Roles & Contributions 

Research/Analysis

Process: As a team, we discussed the various difficulties associated with being an ICU nurse. Through this discussion, we found that alarm fatigue was a common issue experienced broadly by nurses across the world. For example, a 2024 ICU-nurse study reported 62.6% of participating ICU nurses had average-to-high alarm fatigue, and alarm fatigue correlated with higher perceived stress. So we then set out to research and note the various sources of alarm fatigue and brainstorm ways we could resolve each of those issues one-by-one. We wrote those issues and our solutions down on a Google document. We then considered how we could fit those solutions together, discussing and evaluating our ideas. At the end of our discussion, we naturally settled on the concept of creating a wall display that visualized patients’ vitals and needs, while also improving the pagers that notify nurses about potential emergencies through visual and haptic feedback. To validate our direction, we interviewed our client to understand current industry pain points. This feedback informed our pivot toward a "peripheral" wall display rather than just another mobile app or software dashboard. Below are notes we obtained from our research. Sources for this data can be found here. 

• What is alarm fatigue?

  • Reduced attention and slower/less reliable responses to alarms caused by chronic, excessive exposure to frequent, mostly nonactionable alarms.

  • Key point: it’s not just “too many alarms.” It’s too many alarms that don’t require action, which trains clinicians to treat alarms as noise.

• The Identification Problem:

  • Client Insight: "The pager might be a phone type... these are less advantageous because you don’t have access to patient info immediately; the first couple sentences are usually confirming who the patient is and what room they have (and they sometimes get those wrong/mixed up)."

  • NTW Solution: The Tracking Wall provides immediate spatial context. Instead of a nurse spending precious seconds on a phone clarifying "Which room?", the wall's color-coded blocks and kinetic spinners provide an instant visual map of the ward's status, ensuring the nurse knows exactly where to run before they even check their pager.

• Why do alarms go off when not needed?

  • System is optimized for high sensitivity (don’t miss deterioration), which creates low specificity (lots of alerts that don’t require action)

  • Common causes:

    • Threshold violations that are transient or irrelevant (brief dips/spikes that self-resolve). 

    • Artifacts / bad signals (movement, poor contact, disconnected sensors). 

    • Algorithm limitations (e.g., arrhythmia detection generating many false positives).

• What information sets off an alarm?

  • A vital sign crosses a preset high/low limit (HR, RR, BP, SpO₂, etc.)

  • Arrhythmia alarms: algorithms detect rhythm changes (e.g., VT/VF, asystole)

  • Technical alarms: device/sensor problems (lead off, poor signal quality, low battery, disconnected probe)

• What information do nurses need when taking care of patients?

  • Patient-specific baseline and targets (COPD SpO₂ goal, permissive hypertension/hypotension targets, post-op parameters)

  • Trends and trajectory (is the patient improving or deteriorating over minutes/hours).

  • Immediate threats (airway/ventilation failure, severe hypotension, malignant arrhythmias, rapidly falling SpO₂).

  • What to do next (clear escalation/response expectation for high-priority alarms)

• Why do alarms exist?

  • To prevent harm by detecting physiological deterioration early and responding to them through timely intervention 

  • Continuous monitoring can detect deterioration faster than intermittent checks, but humans cannot stare at monitors continuously

• What is the current problem with alarms?

  • Noise and workload burden from false alarms

  • Desensitization and delayed response due to alarm fatigue

  • Risky workarounds (silencing and disabling alarms, overly wide limits)

• What are the current solutions?

  • Pagers

  • Signal-quality improvement (better skin prep, electrode/probe placement, replacing worn leads, addressing motion sources)

  • Smarter algorithms to detect patient emergencies

  • Patient specific alarms

  • Education and training (how to set limits appropriately; what alarms matter most)

• Why hasn’t this been solved yet?

  • Sensitivity vs specificity trade-off: making alarms “quieter” can increase missed detections; making them “safer” can increase nuisance alarms. 

  • Patient heterogeneity: ICU patients have wildly different baselines; one-size thresholds generate noise. 

  • Change management and liability concerns: hospitals fear reducing alarms and then missing rare catastrophic events.

Some feedback from the Client:

[7:04 PM, 1/26/2026]: As of right now the best method is using the patients chart to directly communicate between nurses and other providers. It keeps everyone on the same page and the patient chart is easily accessible

[7:05 PM, 1/26/2026]: Other hospitals use pagers that can either be a number they have to call or room they have to meet in or the pager might be a phone type these are less advantageous bc you don’t have access to patient info immediately the first couple sentences are usually confirming who the patient is and what room number they have (and they sometimes get those wrong/mixed up)

[7:06 PM, 1/26/2026]: Secondly the phone pagers lose service constantly unless you’re in an area that has service like hallway or room you’ll be fine but in staircases or elevators they often get the call cut and disrupts workflow

[7:07 PM, 1/26/2026]: Recently what’s become big is patients accessing their charts virtually and talking to drs directly

[7:07 PM, 1/26/2026]: We’re also supposed to reach out to our patients and tell them what their results are on this online thing whenever we get the results back

We repeatedly kept hearing things along the lines of: “The best thing that nurses can monitor is their patient chart,” and decided to make it easier for them as well using the Nurse Tracking Wall.

Design

Process: We used Canva to create the design. We explored different formats and tried using various shapes before deciding on this final combination. We carefully considered each aspect of the design to make a decision on how each feature should look and behave.

• Design features and rationale

  • We consulted a nurse throughout the design process who helped us understand the type of information that can/ should be shared through our interface. We also tried to view this from the users’ perspective to determine how effective the features would be.

  • Each unit displays basic but important information regarding room, assigned doctor and nurse, and patient stability. Screen colors range from gray (stable), green (slight fluctuations within expected range), yellow (early instability, check soon), and orange (unstable) as the main set, with blue and red showing the aftermath of codes (the tiles would keep flashing). The dot in the middle also serves as a stability indicator: jittery dot (severe instability), slow pulsing dot (small fluctuations), slow rotation (early instability) – more motion indicates higher instability.

  • For the pagers, the information received gets a bit more specific in terms of what the issue. The color, as well as the duration and intensity of the vibrations and volume will vary to help distinguish the severity of the issue.

• Design artifacts

  • We made a few example screens to represent how the interface can look at different times: a unit view of the interface wherein there are different points of patient stability, important code that takes up the whole screen for a few seconds, and pagers behaving differently depending on the issue.

Evaluation 

Process: Due to a very tight turnaround time, we conducted a rapid, informal review by sending the design to relatives who currently work within the healthcare system. We asked them to evaluate the tool based on their day-to-day professional experience to see if the core concept addressed the actual needs they see on the job. 

While this provided a quick gut-check on the design’s relevance, it was primarily focused on high-level feedback rather than deep, technical usability testing. In all honesty, their positive remarks and our ideas we currently have are not sufficient and we need to conduct many interviews to gain more context on the entire area.

• Findings

  • The feedback was interestingly cautious since the people we talked to mentioned that the prototype is similar to the types of solutions they frequently discuss in their own workplaces as “nice-to-haves”, which suggests we are on the right track. However, they were clear that they couldn't give a final verdict on its effectiveness until the tool is fully implemented and operational.

  • A core takeaway was that the hospital environment is incredibly complex and requires every tool to be perfectly coordinated with existing systems. Because of this, they expressed doubt about whether such a design is actually possible to deploy, as the high level of required integration might make it difficult to fit into their current tightly managed workflows.