As Australia goes through the inevitable digitisation of hospitals, moving away from paper into the digital environment, data-driven approaches are key to informing system evaluation and redesign, improving flow through hospitals and delivering improved patient outcomes. 
CSIRO Australian E-Health Research Centre Principal Research Scientist Justin Boyle told HITNA that this digital environment is a “gold mine” for the industry in terms of improving hospital operations and offering better, more personalised patient care. 
“Data analytics is not just icing on the cake as a nice-to-have, but in the current climate of value-based care and performance frameworks in hospitals, particularly in the areas of safety, quality and access in the face of a tsunami of patient demand, there is an urgent need for analytics-based solutions to deliver high value care,” he said. 
According to Boyle, who will be speaking at the upcoming HIMSS19 conference in Orlando, to meet this demand, healthcare organisations need to work on strategies that will help “decongest the patient flow system”. 
“There is no one silver bullet that can solve all problems and implementing several capacity management strategies is a good point to start. Most hospitals in the public system have low numbers of beds per capita, so given that constraint, we need to figure out how to use data to improve flow efficiency and productivity.”
As capacity management is one of the more visible barometers of the performance of a health system, Boyle said “statistical rigour” needs to be put into analytics of patient flow, but it also needs the engagement of clinicians. 
“It needs to start with behaviour changes within an organisation. Clinical champions are needed to get staff behind an initiative as the early stages can be hard. It’s also important to have a solution that’s scalable and personalised to an organisation's needs,” he said. 
“There are three steps to this: determining the predictive accuracy of a particular tool, embedding it into workflows, and as a result, enabling healthcare professionals to make better decisions in support of clinical judgements.” 

Boyle provided a few examples where data analytics can come into play to improve clinical workflows: 
The first is to use historic clinical data to get efficiency improvements. He suggested healthcare organisations look at target occupancy rates for hospitals as it’s not a “one-size-fits-all” approach. 
“It’s important to have targets that are specific to hospitals that allow us to identify specific occupancy bottlenecks where flow performance declines. The whole goal is to avoid patients waiting and adverse outcomes that include mortality,” he mentioned. 
Another approach is by accelerating the time of patient discharge. Boyle said this is based on the idea that patient flow can be improved when patients are discharged an hour or two earlier, which is then quantified by the impact it has on the emergency department.  
“We can’t add more beds so we need to look into what else we can do, such as analyse situations around discharge timing or the configuration of beds.”
Predicting demand is the third approach. 
“What we do here is forecast patient arrivals and departures resulting in hospital preparedness. For example, we work with health departments on early warnings of outbreaks like influenza from a number of data sources. So in this case, being aware of the timing of the flu season and its magnitude from numerous data sources could provide early warning,” he said. 
Moving into the future 
Boyle said there’s potential for this technology within areas such as telehealth, mobile health and genomics as the intelligence derived from the data enables clinicians to prescribe healthcare that’s targeted to individual patients. 
“It’s great to get insight about what happened last week or is happening now, but the aim should be to use this data to predict what is going to happen in future,” he said.      
“Looking at things like vital signs data that are recorded routinely within the digital environment, for example, will provide lots of potential in future. Not feasible in the paper environment, this would provide a significant advantage with regards to early detection and risk stratification. 
“In the long-term, this may support the delivery of advancements like clinical genomics, where genome-based outcomes can be tailored to an individual. 
“But in the foreseeable five-year future, data analytics will revolve around delivering care for a specific patient – so using their health contact data, social media data and internet footprint as useful indicators will aid in personalising their healthcare experience.”
Boyle mentioned that the biggest challenge, looking into this future, is the change management that needs to happen to make it work – who needs to use that information, what time they need to use it, and workflow procedures that need to be created from it. 
“Sustainably embedding solutions into business-as-usual workflows is the greater challenge compared to the mathematics of developing and validating predictive models,” he added.
Boyle will be joined on stage by CSIRO Australian E-Health Research Centre Senior Research Scientist Sankalp Khanna and together, they will be presenting on the topic of Deriving Value from Patient Flow Analytics at the upcoming HIMSS19 conference in Orlando. Their session is on February 14, at 8.30-9.30am. 



White papers