Article Preview
TopIntroduction
Over the years, healthcare information technology has rapidly advanced, and with its growth, has led to industries shifting their business practices towards accommodating more clinical services connected to the mainstream of the electronic health records. Incidentally, as the industry moves towards solving the interoperability bottleneck between different clinical data formats and silos, more focus is given to the integration of the clinical workflow as a mean for solving this bottleneck. Workflows are made up of a sequence of tasks, consuming resources, and achieving goals. So, it makes sense that workflow interoperability requires task interoperability.
When taking a step back and examine the movement towards workflow interoperability, studies indicate clinicians consider what sort of impact electronic healthcare records (EHR) will have on their typical workflow. It was found that if there is too much resistance by physicians due to the workflow being impacted too much, then the adoption rates will be poor (Bowens, Frye, & Jones, 2010). To mitigate this resistance, the authors use Cascading to extract, transform, and load (ETL) discharge summaries. The reason for using Cascading is that it is built upon the idea of streamlining workflows. What is meant by this is that Cascading offers the scalability of Hadoop but abstracts the complexities that come with MapReduce tasks. This creates an environment where developers adhere to new programming workflows, making it easier than ever to construct complex data processing workflows without knowing the underlying complexities of the MapReduce infrastructure. This is important to note because developers, just as clinicians, are hesitant to adapt to new technologies if their current workflow is disrupted. Using the proposed system, the resistance many clinicians may feel by adopting to EHRs and order sets can be removed.
As part of this research, the authors use Scalding, a Scala-based Cascading API, as the benefits of Scala and MapReduce can be combined to create an intelligent wrapper allowing seamless integration between a clinicians existing workflow and an EHR. A mix of Scalding and Scala is used to implement natural language processing (NLP) techniques such as Term Frequency-Inverse Document Frequency (TF-IDF) with cosine similarity to perform the extraction and transformation into a standardized medical messaging format developed by the Health Level Seven International (HL7) organization. In this article HL7's Fast Healthcare Interoperability Resources (FHIR) along with JavaScript Object Notation (JSON) is used as the primary medical messaging format. Using HL7's FHIR with JSON, loading the message into any EHR using Slick, a new database query and access library from Scala increases the interoperability of the proposed system.
In (Schnipper et al., 2008), researchers introduce the concept of smart forms, requiring the user to follow a set of standardized input fields meticulously. The authors implemented a similar concept, known as order sets, to showcase how inefficient the workflow becomes when a user must traverse pages of input fields manually. To overcome this inefficiency, and enhance the workflow, the authors bring forth the idea of transforming clinical narratives into a standardized format allowing full interoperability between systems. This allows population of the designed order set from semi-structured data, removing the requirement to fill out pages of data on a smart form or order set.
A workflow can be defined as an ordered procedure of tasks followed by an individual in the environment (Brixey, Robinson, Turley, & Zhang, 2010). In the context of clinical workflow, inferences can be made that the same definition applies to clinicians. In a chaotic live clinical environment, it becomes a challenge to adhere to every rule, regulation, and policy set in place. Clinicians are responsible for following a series of steps outlined by the organization they work in to ensure patient safety and health. Modelling an ideal clinical workflow becomes a complex challenge as structuring dynamic human behaviour is no easy feat. Analytical analysis of existing workflows can help prevent variabilities between clinical practices and improve patient outcomes (Rotter et al., 2008).