From Big Data to Quick Decisions

Abstract

The department of Critical Care Medicine at SickKids Children’s Hospital is an ever changing environment. In the department’s units one finds the most vulnerable patients, surrounded with many monitoring devices, intravenous pumps and life support devices. Sudden changes in their physiological states require clinicians to make quick decisions concerning treatment of the patients. However, the large variety of monitors makes it more complex to see, analyse and interpret the right data. Clinicians are often required to make calculations in their head and make quick and efficient decisions, mainly based on expertise. To facilitate clinicians in decision making, the hospital adopted a system called T3. This so called clinical decision support system integrates the data streams from the different monitoring devices into one user interface. In addition, it contains algorithms that can predict events. This should shift decision making from prescriptive to predictive, which implies that clinicians can now prevent events from happening when the right treatment is given. T3 was implemented to support clinicians. However, in practise, the majority of clinicians do not use the system. One of the assumptions regarding why this is occurring is that the system is not user friendly. This produced the initial assignment of this thesis: to redesign T3 in such a way that it is easy to use. Ultimately the assignment was supplemented with the goals of adding more functionalities to the current system and eventually exploring conceptual scenarios for its future use. A total of five iterations, based on a tablet-first principle, led to a new design, built around three main use scenarios. Extensive decomposition and analysis of the current T3 system, together with the application of a heuristic evaluation formed the basis of all iterations. To reinforce the system’s usability, new functionalities were extracted from trend analysis literature. The last two iterations, the concept and final design, elaborate on the first three iterations and add new perspective on T3’s potential use in the future. Ultimately, the final designs pursue a future where the critical care field makes a big shift from experience based to evidence based decision making. The vision describes a T3 that does not only predict events, but it also recommends evidence based suggestions for treatment. Ideally, T3’s standards will eventually be adopted by other health technology companies. This would facilitate and improve the development of evidence based healthcare. Moreover, this integration would create more consistency among various medical devices. This would improve the general usability of medical equipment. The future vision creates an ideal scenario for the future, but as the word implies this is only a vision for now. Before this vision might become reality, multiple iterations of thorough research should be conducted. Only if the research results are watertight, could the creators of T3 start implementing algorithms for treatment advice. Two other present- oriented scenarios distinguish the use of T3 at the bedside and away from the bedside. The corresponding designs are much more likely to be feasible for implementation on the short term.