No One Left Behind: Adaptive Tablet Modalities for Digitally Excluded Emergency Department Patients Design, Implementation, and Social Evidence for an Impairment-First Interface
1MSc Engineering — Innovative Communication Technologies and Entrepreneurship Aalborg University, Copenhagen, Denmark
2Department of Ophthalmology, Bangladesh Medical University, Dhaka, Bangladesh
*Corresponding author. Email: achowd24@student.aau.dkAbstract
Background
The urgent care departments in Europe face a structural paradox: accelerating digitalisation is accompanied by a patient population that is disproportionately unable to engage with standard digital tools. An internal analysis at the Emergency Department (Akutafdelingen) of Nordsjællands Hospital in Hillerød, Denmark found that 43% of emergency patients struggle with digital solutions — a figure that reflects the predictable composition of acute care populations rather than any individual failing.
Objective
This paper presents the design, iterative development, and secondary validation of the ED Adaptive Interface (v5): a prototype adaptive patient terminal developed in response to this challenge. The system operationalises what the author terms impairment-first design — a methodology that treats the most constrained patient experience as the primary design problem and derives the standard experience as a subset. The interface configures itself in under ten seconds via nurse-led setup, adapting across four axes of impairment: visual, motor, speech, and cognitive.
System
Version 4 supports five accessibility modes, a heatmap pain assessment grid, a Privacy and Dignity panel, a live workflow tracker with care notifications, structured dual-category help requests, and plain-language medical term definitions across four languages. Version 5, reported here for the first time, introduces a Condition Worsening Escalation button, a Referral Pathway Display, a "Why Am I Waiting?" triage explainer, a Symptom Progression Log, MinSP/Yellow Card Scan simulation, expanded language support (seven languages: English, Danish, Arabic with full RTL layout, Turkish, Romanian, Polish, and Somali), and an expanded ten-item Communication Board. The entire system runs as a single 79-kilobyte HTML file with zero infrastructure requirements.
Methods
To base the design on patient-generated evidence, two independent social media threads were subjected to an inductive thematic analysis (Braun and Clarke, 2006) a primary corpus of 83 entries in the Facebook group: Foreigners in Denmark (collected March 2026) and a corroborating corpus in an international community group in the Aarhus region (collected April 2026). All identifiers in both datasets were fully anonymised under GDPR Article 89 research provisions prior to analysis. No participants were contacted. Generative AI tools were used to assist with drafting, writing, and prototype code development in the preparation of this manuscript; all scientific content, data collection, analysis, and conclusions are the sole responsibility of the authors.
Results
The first discourse corpus produced five major themes in relation to the five general problem areas that the prototype was intended to cover: system navigation and triage literacy gaps (31 entries); language and cultural barriers (6 entries); communication failures during care (5 entries); staff overload and capacity constraints (8 entries); and pain and severity assessment failures (14 entries). The supportive dataset supported all five themes on its own and presented two new themes: the different treatment of international patients and medical gaslighting as a long-term trend of patient advocacy failure. One of the major structural discoveries the five most-liked comments were critical of the original poster being self-referring to the ED when she had in fact been explicitly triaged to receive 1813 telephone referral to the ED directly inspired the Referral Pathway Display and Why Am I Waiting? features in v5.
Conclusions
The convergence of design rationale and independent social evidence across all five problem categories suggests that impairment-first design is not a niche accessibility concern but a structural approach to healthcare interface quality. The prototype is ready for a structured clinical pilot using the System Usability Scale (SUS) and semi-structured staff interviews. The long-term roadmap includes full MinSP integration, hospital PMS connectivity, and clinical validation.
Article notes
Competing Interest Statement
The authors have declared no competing interest.
Funding Statement
No external funding was received for this research. The work was conducted independently by the authors without institutional financial support.
1.Introduction
Healthcare systems across Europe face a structural paradox. The same demographic groups most likely to require emergency care — older adults, patients with chronic conditions, recent migrants, and individuals with cognitive or physical impairments — are also those least equipped to navigate the digital interfaces through which care is increasingly delivered. As healthcare digitalisation accelerates, this asymmetry does not resolve itself; it compounds.
At Nordsjællands Hospital’s Emergency Department (Akutafdelingen) in Hillerød, Denmark, the tension between digitalisation and inclusion arrived as a concrete institutional problem. A hospital-led internal analysis determined that 43% of emergency patients struggle with digital solutions. The Emergency Department sees an unpredictable cross-section of the population: an 85-year-old arriving with acute confusion sits alongside a 30-year-old with a sports injury; a recent refugee with no Danish shares a waiting area with a long-term resident who has no smartphone and no prior encounter with digital healthcare tools. Generic tablet interfaces, calibrated for a digitally fluent adult in a state of baseline calm, exclude nearly half of this population before a single clinician has entered the room.
The solution concept described in this paper originated during the OpenInnovation 2026 Health challenge — a cross-university student innovation sprint co-organised by DTU Skylab, Copenhagen School of Entrepreneurship (CBS), and KU Actory, with Nordsjællands Hospital as the case partner — but the prototype, methodology, and all work reported here were developed independently after that event. The ED Adaptive Interface is a working prototype that configures itself in under ten seconds to serve patients across the full spectrum of visual, motor, speech, and cognitive impairment, in their own language, with zero infrastructure requirements.
The design philosophy at the core of the prototype — impairment-first design — is the paper’s central theoretical contribution. Rather than building an interface for a typical user and retrofitting accessibility as a compliance layer, the design began with the most constrained patient imaginable: one who cannot see, cannot touch the screen, cannot speak, and cannot read Danish. Every design decision was tested against this patient’s needs before any other consideration. The result is an interface that, by solving the hardest problem first, also serves every other patient better.
This paper makes four contributions. First, it documents the architecture and full feature set of v4 of the ED Adaptive Interface. Second, it reports v5 — a set of evidence-driven upgrades developed following secondary validation through social media discourse analysis — for the first time. Third, it formalises the impairment-first methodology as a transferable framework for healthcare interface design. Fourth, it presents an inductive thematic analysis of 83 entries from a public social media forum as a post-hoc validation source — an approach that, to the author’s knowledge, has not previously been applied in this form in health informatics research.
The prototype is technically unremarkable: a single HTML file, 79 kilobytes, running on any modern browser without installation or network connection. The contribution is not the technology. It is the thinking that produced it, and the evidence that the problems it addresses are real, widespread, and structurally predictable.
2.Background and Related Work
2.1Digital Exclusion in Healthcare Settings
The relationship between digitalisation and equity in healthcare is well-established in the literature. Digital health interventions consistently assume levels of literacy, language proficiency, motor dexterity, and cognitive capacity that many patients — particularly those most likely to require emergency care — cannot reliably provide [1, 2]. This is not a minor oversight. Older adults, migrants, patients with physical disabilities, and those experiencing acute psychological distress face compounding barriers that standard interfaces are simply not built to accommodate [3].
In the Danish context, two pressures converge. The country’s commitment to a digital-first public sector has been ambitious and rapid, with electronic patient records, online appointment systems, and digital prescription management all migrating to digital-only delivery within a comparatively short timeframe. Meanwhile, the Emergency Department population is structurally heterogeneous. Eurostat data indicate that between 35 and 50 per cent of adults aged 65 and over in EU member states report significant difficulty with digital devices [4], a proportion that rises sharply in acute care settings where IV lines restrict hand movement, sedation reduces response precision, pain degrades concentration, and anxiety compounds every other barrier.
The 43% figure cited by Nordsjællands Hospital is the predictable outcome of deploying technology designed for baseline conditions in a setting where baseline conditions are the exception. Designing around it is a clinical necessity, not an optional accessibility accommodation.
2.2Accessibility Frameworks and Adaptive Interface Design
The human-computer interaction literature offers well-developed frameworks for accessible design, most prominently the Web Content Accessibility Guidelines (WCAG 2.1) [5] and the WHO’s global assistive technology standards [6]. However, most digital health implementations treat these frameworks as compliance checklists applied after the primary interface is built — a set of retrofitted accommodations for an assumed minority, rather than foundational design constraints.
Universally designed interfaces — those built from the outset for the widest possible range of users — consistently outperform retrofitted accessible variants across usability and satisfaction metrics [7]. The curb-cut effect, originally observed in built environments, is well-documented in digital contexts: features introduced for motor-impaired users (keyboard navigation, large tap targets, high contrast modes) reliably improve the experience for all users [8]. Switch-access devices, sip-and-puff technology, and augmentative and alternative communication (AAC) boards are established assistive tools with strong clinical evidence bases [9], yet they remain largely absent from hospital-issued patient-facing interfaces. The ED Adaptive Interface integrates compatibility with these devices natively through its Auto-Scanning Focus mode, at no additional hardware cost.
2.3Alarm Fatigue and Structured Clinical Communication
Emergency departments present a distinctive communication environment. Patients are frightened, frequently in pain, and often unable to articulate their needs clearly. Nursing staff operate under persistent time pressure and high cognitive load. The combination creates conditions for what the clinical literature terms alarm fatigue: the desensitisation to alert signals that develops when the same notification mechanism is used regardless of clinical urgency [10, 11]. When a request for a blanket generates the same alert as a report of chest pain, response thresholds rise over time and genuinely urgent signals risk being missed. Structured communication approaches separating non-urgent comfort needs from clinical urgency have been proposed in nursing informatics [11], but their implementation in patient-facing interfaces has been limited. This gap directly informed the division of all patient help requests into two structurally distinct categories — Comfort and Medical — each generating a different signal type to the care team.
2.5Pain Assessment in Emergency Populations
Standard pain assessment tools — the Visual Analogue Scale (VAS) and Numeric Rating Scale (NRS) — were validated primarily in research populations. Their performance degrades in patients with limited literacy, language barriers, motor impairment, or cognitive distress, precisely the populations most common in emergency settings [13]. Multi-modal pain assessment tools presenting simultaneous numerical, colour, and facial expression cues have demonstrated superior reliability in these populations. The heatmap pain grid in the ED Adaptive Interface combines these three cue types with large tap targets compatible with tremor, IV lines, and agitation.
2.6Health Literacy and Patient Engagement
Health literacy is a well-established predictor of healthcare outcomes. Paasche-Orlow and Wolf [14] identified low health literacy as a causal pathway to worse clinical outcomes, lower medication adherence, and higher emergency readmission rates. In an emergency department context, patients frequently encounter medical terminology without explanation, compounding anxiety and reducing cooperation. Plain-language definitions of every medical term appearing in the ED Adaptive Interface, available in all supported languages, represent a low-cost, high-impact response to this documented problem.
3.Design Methodology: Impairment-First Design
3.1The Core Principle
The central design principle of the ED Adaptive Interface is impairment-first design. The principle is deceptively simple: identify the most constrained user scenario and solve that problem first. Every subsequent design decision is evaluated against this primary constraint before any other consideration. Typically-abled, digitally literate, linguistically fluent patients are treated not as the default user but as a beneficiary subset — they receive the same interface, which turns out to be better for them too.
This represents a deliberate inversion of the conventional product design hierarchy. In most digital product development, accessibility features are added after the core product is built for a typical user. Impairment-first design inverts this entirely: every feature is built from the ground up to serve the most constrained user, and the standard experience is derived from that. In the words of the design philosophy that guided every decision:
"We don’t ask the patient to learn our technology; our technology learns the patient."
This is not merely a tagline. It describes a specific design protocol. When any feature was under consideration, the first question asked was: can a patient who is blind, paralysed, mute, confused, and reads no Danish use this? If not, the design was rejected or modified until that patient could engage with it. Only then was the feature evaluated against any other criterion.
3.2Defining the Four Impairment Axes
Four dimensions were identified along which patients may experience reduced digital interaction capacity in an emergency context. These are not clinical diagnoses but functional states — the actual capabilities that determine whether a patient can engage with a tablet during an ED stay:
Visual
Reduced ability to perceive screen content, due to low vision, blindness, sedation, poor lighting, anxiety, or the absence of reading glasses.
Motor
Reduced ability to touch or manipulate a touchscreen, due to tremors, IV lines, post-stroke weakness, paralysis, cerebral palsy, sip-and-puff dependency, or post-operative restriction.
Speech
Reduced or absent spoken communication capacity, due to tracheotomy, stroke, severe pain, acute respiratory distress, laryngectomy, or psychological trauma.
Cognitive
Reduced ability to navigate multi-step digital interfaces, due to dementia, acute confusion, sedation, high anxiety, acquired brain injury, or simple unfamiliarity with digital tools.
These axes are not mutually exclusive and must be treated as independently composable. The nurse setup interface allows any combination of accessibility flags to be set simultaneously.
3.3The Nurse-Led Configuration Approach
A critical design constraint was time. An admitting nurse typically has seconds, not minutes, to configure any patient-facing tool. Self-configuration approaches are inappropriate for populations that are, by definition, struggling with the technology. The solution was a dual-tab nurse setup interface. The first tab asks what the patient can do (abilities); the second asks what barriers they face (disabilities). Leading with capability rather than limitation produces a more complete and more dignified patient profile. The complete configuration requires fewer than ten seconds, after which the system transforms automatically to the configured mode.
3.4Iterative Development Process
The prototype was developed iteratively, progressing from v1 through v4 over the course of the OpenInnovation 2026 challenge, with each version documented in external-facing specifications — the Feature Guide and Jury Q&A documents — that maintained design coherence. Version 5, reported in this paper, was developed independently after the challenge period, driven directly by findings from the social media discourse analysis described in Section 5. The v5 features are not additions motivated by developer preference but upgrades traceable to specific themes in patient-generated discourse data.
4.System Architecture and Feature Implementation
4.1Technical Architecture
The ED Adaptive Interface is implemented as a single HTML file using React 18.2 (loaded via CDN) and Tailwind CSS for styling. The compiled application is approximately 79 kilobytes — smaller than the average web page image. It runs fully offline in any modern browser without installation, server infrastructure, or network connectivity. There is no database, no user account system, no app store, and no IT department involvement required. The file can be transferred to a hospital tablet via USB drive and opened directly.
This zero-infrastructure architecture was a deliberate design constraint, not a technical limitation. Hospital IT environments are typically slow to onboard new software, requiring security audits and presenting significant barriers to rapid deployment. A self-contained HTML file bypasses all of these barriers: any ward with a browser can run the prototype immediately. The absence of server-side data persistence is acceptable at prototype stage and is addressable through future integration with the hospital’s patient management system (PMS).
4.2Nurse Setup Interface
The nurse setup screen is divided into two tabs: Abilities (what the patient can do) and Disabilities (barriers they face). Language selection is a global toggle affecting every string in the entire interface. Accessibility mode selection is checkable in any combination. From v5, the nurse also selects the patient’s referral source — 1813 telephone triage, self-referral, ambulance, or GP — which then appears as a persistent badge on every patient-facing screen for the duration of the visit.
A MinSP / Yellow Card Scan button in v5 simulates the integration concept for the long-term roadmap. The nurse taps "Scan Now," an animated spinner plays, and the patient profile is pre-populated with placeholder data. This is a proof-of-concept demonstration, not a live integration; its purpose is to show stakeholders what the completed integration pathway would look like in practice.
4.3Accessibility Modes
Five accessibility modes transform the patient dashboard automatically based on the nurse’s configuration. Each mode is independently activatable and composable with others. Table 1 summarises the five modes, their target impairment axes, patient experience, and technical implementation.
The Blue/Orange high-contrast palette warrants specific explanation. Most high-contrast implementations use black-and-white, which fails for the most common forms of colour vision deficiency (red-green deficiency, affecting approximately 8% of males of Northern European descent). Blue-orange combinations are discriminable by approximately 99% of the population, including those with deuteranopia and protanopia, while still achieving the luminance contrast ratios required by WCAG 2.1 Level AA [5].
The Auto-Scanning Focus mode is fully compatible with external switch devices, including sip-and-puff systems used by patients with severe motor impairment (ALS, high spinal injury, locked-in syndrome). A glowing border cycles through interactive elements on a three-second interval; any external keypress event — from a keyboard, a Bluetooth switch, or a sip-and-puff device — triggers selection of the currently highlighted element. The interface is fully operable by patients who have zero direct contact with the touchscreen, at no additional hardware cost.
4.4Clinical Communication Features
4.4.1The Divided Help Request System
All patient help requests are divided into two structurally distinct categories. Comfort requests water, blanket, temperature adjustment, privacy, dimmed lights — are non-urgent and generate a low-priority signal. Medical requests — pain, difficulty breathing, dizziness, chest pain — are urgent and generate an escalated signal requiring prompt attention. The clinical rationale is alarm fatigue prevention [10]: when every request generates the same alert regardless of urgency, response thresholds rise over time and critical signals risk being missed.
4.4.2Heatmap Pain Assessment Grid
The standard Visual Analogue Scale requires fine motor control and presents a single positional cue that may not carry consistent meaning across language and cultural contexts. The ED Adaptive Interface replaces this with a grid of eleven large tap targets (0–10) providing three simultaneous cues per rating: a colour (green to red gradient), a number, and an emoji icon. The targets are sized to be reliably activatable by patients with IV lines, tremors, or motor agitation. This approach is consistent with recommendations in the pain assessment literature for multi-modal representations in linguistically and cognitively diverse populations [13].
4.4.3Privacy and Dignity Panel
The Privacy and Dignity Panel provides one-tap access to six requests without requiring speech, explanation, or waiting for a nurse to notice: Close Curtain, Cover Me, Lower Lights, Need Privacy, Call Family, and Feel Unsafe. The last of these is a discreet escalation mechanism — a silent alert to staff that the patient feels unsafe — without requiring the patient to speak, raise their hand, or attract attention within the department. To the author’s knowledge, this feature is absent from all existing commercial ED patient terminals.
4.4.4Live Workflow Tracker and Care Notifications
A five-stage workflow tracker (Triage → Tests → Doctor → Treatment → Discharge) provides patients with a persistent visual representation of where they are in the care process. At each stage transition, a banner notification appears in the patient’s own language explaining what just happened and what comes next, written in plain language. Proactively answering "What is going on?" before the patient has to ask eliminates a category of nursing interruption that consumes significant time in high-volume ED settings.
4.4.5Communication Board
The Communication Board provides tap-to-talk access to frequently needed phrases for non-verbal or speech-impaired patients. Version 4 included six items: Thirsty, Bathroom, Cold, Yes, No, and More Pain. Version 5 expanded this to ten items, adding Scared, Too Loud, Allergies!, and I Need Help — additions directly motivated by entries in the social media discourse corpus describing patients who had no means to communicate emotional distress or urgency escalation. When the speech-impaired mode is active, the Communication Board is promoted to the home screen.
4.4.6Plain-Language Medical Term Definitions
Every medical term appearing in the interface — Triage, IV Line, Vital Signs, ECG, and others has an expandable plain-language definition available in all supported languages. Health literacy research consistently shows that patients who understand their clinical situation are more cooperative care partners and report better satisfaction outcomes [14]. Providing this information passively at the point of encounter reduces both cognitive burden on the patient and the frequency of explanatory interruptions to nursing staff.
4.5Version 5 New Features
Version 5 introduced seven substantive upgrades, all directly traceable to themes identified in the social media discourse analysis reported in Section 5. Table 2 summarises these additions.
Two v5 features deserve particular elaboration. The Condition Worsening Escalation button addresses the most serious failure mode in the discourse corpus: patients whose clinical deterioration was not communicated to staff, either because they could not attract attention or because their self-reports of worsening were not taken seriously. The button is prominently placed on the home screen, coloured red, and opens a structured five-item checklist before generating a timestamped alert. The structured format reduces false activations and provides staff with specific clinical information before they enter the room.
The "Why Am I Waiting?" module addresses Theme T1 directly. It presents a collapsible explainer showing the triage priority system (RED/ORANGE/GREEN levels) with the patient’s own assigned level highlighted, alongside the referral source badge set by the nurse at admission. A patient who arrived with 1813 approval can see, on their own screen, that they arrived via the telephone triage service and that their priority level is ORANGE. This does not change the wait time, but it fundamentally changes the information environment in which the patient waits — and, as the discourse corpus demonstrates, the absence of this information generates significant distress, confusion, and unjustified social sanction from other patients who assume self-referral.
6.Discussion
6.1Impairment-First Design as a General Framework
The impairment-first principle is not specific to emergency department tablets or to the Danish healthcare context. The core claim — that designing for maximal constraint produces better outcomes for all users — is consistent with a substantial body of evidence in universal design research [7, 8]. Wheelchair ramps benefit parents with pushchairs, delivery workers, and cyclists. Closed captions benefit viewers in noisy environments. Automatic door openers benefit patients with trolleys and full hands. The curb-cut effect is a structural consequence of solving hard problems thoroughly.
In digital health, investment in accessibility for the most constrained users is not a cost imposed by a niche use case but a structural quality investment that improves the product for every user. Large tap targets make interfaces easier to use for anyone in a stressful, time-pressured environment. Plain-language definitions reduce anxiety for all patients, not only those with limited health literacy. Healthcare institutions that ask "which interface works for the most patients?" systematically underinvest relative to those that ask "which interface works for the patient who arrived alone, in pain, unable to speak, unable to read Danish, and currently at the limits of their cognitive capacity?" The latter question produces better software.
6.2The Social Media Discourse Method: Advantages and Limitations
The use of public social media discourse as a post-hoc design validation source is unusual in health informatics. Its genuine methodological advantages — unsolicited content, unfiltered emotional register, multi-perspective structure — cannot be replicated in formal patient satisfaction surveys. However, several limitations must be acknowledged. The "Foreigners in Denmark" group is not representative of the full ED patient population: its members are younger, more digitally active, and more likely to experience language barriers than the average Danish emergency patient. The dataset amplifies certain themes and almost certainly underrepresents others, particularly the experience of elderly monolingual Danish patients with cognitive impairment. The thematic analysis should be understood as corroborating rather than establishing the significance of these themes.
A further limitation is source diversity. Although two independent threads were analysed, both were drawn from the same platform (Facebook), and each was captured at a single point in time in a group with evident moderation tendencies. The corroborating Aarhus thread partially addresses the single-group limitation, but both groups skew toward international residents who are digitally active. Future research using this methodological approach should seek multi-platform corpora, ideally including Danish-language sources, alongside explicit archival strategies that account for moderation effects.
6.3The Prototype in the Broader Digital Health Landscape
The ED Adaptive Interface sits within a growing landscape of patient-facing digital tools deployed in clinical settings. Commercial proprietary ED tablet solutions have been implemented in larger European hospital groups, typically with single-language interfaces, standard VAS pain scales, and call-nurse buttons without categorical differentiation. The absence of the features documented in this paper from commercial implementations suggests either that the problem has not been identified or that design priority has not been placed on the most constrained users. The prototype demonstrates that building this kind of system is neither technically complex nor expensive: it required no specialist equipment, no proprietary software, and no institutional budget. The primary investment was in design thinking.
6.4On Convergent Validity
The convergence between the design rationale developed independently from the literature and the themes emerging independently from the social media corpus is meaningful evidence. The Workflow Tracker was not designed because of the "Foreigners in Denmark" thread — it was designed because alarm fatigue research indicated it would reduce nursing interruptions. The thread then independently produced 31 entries describing exactly the kind of navigation and information confusion the Workflow Tracker is designed to address. This convergence does not prove efficacy, but it does suggest that the problems being addressed are real, widely experienced, and not artefacts of the specific design context.
The most striking convergent finding is Theme 1’s structural pattern: the most highly-engaged comments in the thread were factually incorrect about whether the original poster had appropriate grounds for her attendance. This is a finding about the information environment in which patients experience and discuss emergency care. The Referral Pathway Display and Why Am I Waiting? module change the information available to the patient — a modest intervention with potentially significant effects on patient anxiety, social sanction, and self-discharge rates.
7.Limitations
Several limitations of this work must be stated clearly. First and most significantly, the ED Adaptive Interface has not been tested with real patients or clinical staff. All design decisions, however well-grounded in published literature and social evidence, remain unvalidated in clinical practice. The prototype is ready for a structured pilot; it is not yet evidence-based in the clinical sense.
Second, the prototype’s current architecture lacks server-side data persistence, meaning that symptom logs, escalation alerts, and pain scores captured during a patient visit are not transmitted to or stored in any clinical information system. This limits the clinical utility of the Symptom Progression Log and Condition Worsening Escalation features in real deployment, pending the PMS integration described in Section 9.
Third, both social media corpora have the limitations described in Section 6.2: demographically skewed toward digitally active international residents, Facebook-platform only, and subject to group moderation effects. The thematic analyses represent qualitative corroboration from two independent sources, not quantitative evidence of design impact.
Fourth, language support, while broader than most comparable systems (seven languages in v5), remains substantially less than the full range of languages spoken by patients in Danish emergency departments. The prioritisation of Romanian, Polish, and Somali for v5 was informed by census data on non-Danish-speaking populations in the Region Hovedstaden catchment area, but necessarily excludes other communities.
Finally, WCAG 2.1 AA compliance of the high-contrast mode, while technically correct, has not been validated with patients who have specific colour vision deficiencies or complex visual impairments. WCAG compliance establishes a minimum standard; it does not guarantee usability for all members of the target population.
8.Conclusion
Forty-three per cent of emergency patients at Nordsjællands Hospital struggle with digital tools. This figure is the predictable consequence of deploying interfaces designed for baseline conditions in a setting where baseline conditions are routinely absent. The appropriate response is not to slow down digitalisation but to change how it is designed.
The ED Adaptive Interface demonstrates that it is technically and financially feasible to build a patient-facing digital tool that configures itself in under ten seconds to serve patients across the full spectrum of visual, motor, speech, and cognitive impairment — in seven languages, with zero infrastructure requirements, running on any existing hospital tablet. The impairment-first design philosophy that produced this prototype offers a transferable framework: begin with the most constrained user, solve their problem rigorously, and discover that you have also substantially improved the experience for everyone else.
The social media discourse analysis presented in this paper adds a form of evidence that formal patient satisfaction research rarely captures. Two independent social media corpora — 83 entries from the primary dataset and over 100 comments from the corroborating Aarhus thread — drawn from forums not designed for research, independently surfaced the same five categories of failure the prototype was built to address. That convergence is not proof of clinical efficacy; it is evidence that these problems are real, widely experienced, and structurally reproducible. Version 5 of the prototype, reported here for the first time, translates the findings of that analysis directly into implemented features.
The prototype is ready for a structured clinical pilot. The questions that remain — does it reduce nursing interruptions? Does it improve patient-reported anxiety? Does it increase pain score accuracy? — are answerable, and the roadmap for answering them is clear. The question is not whether a system like this can be built. It has been. The question is whether the institutions that need it most are ready to move.
"We don’t ask the patient to learn our technology; our technology learns the patient."
9.Future Work
9.1Clinical Pilot Design
The immediate next step is a structured clinical pilot at Nordsjællands Hospital. Phase 1 would involve usability testing with five to ten ED nurses and a comparable number of patients across two dedicated testing days, using the System Usability Scale (SUS) [16] as the primary quantitative instrument. Task completion rates — nurse setup time, first successful pain rating, help request submission — would provide supplementary evidence. Semi-structured interviews with nursing staff would surface workflow integration concerns that desk-based prototyping cannot anticipate.
Phase 2 would involve a four-week ward pilot on existing tablet hardware in the Emergency Department, without modification to the ward’s IT infrastructure. Patient-reported outcomes (anxiety, perceived communication quality, understanding of care process) would be collected at discharge using brief validated instruments. Nursing workload indicators — specifically the frequency of call-button activations and status-update requests — would be tracked against baseline data to provide a preliminary estimate of workflow impact.
Phase 3, contingent on Phase 2 findings, would address PMS integration for live workflow stage updates and export of patient-reported symptom data to nursing dashboards.
9.2Full MinSP Integration
MinSP is Denmark’s national citizen health record platform. If a patient’s accessibility profile, language preference, chronic condition history, and relevant medical context could be pre-populated from MinSP at arrival — triggered by scanning the patient’s yellow card (Sundhedskort) with an NFC reader or barcode scanner — the nurse setup step would be reduced to a single confirmation. This is a significant technical undertaking involving API access agreements with the Danish Health Data Authority, GDPR-compliant authentication protocols, and integration testing with the hospital’s existing PMS. The MinSP simulation feature in v5 demonstrates this concept and serves as a stakeholder communication tool for the integration roadmap.
9.3Additional Language Support
Version 5 supports seven languages. The most common non-Danish, non-English languages spoken by patients in the Region Hovedstaden catchment area include Urdu, Dari/Farsi, Tigrinya, Vietnamese, and several West African languages not currently supported. Expanding language coverage to the top fifteen non-Danish languages in the region is a realistic near-term goal; the translation framework in the current prototype is structured to accept new language objects without requiring changes to the interface logic. Any new language additions should be verified by bilingual clinicians or professional medical translators before deployment.
9.4Hospital PMS Integration for Live Workflow Updates
Currently, workflow stage transitions are triggered manually via the Staff Simulation Mode. In a deployed system, these transitions should be triggered automatically by the hospital’s PMS — specifically by status updates already being logged in the clinical workflow (blood test ordered, results received, doctor assigned). PMS integration represents the single highest-impact technical upgrade to the deployed system, allowing the workflow tracker and care notifications to function continuously without imposing any additional staff burden.
9.5Paediatric and Companion-Assisted Modes
The current prototype is designed for adult patients. Paediatric patients present distinct interface requirements: simpler language, larger visual elements, and a companion-mediated configuration where a parent or guardian interacts with the system on the child’s behalf. A companion-assisted mode — where a family member or carer is explicitly designated as the interface operator — would also support patients with severe cognitive impairment for whom nurse setup cannot anticipate all interaction needs.
Data Availability
The anonymised social media discourse dataset is archived by the corresponding author and is available to reviewers upon reasonable request. The prototype (ED Adaptive Interface v5) is a self-contained HTML file available from the corresponding author upon request. No other primary data were generated.
Acknowledgements
The initial solution concept was developed during the OpenInnovation 2026 Health challenge (February 2026), a cross-university student innovation sprint co-organised by DTU Skylab, Copenhagen School of Entrepreneurship (CBS), and KU Actory (University of Copenhagen), with Nordsjællands Hospital Emergency Department as case partner. The prototype, research methodology, social media analysis, and all work reported in this paper were developed independently by the author after the challenge period (February–April 2026). The author thanks Professor Thomas Andersen Schmidt and Jens-Peter Baatz Kristensen (Head of Innovation, Nordsjællands Hospital) for presenting the challenge brief and for their engagement with all participating teams. The social media discourse dataset was collected and anonymised by the author independently; no institutional data collection infrastructure was involved. Manuscript preparation assistance was provided by Atashi Irtiza, MBBS, DCO, Bangladesh Medical University. Generative AI tools were used to assist with drafting, writing, and prototype code development during the preparation of this manuscript. All scientific content, data collection, analysis, and conclusions are the sole responsibility of the authors.
Data Ethics Statement
The secondary evidence reported in Section 5 consists of publicly accessible social media content collected from a Facebook group ("Foreigners in Denmark") on 14 March 2026. The data collection and analysis were conducted in accordance with the following ethical provisions:
- Public accessibility: The group and the thread analysed were publicly accessible at the time of collection. No private group membership was required to view the content, and no authentication credentials were used to access it.
- No participant contact: No commenter was contacted before, during, or after the analysis. No consent was sought from individual participants, consistent with GDPR Article 89 provisions for research processing of publicly available data.
- Full anonymisation: All identifiable information — names, usernames, profile images, and any biographical details — was removed from the dataset and replaced with non-identifiable reference codes (C01–C81 for commenters, OP for the original poster) prior to any coding or analysis. No identifiable information appears in this paper or in the research archive.
- No participant harm: The analysis did not involve any interaction with participants and produced no information that could affect any individual’s healthcare, employment, legal status, or reputation.
- GDPR compliance: The processing of this data for research purposes is conducted under GDPR Article 89, which permits the use of personal data for scientific research where appropriate safeguards — including anonymisation — are applied. No institutional ethics approval is required for retrospective anonymised analysis of publicly posted content under Danish research practice for this class of data.
- Data archival: The anonymised dataset is archived by the author and is available to reviewers upon reasonable request. The original non-anonymised data has been permanently deleted from all author systems following the anonymisation process.
A second sample was gathered in April 2026 in a Facebook group that serves the international community in the Aarhus area of Denmark. All the above-described anonymisation, non-contact, and GDPR Article 89 provisions are applicable to this dataset. All identifiers such as names, usernames, profile pictures and any location or biographical information were deleted and substituted with non-identifiable reference codes before analysis. None of the participants were contacted. All author systems have been permanently destroyed with all the non-anonymised information. The researchers wrote this Data Ethics Statement, which is not an official institutional ethics board decision. The authors verify that this study design did not need any institutional ethics review according to the existing regulations.