Design by data: the future of healthcare facility design

Using predictive analytics to design your future health facility will streamline the design process and outcomes — and save costs. 

Predictive analytics (PA) uses quantitative and qualitative clinical operations and financial data to analyse and optimise the current and future performance of a healthcare facility, thereby guiding strategic project investment and design decisions.

Historical and current state data and competitive market demand analyses, which provide insight into upcoming patient volume and type mix opportunities and challenges, are brought together in a common digital environment to create a virtual simulation model. The model can be used to review and rank-order hundreds of potential design and operational alternatives to maximise the entire solution space.

PA can be used on any renovation, upgrade or construction project and runs concurrently with the traditional project process. It can be introduced as early as the concept design stage but should not be used beyond the completion of schematic design. This is because it can often have far reaching design implications, so should be fully integrated by this stage.

Once a facility is operational, the model becomes a powerful reengineering tool, supporting continuous process improvement over time. The model can be used to simulate changes to existing processes, identifying their potential impact on operational performance and patient and client satisfaction in the virtual environment prior to physical implementation.

Figure 1: PA is underpinned by a rigorous quantitative and qualitative data collection process. Image credit: ©Jacobs

Taking the first step

The first step in any PA process is data collection and this can take anywhere from a single day to 3 months, depending on data availability, volume and integrity. Every inpatient department and outpatient clinic has its own set of requirements, so the initial data collection process can be extensive. All available data relating to physical space, materials management and transport systems, medical equipment, IT and clinical communication technologies, staffing models, patient scheduling and arrivals protocols is collected.

Complete datasets for every patient encounter for the last 12 months and, where possible, the last two years are gathered. Ideally, this will include detailed, time-stamped data for each element of each unique patient journey through the department being studied. It may also include data for up- and downstream departments whose performance impacts on the primary target department.

If data is not immediately available, a short-term data collection effort (two to eight weeks) can be completed to gather enough data to support the study. Sometimes, the first PA effort at an organisation uncovers the fact that the key metrics are not being gathered or reported. While this is an obvious impediment in the short term, it can be an impetus for change to better support ongoing operations.

The simulation model

Once the data collection phase is complete, a simulation model that captures all the work that occurs within and around a healthcare facility is created. The model is used to facilitate and inform decision-making in numerous ways.

The clinical and financial performance of preliminary design concept plans are assessed using the same key performance indicators that will be used to evaluate the actual built facility once it is operational. This eliminates the possibility of unmet expectations post construction and gives those responsible for large capital investment decisions a much greater degree of confidence and certainty.

While there is an initial cost collecting the data and creating the simulation model, once complete, hundreds and thousands of high-fidelity solutions can be tested rapidly, at no risk and very low cost, and without any construction or staff retraining expense.

Examples of US health facilities that reduced cost through PA

Example 1: University Health System San Antonio, Texas, USA PA eliminated unnecessary space, reduced the construction cost of a 44,000 SF clinic by US $623,000.

Example 2: Baylor Scott & White Health Dallas, Texas, USA PA increased the clinic patient throughput volume per exam room per day by 50% without increasing staff or clinic operating hours.

 

The model includes both first costs and lifecycle full time equivalent (FTE) costs, helping the project team make intelligent trade-offs between the two. When you consider initial construction costs represent between just 6.5% and 8% of lifecycle costs while staffing costs represent between 72% and 76%, optimising the latter is key to any long-term healthcare cost reduction strategies.

The data-driven approach that underpins data analytics can help decision-makers reach a consensus and break deadlocks faster than any other methodology because it is data driven, transparent and inclusive. Emotions and prejudices are removed from the equation and decisions can be made on fact.

Once consensus is reached, decisions tend to stick. PA is thorough and rigorous, leaving nothing for later, meaning decisions are made once and not revisited.

Longer term, PA offers an opportunity to implement a powerful, quantitative approach to continuous process improvement across time (existing, renovated, and new spaces) and to assure consistent and high-performing operations across space (a department, a facility, a campus and a system).

*David Morgareidge is Director of Predictive Analytics at Jacobs, a global professional services sector that aims to deliver solutions for a more connected, sustainable world.

Image credit: ©stock.adobe.com/au/VIGE.co