided by PaaS)
Servers (provided by IaaS, provided by PaaS)
Storage (provided by IaaS, provided by PaaS)
Networking (provided by IaaS, provided by PaaS)
The combination of flexibility and scalability provided by Amazon Web Services PaaS allowed the Obama campaign to move their entire mission critical collection of 200 applications from the East Coast to the West Coast the day before Sandy hit the East Coast.
All this automagical creation of servers led to lots of humorous comments and tweets about the long lines at the men’s restroom (didn’t scale) and the remarkable conference food and beverage service (scaled).
If you are not a programmer you have no idea what excitement “Everything is programmable” causes in a developer. Programs are these fragile little ghettos of computation over which programmers are Lord and Master (or Lady and Mistress). Programs have all kinds of arbitrary limitations. They can’t start themselves. Though they can usually stop. Sometimes they can’t even access files in the surrounding file system due to security concerns. “Everything is programmable” means that not only can you clickity-clickity-click create servers and provision them with operating systems and databases and other wonderful things, you can make a computer program do that clicking for you. Now, obviously, with great power, comes create capacity to wreak havoc. So what? As long as you have the appropriate backups, you’re free to experiment and to break and to succeed in unusual ways.
RT @jinman: Werner #reinvent“Everything now is a programmable resource – Data centers, networks, compute, storage dataase, load balancers” — José Papo (@josepapo) November 29, 2012
@werner: Over 30% large scale it projects fails – because of faults in resource estimation. 50% that succeed over budget. #AWS#reinvent — Holger Mueller (@holgermu) November 29, 2012
30% IT projects fail? 50% over budget? Sound familiar? I’ve heard ballpark similar statistics for EHR implementations.
Werner #reinvent The commandments of 21st century architectures 1) Controllable, 2)Resilient, 3)Adaptive and 4) Data Driven #cloud — Jinesh Varia (@jinman) November 29, 2012
I’m familiar with a long list of “-awares” in computer technology: context-aware, process-aware, safety-aware, user-aware, to name a few. “Cost-aware” architecture really made me perk up my ears. I was a pre-med accountancy major. I took courses in cost accounting, economics of the firm, engineering economics, and the slides and terminology remind me a lot of those classes. At a very, very high-level, cost-aware software intelligently knows how to configure, behavior, monitor, etc, so as to minimize total cost of operation. At least that’s my take. Seems sort of relevant to healthcare, don’t you think?
There’s a popular saying in software, “It’s a feature, not a bug.” It’s usually said sarcastically, sardonically, even sadly. It acknowledges that all sophisticated software ships with bugs. The disingenuous side of this coin is the attempt to convince a user that the software really is supposed to work that way.
“Everything fails all of the time. Not just our stuff.” Werner ordering everyone to use at least two availability zones. #reinvent — James Gill (@jamesjgill) November 29, 2012
“Everything fails all of the time” and “Failure is just another form of deployment” is the ultimate “It’s a feature, not a bug” but in a good way. Instead of too expensive to be practical strategies to eliminate all bugs (or human error, for that matter), focus on building systems that tolerate bugs and error well. EHR design could learn from this philosophy.
S3 is Amazon’s Simple Storage Service. It’s a popular place to store webpages and media such as images, videos and sound files. HA stands for High Availability. If you multiple out times how many seconds in a year I get about 3 milliseconds (I might be off a digit or two, how many nines are in 99.99999999?). Of course a lot of other things not under Amazon’s control can go wrong between you and your blog. But still.
Here’s an overview of the major layers and categories of the 30 some Amazon Web Services.
I was going to create imaginative and dramatic health IT use cases, listing benefits for all the AWS services. For example, I was watching the movie Contagion last night and thought about time-to-deploy and scalability issues of supporting a medical app for use in a worldwide pandemic. Then, what if people didn’t just die, but they turned into zombies and began taking down data centers. That sort of thing. But this blog post is getting awfully long.
However, there is one particular Amazon Web Service that I’m particularly interested in…. Want to spend some time on…. It lives in the application services layer….
Chuck: Your Blog is Called EHR Workflow Management Systems…Well?
You’re right. This wouldn’t be a blog post in EHR Workflow Management Systems without a section on healthcare workflow. Amazon Simple Workflow is a workflow engine in the cloud that conveniently interoperates other Amazon Web Services. Recently I interviewed the founder and CTO of KISSFLOW, a workflow engine for Google Apps. That blog post, Healthcare Workflow-as-a-Service may be worth a review.
This is the workflow for finding lost packages. I’m sorry that is is not healthcare specific. But I don’t think it takes a lot of imagination to replace “pages” with medical tasks or even patients (got temporarily lost all the time at Cook County). This example has three potential steps (top). There are three activity workers. The first searches the warehouse (perhaps by querying databases, perhaps by assigning staff the job of a visual search). The second arranges for shipment of the package, if found. The third arranges to cut and send a check to the customer, if not found. There is one Workflow Worker (bottom), who job it is to coordinate the activity workers.
This talk on Amazon Simple Workflow was a tour-de-force. I believe it will be posted to Amazon Web Services YouTube channel, so I won’t rehash it. But I will touch on one portion of special interested to me. In the next slide, see the “00.21.32”? And the “1”, “2”, and “3”? And the file symbol containing lines 1, 2, 3? It’s the event log, which can be data-mined (called process mining) for workflow insight. I recently gave a presentation and published a proceedings paper on process mining of EHR event logs. The time-stamped data can be used to understand otherwise invisible workflow (often very different from what people think) spot stuff that shouldn’t happen but does (or should happen but doesn’t), and generally enhance workflows to make them more effective and efficient.
As debate about EHR usability turns to EHR design, medical error, and patient safety, EHR auditability will be increasingly important. Once problematic workflows are identified, the most effective way to improve them will be to improve the process definitions consulted by a workflow engine to drive workflow behavior. Speaking with representatives from Amazon Web Services and indicates that they are enthusiastic about the potential for EHR vendors and health IT systems to embed Amazon Simple Workflow into their applications. I enthusiastically and strongly endorse this use for Amazon Simple Workflow.
If you are interested in learning more about Amazon Simple Workflow these two slides provide links to a variety of documentation, communities, and videos. If you do anything interesting, I hope you’ll come back here. I’d be delighted subject you to one of my 10 Questions About X Plus a One-Minute Interviews.
By the way, Amazon Simple Workflow isn’t the only process-aware technology living in the AWS cloud. There’s also the Amazon Data Pipeline, which was introduced at a keynote. Using its graphical editor, you draw workflows from data sources, such as event logs, to various uses such as daily reports or weekly archiving. Big data meets workflow engines in the cloud.
One aspect of cloud computing I’m particularly enthusiastic about is the same aspect I’m particularly enthusiastic with respect to mobile. As cloud and mobile (and social too, but different future post) technology defuses into healthcare, it will bring along workflow engines-as-a-service. These will be quick to deploy, scalable, and reliable backends that make mobile apps smart about which screen and what to present to which user when, where and why. I wrote about this in Contextual Usability, My Apple iPad, and Process-Aware Clinical Groupware.
Amazon Web Services: Not Just a Fish Story
Not many years ago, if I read this blog post, I might have thought it was a real fish story. Scalability this big? Ran the line how fast?
But this was no fish story. And it didn’t get away. In fact, even if you want to avoid it, and you’re in IT, or health IT, it’s coming to get you. You might as well string the high-test line, strap yourself in, brace your feet against the back of the boat, and get ready for the ride of your life. (Ouch! I might have taken the fish metaphor too far! Wait, scalability, scales….)