Visualizations for a smarter municipal ­community

Subscribe to ABB Review

Relying on user-centered design, ABB created intuitive data visuali­zations for two essential community systems: district ­heating and health-care. Promising results set the stage for ­further research into smart visualization and artificial intelligence.

Veronika Domova, Shiva Sander Tavallaey  ABB Corporate Research Västerås, Sweden, veronika.domova@se.abb.com, shiva.sander-tavallaey@se.abb.com

Advanced analytics and information visualization can foster awareness and a better understanding of a community’s industrial processes, thereby creating guidelines and contributing to increased productivity and energy savings. For example, well-designed visualizations can help uncover bottlenecks in the patient flow of a hospital, potentially increasing efficacy of the healthcare system. Effective data visualization of a district heating system can help industrial operators run the energy production process more efficiently, leading to more satisfied community residents while lowering regional environmental impact.

However, presenting essential data in a comprehensive visual form to the right users is a non-trivial task. Relying on collaborations with multiple stakeholders, ABB’s technology and domain experts, researchers utilized the user-centered design paradigm1 to develop several innovative visualizations for Swedish community hospitals and district heating systems that can do just that.

District heating and cooling (DHC) is an efficient, fuel-flexible, and sustainable way to produce and deliver energy in today’s market →1. In Sweden, for example, this popular heating process serves more than 50 percent of all homes [1].

01 ABB works closely with public and private stakeholders to explore digital visualization solutions that could help cities like Västerås, Sweden to increase efficiency and sustainability.
01 ABB works closely with public and private stakeholders to explore digital visualization solutions that could help cities like Västerås, Sweden to increase efficiency and sustainability.

ABB, a pioneering technology leader in automation and control systems, contributes to the efficiency of district heating systems with state-of-the-art SCADA (Supervisory Control and Data Acquisition) systems, intelligent pump control (IPC) for drives, and communication and interface solutions.

The control of a district heating production and delivery system is a consequential task: To control the district heating system, operators rely on process graphics and a myriad of numerical parameters. Dependence on weather conditions introduces uncertainties and unpredictability and non-user-friendly legacy interfaces, overloaded with numerical data and process graphics, do not alleviate the challenge. As a result, industrial operators often run the production process with their “gut feeling” or previous experience.

To research the design space of user-interfaces (UI) for industrial operators, ABB collaborated with several Swedish energy companies, Sweden’s independent state-owned research institute, RiSe, and others, eg, PiiA (Process Industrial IT and Automation Agency). Experts from ABB Corporate Research and RiSe employed the user-centered design paradigm 1 to investigate ways to transform runtime data of DHC processes to insightful and aesthetically pleasing visualizations to foster understanding of the DHC system status.

Initially, extensive field studies (interviews with operators, production planers, service technicians, environmental engineers, private customers and related companies) led to gains in domain knowledge. Multiple analytical techniques applied to collected data allowed researchers to identify operators’ challenges.

By conducting interactive workshops with operators and business stakeholders alike, ideas for potential solutions could be generated. Subsequently, iterative low- and high-fidelity prototyping occurred: designers could continually refine the development process through qualitative user-evaluation sessions in which feedback was incorporated in the development process.

Many ideas and concepts were considered initially; those that intended to help industrial operators obtain an instant overview of the district heating system status were, however, strongly favored. The development of such visualizations requires the aggregation of large amounts of data to form a meaningful summary represented visually as aesthetically expressive and pleasant metaphors and symbols.

The resultant pipeline visualization is a light-weight web solution that runs in a modern browser. It displays the three main constituents of a district heating system: production, distribution and consumption →2.

02 The display makes an attempt to visualize the cornerstone constituents of a district heating system: production, distribution and consumption.
02 The display makes an attempt to visualize the cornerstone constituents of a district heating system: production, distribution and consumption.

The visualization is intended to serve as the starting point of the operator’s workflows, who should then be able to quickly grasp whether the district heating system is performing properly or malfunctioning. Moreover, the operator can drill down into any aspect of the visualization by clicking on the corresponding UI elements, which will switch to a more detailed process representation, eg, process graphics or the city map view.

Several industrial operators who evaluated the pipeline visualization were clearly interested in the potential of future visual solutions and fascinated by the animation and use of bright colors. Although some individuals were skeptical that such a scheme could be useful in the near future, most could envision the visualization as an overview screen in the industrial control room. These preliminary results are indeed promising.

Quality health-care gains through visualization
The projected demographic boom and limited medical resources stress healthcare systems. In Sweden, as elsewhere, medical personnel dedicate time and effort to administrative and maintenance work, eg, paperwork, scheduling, cleaning and sorting medical instruments, working with contagious bacteria or hazardous substances, etc. [2]. Such necessary routine work is repetitive, monotonous and potentially dangerous. Time is squandered that is better spent elsewhere.

What if hospitals could find more efficient means to perform this work? ABB is convinced that robots and algorithms can improve organization to minimize the need for people to perform hazardous and monotonous tasks. Seizing on their expertise in robotics, automation and control systems, ABB explored visualization solutions that could free staff to do what they do best – provide expert medical care and service to patients.

AutoMed: Collaboration is the key
Initiated in 2015, AutoMed is a long-term collaborative medical domain research project funded by the government agency, Vinnova. ABB joined forces with two Swedish hospitals and multiple industrial and academic partners to investigate how process automation, optimization, scheduling, simulation and modeling approaches can lead to smarter resource allocation and patient flow organization.

In the scope of the collaboration, ABB researchers, due to the lack of real-time data, worked closely with Swedish hospitals to obtain historical data of patient flows to identify potential bottlenecks and develop a simulation tool capable of their prediction. A key feature for this purpose is the creation of a UI for extensive data exploration and analytics.

Furthermore, by visualizing waiting times, ie, how long patients have to wait before receiving the necessary medical service, staff could identify bottlenecks in the patient flows that result in overall medical care system delays. Unfortunately, it was impossible to obtain a complete picture of patients’ waiting times due to a lack of reliable data. Extra effort was, however, invested to approximate waiting times for each department by assessing mean times, standard deviation and identifying outliers, ie, patients who waited longer or shorter than normal.

ABB also explored visualizations that would help identify a unique process flow for specific diseases. Surprisingly, when a person receives a diagnosis, their treatment plans can vary from case to case, ie, the departments to be visited and procedures to be followed depend in some degree on the judgement and spontaneous decisions of the doctor(s) responsible for the patient’s case. It would be helpful to have a relatively common flow for each disease: currently the flow differs from patient to patient. Researchers found some commonalities and approximated, somewhat, such a flow for certain diseases.

Innovative visualization design for healthcare
A web-based portal with several views was designed and created to enable interactive medical data exploration: patient flow overview, patient cases overview and patient transitions between departments overview →3.

03 The main overviews are: patient flow overview, patient cases overview and patient transitions between departments overview.
03 The main overviews are: patient flow overview, patient cases overview and patient transitions between departments overview.

The patient flow overview is a hierarchical data visualization showing the structure of the hospital and how many acute/scheduled patients visited each department over a selected time →4. The interface allows filtering capabilities according to diagnosis- and time-related attributes. Each department can be expanded or compressed through mouse interaction. The spectrum of waiting times is displayed as histograms around each department (waiting time increases in a clockwise direction). The higher the column, the greater is the number of people that have waited. By hovering with the mouse over a column, a pop-up will appear that explicitly shows this information. Histogram bars are color-coded based on whether the waiting time is shorter, slightly longer or greatly exceeds an estimated threshold. By clicking on a bar, the user is redirected to the overview of the patient cases represented by the histogram column.

04 The patient flow overview display shows the hospital structure and number of patients being served. Users can filter according to diagnosis- and time-related attributes. Histograms provide visual representations of patient waiting time.
04 The patient flow overview display shows the hospital structure and number of patients being served. Users can filter according to diagnosis- and time-related attributes. Histograms provide visual representations of patient waiting time.

The overview of patient cases depicts the disease history of patients. Filters and an identification number allow a particular, yet anonymous, patient to be located. The timeline visualization of a disease shows flow in time, including diagnoses, departments visited, services received and waiting times →5,6. To ascertain why waiting occurred, the waiting period is correlated with the availability of necessary resources, eg, number of free beds during that time.

05 User-friendly visualizations with advanced data exploration and analytics capabilities could help hospital staff uncover the reasons for the formation of patient queues.
05 User-friendly visualizations with advanced data exploration and analytics capabilities could help hospital staff uncover the reasons for the formation of patient queues.

The transition diagram emphasizes the patient’s transitions between various departments during the treatment process. Insight is won by visually enlarging the departments and connections transited by many people. By hovering over the visual elements with the mouse, the user receives precise numerical information. The user can identify departments and transitions that are involved in a specific disease by applying filters →7.

06 The overview of patient’s display visualizes the timeline of a disease flow in time.
06 The overview of patient’s display visualizes the timeline of a disease flow in time.

Because the success of a visualization is measured by its usability, this web portal was demonstrated to medical personnel and tested by individuals hands-on. Preliminary feedback was positive: participants anticipated further system enhancements and were open to a continuation of the project.

07 The patient transition overview visualizes how patients move between the departments: the size of the department symbol, as well as the thickness of the connecting lines, is based on the number of visits.
07 The patient transition overview visualizes how patients move between the departments: the size of the department symbol, as well as the thickness of the connecting lines, is based on the number of visits.

Visualization challenges and future steps
Access to good quality data is essential for successful development of data-driven solutions, yet is challenging for many industries, including health care. Multiple data-related problems were faced in the scope of both projects; these included legal issues during data transfer, poor data quality, missing data, domain-specific jargon, etc. To amend these difficulties, extensive manual analysis and data processing was conducted in tight collaboration with domain representatives before ABB experts could forge ahead with actual project goals.

Having a large number of diverse stakeholders onboard both projects opened up collaboration opportunities and permitted consideration of a large range of perspectives. Conversely, research processes were occasionally hampered because different stakeholders had widely variant agendas.

During the course of field studies, ABB’s experts witnessed entrenched legacy working practices that tended to curtail acceptance of new solutions and methods. Nonetheless, industrial operators and medical staff were interested in the advantages of new visualization solutions and anticipated the possibility of increasing production efficiency in the future.

Inspired by the success of both projects, ABB researchers will continue working on effective big data visualizations. The next logical step introduces machine learning and artificial intelligence to provide the user with hints for the next most appropriate action to take.

Footnote
1) Please refer to “User-centered power network visualizations empower digital twins” article.

References
[1] Euroheat & Power. (2017, May 1). District Heating in Sweden webpage [Online]. Available under: www.euroheat.org/knowledge-hub/district-energy-sweden/

[2] J. Westbrook, et al. (2011, Nov. 24) BMC Health Service[Online]. Available under: www.ncbi.nlm.nih.gov/pmc/articles/PMC3238335/

Links

Contact us

Downloads

Share this article

Facebook LinkedIn Twitter WhatsApp