Multimodal Alarm Distribution System for Intensive Care Units

With the increasing development of biosensor technology and the associated possibilities of patient monitoring, the number of acoustic and visual alarms in intensive care units (ICU) is increasing as well. These alarms (about 250 per patient every day) are audible for every person in the intensive care unit and thus disturb both patients and staff. Each alarm must be analyzed and evaluated by the caregiver in charge of the cause and urgency of the alarm, which is 1) a high cognitive load and 2) the potential risk of alarm fatigue, desensitization and slower response time to alarms. The majority of scientific approaches aimed at reducing alarm fatigue focus on reducing the number of alarms. E.g., such as improved algorithms, intelligent alarm delays, and changes in overall workflow and alarm policy. Significant improvements have already been achieved, but the remaining alarms are still audible everywhere.
This dissertation examines how patient alarms should be designed to alert nurses personally and unobtrusively, thus reducing noise exposure in intensive care units. In participative design studies, alarm patterns for a head-mounted display were designed and appropriate modalities within laboratory studies with ICUs were evaluated to derive a multimodal alarm design for bodyworn patient monitor systems. The results showed that for short-term use, personal alarms were often preferred by nurses compared to state alarms. Based on the results, a multimodal alarm pattern could be derived, which was integrated into a Google Glass.
To confirm alarms, first touchless interaction designs were implemented, which were evaluated in a pilot study. For subsequent research, developers, and engineers, this thesis provides design guidelines for body-worn alarm systems.