In clinical contexts, “automation” is a buzzword. It’s meant to inspire images of doctors’ offices filled with computers as they calculate ways to improve patient outcomes at lower costs. This is where we’re headed, but we’re not quite there. Not yet, anyway.

In its truest form, automation involves taking relevant data, using those data to calculate some action, and implementing (or recommending) that action. Ideally, these steps can be repeated again and again as new information becomes available.

When we start thinking about this decision-making process in a care context, we can start to see how powerful clinical automation can truly be. Consider a relatively simple example of care automation: Rigotti et al. (2014) conceptualized an intervention for smokers who had recently been discharged from a hospital. Specifically, the patients received a series of follow-up phone calls with pre-recorded messages intended to support them while they tried to quit smoking. This meant that each day, a computer algorithm would calculate how long it had been since a patient was discharged. It would then determine whether the patient should receive a follow-up call that day. Finally, it would actually make the call and deliver the pre-recorded message. Another example of automation in a clinical setting is described in Ross et al. (2009). For this, an algorithm would take some measurements of a cardiac intensive care patient’s vitals every 30 seconds. The algorithm would then calculate whether the vitals were outside of acceptable ranges. If so, the algorithm would make the necessary adjustment to pumps that delivered anesthetic drugs to the patient.

Even with these two relatively simple examples, we can see that automation concepts can help us systematically meet a patient’s changing needs, improve outcomes, and limit our resource waste. Because of this huge potential for improving care efficiency, we should hold “automation” in higher regard than as just another buzzword.

But when we try to imagine clinical automation on a massive scale, we’re faced with an information problem: how can an algorithm effectively recommend care decisions if the relevant data isn’t easily accessible? As patient data is typically segmented between different care providers and is siloed in different health record systems, this data accessibility issue is essentially an industry standard. But RexDB® is ready and waiting to fix this.

Being a highly analyst-configurable database platform, RexDB offers a flexible tool to aggregate and curate raw clinical data. This means that RexDB can help put patient data into a form usable by effective automation algorithms.

RexDB’s ability to store and transform data has already proven useful in a clinical setting: Prometheus Research helped create a custom infrastructure that centralized and curated patient data independently collected at New York Presbyterian Hospital (NYPH) and Weill Cornell College of Medicine (WCMC) and stored in two separate medical record systems. The result was a data pipeline that gave a more comprehensive, coherent, and data-driven picture of patients’ needs and circumstances. It’s this sort of picture that’s necessary if clinical automation algorithms are to be most effective.

The collaboration between NYPH and WCMC is just the tip of the iceberg in terms of what RexDB can offer, and we at Prometheus Research are eager to help automated care reach its full potential!

To learn more about our work supporting  NYPH-WCMC, read our partner presentation found here.


N.A. Rigotti, S. Regan, D.E. Levy, S. Japuntich, Y. Chang, E.R. Park, J.C. Viana, J.H. Kelley, M. Reyen, & D.E. Singer (2014). Sustained care intervention and postdischarge smoking cessation among hospitalized adults: a randomized clinical trial. JAMA. 312(7): 719-728.

J.J. Ross, M.A. Denai, & M. Mahfouf (2009). A hybrid hierarchical decision support system for cardiac surgical intensive care patients. Part II. Clinical implementation and evaluation. Artif. Intell. Med. 45(1): 53-62.