menu
As people flow from their 'normal' stat to less typical states. Have drug doses and vital trackers adjust to temporary irregularities as needed. For example, being in surgery or pregnant is a temporary state that has drug ranges that you would not normally give someone.
What is an appropriate scope for standardization?
The Seat that fit no one, in the 1950's the US Air force measured thousands of men to find the average dimensions to use for their cockpit design. The problem was that this seat fit no one, because every one of their solders had some divergence from their 'perfect average'.
In a perfect situation 1 design would work for everyone and you would only have to implement it once. But no one is going to be perfectly average.
So how do we find the balance of reasonable for our team to run, and customizable enough to actually fit the needs of the users?
Originally this project was going to focus on UI in 'ICU Drug Rights'. Addressing risk for wrong drug dispensation and human failure.
But the nursing staff that I was working with highlighted challenges unique in obstetrics and Picosin. So this design focus shifted to their current needs.
My approach is inspired by two biological models. First, epigenetics: unlike DNA, which is fixed, epigenetic markers are adaptive layers that regulate gene expression in response to environmental changes without altering the core genetic code. This is my design metaphor. I am not trying to rewrite the hospital's entrenched 'DNA'—its legacy equipment, protocols, and physical infrastructure. That is impossible. Instead, I am designing an epigenetic layer: a responsive UI that can adapt to new safety problems and information stimuli, activating or deactivating protocols to maintain safety without requiring a total system overhaul.
The mechanism for this layer is inspired by the body's dual-processing model, used for homeostasis. The body doesn't have one setting; it constantly makes micro-adjustments (e.g., raising or lowering heart rate) to achieve a balanced state. Similarly, my UI proposes two primary modes or 'environmental tops' that nurses can switch between—perhaps a 'standard' mode and a 'high-alert' mode—each optimizing information flow and checks for a different context. This shifts the burden of adaptation from the nurse's cognitive load onto the system itself, allowing them to focus on their patient rather than compensating for the hospital's design flaws, thereby directly reducing alarm fatigue.
Inspiration: epigenetics framework
Biochemistry: Dual Processing Model
Users: Obstetrics Nursing Team
A lot of heath standards and ranges were determined with a very small section of the population. For example, the BMI calculations where based off of measurements of Scotsmen and doesn't work for a lot of people.
A really important shift that must be established for personalized treatments, is having homeostasis measured to the individual from the start, finding 'normal' to the individual over years of observation (like annual check-ups). Rather than comparing everyone to an arbitrary measurement of 'normal'.
This shift from a population-based standard to a personalized baseline is fundamental. Rather than using a single, rigid line of 'normal'—a model that inevitably fails anyone who doesn't fit the original narrow sample—we must establish a unique homeostatic set point for each individual.
This is akin to the planetary orbits in the illustration; health is not a static destination but a dynamic state of equilibrium maintained through constant, tiny adjustments. This personalized 'orbit' is established over time through longitudinal tracking at annual check-ups, defining what 'normal' means for that person's body.
The goal of treatment then changes dramatically. Instead of forcing a patient's metrics toward an external, arbitrary average, the focus becomes supporting their body's innate ability to return to its own unique baseline after a disruption, creating a truly responsive and effective framework for lifelong health."
This is a complicated environment and the most important functionality is that information is successfully communicated quickly.The role of this design is to facilitate that by reserving drug alarms for catching fatal errors.
The goal of this design is to make technology a passive, supportive partner in healthcare, moving beyond the anxiety of human-versus-machine competition. Technology is changing society at an unprecedented rate, but in hospitals, the principle must be clarity:
tech should do what it does best—accurately store and process data
humans should do what they do best—adapt and react to the novel needs of the environment
By reducing alarm fatigue through intelligent, prioritized alerts, the role of tech becomes ambient, working seamlessly in the background. Clinicians should only actively notice it when they deliberately interact with it. The ideal system is unnoticed until needed, a symbiotic partnership where technology's analytical power amplifies human expertise, freeing medical professionals to focus their irreplaceable skills on the patient. The current model of having tech make decisions and humans checking spreadsheets is a ridiculous reversal of strengths; our goal is to reverse that reversal."
As alarm fatigue is more relived overtime with an accumulation of designs like this one. I see the role of tech support in healthcare becoming more passive in the process.
Eventually clinicians only noticing all the background support when they are deliberately interacting with tech.
Unnoticed until Needed
Understanding Needs: I started off by meeting with different specialties in healthcare to inquirer current problems in the industry. I met with, compliance lawyers, patient safety specialist and nurses active in the field.
Interviews: I used discovery interviews. When I focused in on a topic I switched to a Cognitive Task Analysis interview to breakdown a specific process.
Analysis: I combined my notes and coded common patterns and themes in recorded interviews
This model directly translates the psychological theory into a clinical tool. The patient's default state ('System 1') is their established, personal homeostasis, and the system operates here with minimal alerts, maintaining a steady course. However, during anticipated physiological irregularities like surgery or childbirth—the 'System 2' events—we know the patient will need different parameters.
The system is designed to proactively shift into this high-support mode, expanding the acceptable drug dose ranges to accommodate the patient's temporary but intense needs. Crucially, as the illustration shows, this isn't an error or an alarm; it's an intentional, guided variation. The goal is to actively support the patient through the event and then systematically guide them back to their personal baseline, making the return to 'System 1' the final, intended outcome of the 'System 2' intervention.
Stakeholders: Obstetrics Team
Codes: alerts, rules, safety, practice, time pressure
overriding redundancy
The obvious priority in healthcare.
specialized uses
Focus on Pitocin
the transition from pre- to post-partum
Challenging to find the sweet spot of regulations to support and not hinder care
obstetrics does not have gentle transitions for nurses to rewrite all the new doses for each stage
This structure was inspired by the dual processing Model in cognitive psychology. That there is a 'system 1' for fast judgment and intuition and a 'system 2' for slow cognitively tasking thought and detail.
When the patient is in homeostasis have the 'system 1' equivalent default.
When a patient is in a special event like surgery or giving birth switch to 'system 2' equivalent. With drug dose ranges specialized to support a patient's return to their homeostasis.
To aid in filling up 'need gaps' of historically neglected population groups. These epi-conditional rules will work as the system adapting to the environment.
In this example, a patient moving from pregnant, to delivering, to just had a baby. The need of the drug Pitocin moves from none, to just a little, to a lot, then back to none.
The epi-conditional rule will adjust the accepted dose range through each phase so nurses don't have to fight the program or override anything. The program just helps in the background and pops up an alarm if something actually alarming is occurring.
To aid in filling up 'need gaps' of historically neglected population groups. These epi-conditional rules will work as the system adapting to the environment.
In this example, a patient moving from pregnant, to delivering, to just had a baby. The need of the drug Pitocin moves from none, to just a little, to a lot, then back to none.
The epi-conditional rule will adjust the accepted dose range through each phase so nurses don't have to fight the program or override anything. The program just helps in the background and pops up an alarm if something actually alarming is occurring.
A sample of an infusion screen for an in-patient with a special variation tag
High Value Contrast (left): when setting up the infusion, a nurse could tag the patient with special tags. For temporary events, like delivering a baby, or long-term tags like diabetes or cystic fibrosis.
Fast Visual Read (right): the specialty tags would have an inverted value contrast, allowing nurses to quickly scan patient status, from multiple distances
This interface breakdown demonstrates the critical principle of redundant coding. Instead of relying on a single visual cue, critical information is communicated through multiple simultaneous channels: a unique symbol (the baby icon), a value-flipped color scheme (inverted contrast), a clear text label ('pre-partum'), and a numerical dose range.
This multi-layered approach ensures the message is received even if one channel fails or is overlooked due to stress or environmental conditions. A nurse across the room might see the inverted contrast. A nurse closer might read the tag. A nurse programming the pump will see the specific dose. This creates a robust safety net, ensuring that the patient's special status is understood quickly and unambiguously from any context, which is paramount for preventing medical errors.
A sample breakdown of the infusion screen
Callouts: samples of tags: cystic fibrosis (65 roses), increase/decrease to homeostasis, pregnancy, heart conditions etc.
Interface Sample: showing how a patient in a pre-partum stage, could have special range for drugs related to delivery at that stage and a normal dose of fluids