An Update on the BPEL4People & WS-Human Task Standards

The BPEL specification focuses on business processes, the activities of which are assumed to be interactions with Web services with no additional prerequisite behavior. But the spectrum of activities that make up general purpose business processes is much broader. People often participate in the execution of business processes introducing new aspects, such as human interaction patterns. Workflow tools already cater for the orchestration of user interactions.

User interactions range from simple scenarios, such as manual approval, to complex scenarios where data is entered by the user. Imagine a bank’s personal loan process. This process is made available on the internet site of the bank using a web interface. Customers can use this interface to enter the data for their loan approval request and to start the approval process. The process performs some checks, and eventually informs the customer whether his or her personal loan request has been approved or rejected. Processing is often automatic and does not require any human involvement. However, there are cases that require bank staff to be involved. An example of such a case is if the online check of a customer’s creditworthiness returns an ambiguous result. In this case, instead of declining the request automatically, a bank clerk could check the request and determine whether to approve or decline it. Another example would be if a request exceeds the amount of money that can be approved automatically. In this case, a manual approval step is required, in which a member of the “approvers” group either approves or declines the request.

User interactions in business processes are not limited to approval steps. They also may involve data. An example of a user interaction that involves data is when an e-mail from an employer is manually attached to the process instance, or when the summary of an interview with an applicant is keyed into the process via a simple form or custom-built application.

To support a broad range of scenarios that involve people within business processes, a BPEL extension is required.

BPEL4People is defined in a way that it is layered on top of the BPEL language so that its features can be composed with the BPEL core features whenever needed. We envisage that additional BPEL extensions may be introduced which may use the BPEL4People extension introduced here.

BPEL4People is the WS-BPEL Extension for People as proposed in a joint white paper by IBM and SAP in July 2005.

In June 2007, Active Endpoints, Adobe, BEA, IBM, Oracle and SAP published the BPEL4People and WS-HumanTask specifications as a follow-up to the whitepaper, describing how human interaction in BPEL processes can be performed.

The OASIS WS-BPEL Extension for People (BPEL4People) Tecnical Committee is working on standardizing the BPEL4People and WS-HumanTask specifications.

Click here for a very engaging podcast that describes the inner workings of Technical Committee’s (something you usually don’t hear much about), describes the work the OASIS TC has recently accomplished and articulates the grand vision for business process management (BPM) and workflow that the committee has been working on.

I strongly encourage you to listen to this podcast. You’ll hear how some the of most important thought-leaders in the IT world, including IBM, SAP, Oracle, Microsoft, TIBCO and Active Endpoints are discussing BPEL and BPEL4People.

SOA, Web Services, BPEL, Human Workflow, User Interaction and Healthcare Systems

Lack of integration among legacy healthcare systems and applications means a continued reliance on manual processes that can introduce high risk errors into critical medical data. And isolated systems can compromise a provider's ability to follow an individual patient's care seamlessly from intake to treatment to aftercare.

While healthcare providers recognize that integration can help them achieve better service levels, many have been reluctant to proceed because of the critical nature of healthcare systems. But the approach to integration need not be a radical one of system rip and replace, nor does it have to precede through the development of system-by-system integration solutions.

Service Oriented Architecture (SOA) is a standards-based approach to integrating IT resources that can enable you to leverage existing assets, while at the same time building an infrastructure that can rapidly respond to new organizational challenges and deliver new dynamic applications. The SOA approach can help free application functionality from its underlying IT architecture and make existing and new services available for consumption over the network.

To derive a new value from existing services and go beyond simple point-to-point integration, you will need to combine and orchestrate these services in a business process. You will want to connect them in a coordinated manner, for example, have the result(s) from one service be the input to another service and have branching logic based on the outcome. Of course, you can use Java, C#, or another programming environment to call the services and manage the processes and data, but there is an easier, declarative way.

BPEL

An important standard in the SOA world is BPEL, or Business Process Execution Language, which serves as the glue to tie SOA-based services (Web services) together into business processes -- at least the portions that can be automated. The resulting BPEL process can itself be exposed as a Web service, and therefore, be included in other business processes.

The BPEL standard says nothing about how people interact with it, but BPEL in the Oracle Inc. BPEL Process Manager (to be discussed in my next post) includes a Human Workflow component (shown in the figure below) that provides support for human interaction with processes.



BPEL and User Interaction

I began an introduction to BPEL and human workflow towards the bottom of my December 14 post. Click here for a good deal more on this topic.

Humans can be involved in business processes as a special kind of implementation of an activity. To facilitate this, a new BPEL activity type called human task is required. From the point of view of the BPEL business process, a human task is a basic activity, which is not implemented by a piece of software, but realized by an action performed by a human being. In the drag-and-drop design pallet shown in the figure above, the actor of a human activity can be introduced into a BPEL process by using your mouse. A human activity can be associated with different groups of people, one for each generic human role.

People dealing with business processes do so by using a user interface. When human activities are used, the input data and output data must be rendered in a way that the user can interpret. More on this in upcoming posts.

Human resolution or disambiguation -- Integrating human workflow in BPEL processes -- Errors in statistical matching of attributes to an individual

To locate health records, statistical matching attempts to string together enough identifying information about an individual to substitute for a unique personal identifier. It involves matching attributes, such as last name, first name, birth date, address or zip code, and gender, and it may use medical-record numbers and all or part of the Social Security number.

The problem with personal attribute keys such as name and address is that they are usually not unique to the individual, change over time, and are often entered into different systems in different formats. And data-entry errors, such as misspellings, add to the difficulties with this type of key. Repeated collection, distribution, storage, and use of these data also represent an important identity-theft risk.

Statistical matching can attempt to correct for some of these changes and errors: The most straightforward process is to tag all of the near matches for human resolution, or disambiguation. Such disambiguation imposes significant costs and operational inefficiencies, particularly if the physician must resolve the ambiguities. Advanced approaches “score” matches on “closeness” to the input set. Those with a high score may be accepted as a match. However, all such efforts are subject to the probabilistic errors inherent in statistical matching systems.

As discussed in earlier posts, there are two types of errors - false positives, in which two different persons’ records are declared to be a match, which can lead to such errors as the wrong patient’s health data being obtained; and false negatives, in which two records for the same person are thought to relate to different people, leading to such consequences as some of the patient’s data being excluded. Both of these errors can lead to serious medical errors, waste (e.g., repeats of tests or the wrong tests), and considerable deviation from the promises of continuity and quality of care postulated for a connected digital health care system.

Disambiguation is a process through which multiple potential identification matches are further parsed until the patient can be matched with his or her data with sufficient certainty to allow for the delivery of a health service with reasonable confidence. The complexity of disambiguation varies according to factors such as the number of potential matches and the type of information available for further analyses. When sufficient digital data are not available to further differentiate potential matches, automated disambiguation may not be possible and may require human involvement. This last case will be the focus of the rest of this post and my next post.

Disambiguation entails implementing significant new workflows and may require substantial time and resources. When human involvement is required, many of the potential benefits of automation are lost. For example, at the point of care, disambiguation is often done by asking the patient further questions regarding personal characteristics and/or health care history. In some situation, disambiguation may not be possible, as when the patient is not present and information needed to further facilitate matching may not be accessible.

I will show how one vendor, Oracle, implements human tasks that can provide workflows such as those identified above. However, these Oracle services can be accessed by applications created with development tools from other vendors (my discussion will use Microsoft Visual Studio).

Introduction to BPEL and Human Workflow

Business Process Execution Language (BPEL), one of the key technologies for Service Oriented Architecture (SOA), has become the accepted mechanism for defining and executing business processes in a common vendor-neutral way. Apropos of this discussion, business processes often require human interactions as well.

User Interaction in Business Processes

BPEL business processes are defined as collections of activities that invoke services. BPEL doesn't make a distinction between services provided by applications and other interactions, such as human interactions. And that's important since real-world business processes often integrate not only systems and services but also users. User interactions in business processes can be simple, such as approving certain tasks or decisions, or complex, such as delegation, renewal, escalation, nomination, or chained execution . . . and matching an ID with an individual.

Task approval is the simplest and probably the most common user interaction. In a business process for opening a new account, a user interaction might be required to decide whether the user is allowed to open the account. If the situation is more complex, a business process might require several users to make approvals, either in sequence or in parallel. In sequential scenarios, the next user often wants to see the decision made by the previous user. Sometimes, particularly in parallel user interactions, users aren't allowed to see the other users' decisions. This improves the decision potential. Sometimes one user doesn't even know which other users are involved - or whether any other users are involved at all.

A common scenario for involving more than one user is workflow with escalation. Escalation is typically used in situations where an activity doesn't fulfill a time constraint. In such a case, a notification is sent to one or more users. Escalations can be chained, going first to the first-line employees and advancing to senior staff if the activity isn't fulfilled.

Sometimes it's difficult or impossible to define in advance which user should perform an interaction. In this case, a supervisor might manually nominate the task to other employees; the nomination can also be made by a group of users or by a decision-support system.

In other scenarios, a business process may require a single user to perform several steps that can be defined in advance or during the execution of the process instance. Even more complex processes might require that one workflow is continued with another workflow.

User interactions aren't limited to approvals; they can also include data entries or process management issues, such as process initiation, suspension, and exception management. This is particularly true in long-running business processes, where, for example, user exception handling can prevent costly process termination and related compensation for those activities that have already been successfully completed.

As a best practice for human workflows, it's usually not wise to associate human interactions directly with specific users; it's better to connect tasks to roles and then associate those roles with individual users. This gives business processes greater flexibility, letting any user with a certain role interact with the process and enabling changes to users and roles to be made dynamically.

BPEL and User Interaction

So far we've seen that user interaction in business processes can get quite complex. Several vendors today have created workflow services that leverage the rich BPEL support for asynchronous services. In this fashion, people and manual tasks become just another asynchronous service from the perspective of the orchestrating process and the BPEL processes stay 100% standard.

In my next post, I’ll talk about some of the specifics of how you might implement a BPEL process that includes human workflow/tasks for disambiguation using tools such as Oracles JDeveloper and Microsoft Visual Studio. The next two figures are meant to give you a preview of that discussion.

{ click the figures for a larger view }


Click here for more on Microsoft Visual Studio 2010

Click here for more on Oracle BPEL and Human Workflow

EHR, interoperability, etc. -- Informatics set to music

Human Resolution or Disambiguation -- False Positive and False Negative Identification: Heath Care and Information Technology Perspectives

Disambiguation of IDs is the process of resolving multiple potential matches into a match with the correct person. In general, statistical matching algorithms are likely to require substantially more-frequent disambiguation compared to that required by a system that uses theoretically perfect universal IDs; often, disambiguation is done by human intervention. Such disambiguation imposes significant costs and operational inefficiencies, particularly if, for example, a physician must resolve the ambiguities.

Note 1: Many of the efficiency and safety benefits theoretically possible with health information technology (HIT) systems depend on eliminating such human involvement and its concomitant slowness, expense, and propensity for error.

Note 2: What follows applies to IDs in general, even though I’ve chosen the healthcare industry for much of this discussion.

Disambiguation sometimes entails implementing significant new workflows that may require substantial time and resources. When human involvement is required, many of the potential benefits of automation are lost. For example, at the point of care, disambiguation is often done by asking the patient further questions regarding personal characteristics and/or health care history.

The potential for error in the statistical matching methods (see my December 1 post on unique patient IDs) has important safety implications, which are a chief concern for many in the health care profession. Two types of errors are involved in statistical matching: false positives, in which there is a link to the wrong patient’s records, and false negatives, in which not all of a patient’s records are found. A graphic representation of these types of errors and of how they relate to the probabilities and threshold for matching is shown in the figure below.



The horizontal scale shows the score of a particular match. As more and more attributes match and as the match is weighted by its score, or value, the higher is the probability that the patient is correctly matched to that record. A low score indicates a low probability of match (and a high probability that it does not match). It is possible to use a threshold above which the record is assumed to match and below which it is not assumed to match, which leads to the shaded areas above and below the threshold.

The area shaded to the right of the threshold is the region corresponding to false positives, or picking up the wrong patient’s records. The shaded area to the left of the threshold is the region of false negatives, or the records of the patient that are not picked up because of some non-matching personal attributes. Setting a balance between the two types of errors involves tuning.

Another approach illustrated in this figure is to define a region of ambiguity within which possible matches are tagged for human resolution, or disambiguation. Whether matching uses a single threshold or two thresholds, it is not possible to avoid encountering false-positive and false-negative matches. Adjusting the threshold or thresholds can result in a different proportion of false-positive and false-negative errors, but cannot be used to eliminate them because they result from the inherent characteristics of the population that lead to the two S-shaped curves.

As stated above, many end-to-end business processes require human interactions with the process.

Task Assignment and Routing

Human workflow supports declarative assignment and routing of tasks. In the simplest case, a task is assigned to a single participant (user or group). However, there are many situations in which more detailed task assignment and routing is necessary (for example, when a task must be approved by a management chain or worked and voted on by a set of people in parallel, as shown in the figure below). I’ve chosen tools in the Oracle SOA Suite to illustrate (in the figures below) human workflow that can provide declarative pattern-based support for such scenarios.





I’ll briefly elaborate here with an introduction to human workflow and continue the discussion in my next post, where I'll talk about how you might implement such a system.

Participant Type

In simple cases, a participant maps to a user, group, or role. However, workflow supports declarative patterns for common routing scenarios such as management chain and group vote. The following participant types are available:

Single approver

This is the simple case where a participant maps to a user, group, or role. Since at least one human being is involved, much more than his or her looking at a monitor screen and clicking with a mouse is involved.

For example, a vacation request is assigned to a manager. The manager must act on the request task three days before the vacation starts. If the manager formally approves or rejects the request, the employee is notified with the decision. If the manager does not act on the task, the request is treated as rejected. Notification actions similar to the formal rejection are taken.

Parallel

This participant indicates that a set of people must work in parallel. This pattern is commonly used for voting.

For example, multiple users in a hiring situation must vote to hire or reject an applicant. You specify the voting percentage that is needed for the outcome to take effect, such as a majority vote or a unanimous vote.



Serial

This participant indicates that a set of users must work in sequence. While working in sequence can be specified in the routing policy by using multiple participants in sequence, this pattern is useful when the set of people is dynamic. The most common scenario for this is management chain escalation, which is done by specifying that the list is based on a management chain within the specification of this pattern. More on routing later.

FYI (For Your Information)

This participant also maps to a single user, group, or role, just as in single approver. However, this pattern indicates that the participant just receives a notification task and the business process does not wait for the participant's response. FYI participants cannot directly impact the outcome of a task, but in some cases can provide comments or add attachments.

Readers who are interested in learning more about the subject of human resolution or disambiguation in an otherwise automated system might look at the following two books, while waiting for my next post.

Costs and Benefits of a Unique Patient Identifier for The U.S. Health Care System

In the healthcare industry, misidentification errors are not restricted to diagnostics and therapeutics but also may affect documentation. So, my earlier posts on semantics, ontologies, interoperability and the like notwithstanding, all is for naught when a given document doesn't provide information about a given patient. A chain is only as strong as its weakest link and patient identification is usually the first link in the healthcare chain.

Complicating the issue, not everybody can participate to the same degree or in the same way in the process of identifying a patient uniquely. Neonatal and senile patients are two groups where health providers and technology are on their own, when it comes to identifying the patient. Naturally, readers of this post fall into neither of these groups.

See, for example, Patient Misidentification in the Neonatal Intensive Care Unit: Quantification of Risk at

http://pediatrics.aappublications.org/cgi/reprint/117/1/e43.pdf

which provides a rather thorough study of errors in the first of these three groups.

The information that is used routinely for patient identification is frequently similar but often not recognizably unique.

In my November 20, 2009 post, Biometric and Other Identification Technologies, I discuss some leading technologies.

Although widely touted as “great” in security circles, all biometric devices (i.e., fingerprint, palm outline, iris, retina, et al) used for unique identification produce false positives and false negatives.



For example, an episode of Fox's "24" last season showed a White House visitor placing her thumb on a fingerprint scanner, a type of screening that is not typically used at the White House.

Fingerprint: false positives or negatives with scars, calluses, cracks in the skin, dirt, household cleaners and other variables.



Retina scan: susceptible to diseases such as glaucoma.



At the same time, non-biometric technologies have their own sources of error.



For a widely discussed examination of the costs and benefits of a unique patient identifier for the U.S. health care system, see

http://www.rand.org/pubs/monographs/2008/RAND_MG753.pdf

This recent study says using unique patient identification numbers for U.S. citizens would reduce medical errors, make electronic health records simpler and protect privacy.

The study says that despite a potential cost of $11 billion to create unique patient ID numbers, the effort "would likely return even more in benefits to the nation's health care system."

Most health care systems use statistical matching to find EHRs, according to the study by RAND Health, a research division of the RAND Corp. Statistical matching looks for demographic information, including names, birth dates and all or part of Social Security numbers.

See my November 17, 2009 post, Unique Patient Identification Numbers, Electronic Heath Records (EHR), Electronic Medical Records (EMR), and Social Security Numbers (SSN).

RAND researchers, who reviewed past studies, said that method causes errors or incomplete results about 8% of the time and leaves patients more exposed to privacy breaches.

"Assuming every health care system would have these [ID] numbers, then you'd be more likely to pick up all of the person's information," said Richard Hillestad, PhD, the study's lead author. "It would certainly make a lot of things easier."Using demographic information to locate EHRs causes errors or incomplete results about 8% of the time.

But critics expressed concerns.

"It's an absolutely terrible idea," said Deborah Peel, MD, a psychiatrist and chair of the Patient Privacy Rights Foundation, a watchdog group based in Austin, Texas. "Any database that has these numbers is bound to be a treasure trove for identity thieves."

The study was funded by a group of health information technology and IT companies, but Hillestad said that didn't influence the outcome. Dr. Peel is skeptical. "The combination [of data] is really deadly," she said. "That's why I say this is a data miner's dream."

The American Medical Association advocates prohibiting the sale and exchange of personally identifiable health information for commercial purposes without a patient's consent. The AMA also advocated in 1999 in favor of legislative action to repeal the portion of the Health Insurance Portability and Accountability Act of 1996 that mandated use of a unique patient identifier.

Hillestad said privacy is a big issue, but touted the ID numbers as a security boost.

"You're not sending all of the name and demographic information through the line to get connected," he said. "[Privacy] would depend on how much you protect the numbers."