Written by Technical Team | Last updated 17.10.2025 | 11 minute read
Designing digital health products is unlike building any other app. Lives, livelihoods, and clinical workflows are at stake, so the margin for error is slim. Great aesthetics and neat interactions are not enough; teams need a robust, ethical, and repeatable approach to understanding their users and the contexts in which care happens. That is the promise of UX research in healthcare: to reduce risk, surface real needs, and turn those needs into safer products that clinicians trust and patients actually use.
Yet the digital health landscape is complex. Patient journeys stretch across services and settings. Clinicians switch between paper notes, hospital EHRs, and mobile tools. Carers—both formal and informal—quietly stitch the experience together. Meanwhile, regulations, clinical safety processes, and data governance introduce constraints that can make research feel slow or intimidating to newcomers. Done well, however, those constraints sharpen the work: they force clarity about risk, value, and outcomes.
This article sets out a practical playbook for UX research in digital health, showing how to choose methods that fit clinical reality, handle regulatory demands without grinding to a halt, and translate insights into design decisions that move patient and system outcomes in the right direction. It is written for product managers, designers, clinicians, founders, and delivery teams who want research that is both humane and high quality.
The first rule of digital health research is deceptively simple: define “the user” precisely for every workflow and risk class. A medication reconciliation feature used by pharmacists in an acute admissions unit has very different requirements to a self-monitoring module used by people newly diagnosed with hypertension. In practice, most digital health products serve multiple audiences across a single journey—patients, clinicians, administrators, and carers—each with distinct tasks, constraints, and definitions of success. Treat them as separate user groups with their own jobs-to-be-done and you will avoid the common pitfall of designing a product that pleases everyone in the abstract and helps no one in reality.
For patients, the most important contextual factor is the lived experience of illness. Symptoms fluctuate; anxiety and fatigue impair attention; dexterity and vision can be compromised; stigma shapes disclosure. A “frictionless” task in a lab becomes a high-cognitive-load task when a patient is managing pain, negotiating transport, or caring for a child. UX research must go beyond idealised personas and explore how disease severity, comorbidities, and social determinants of health shape behaviour. That includes mapping not just what people say they do, but how they actually do it amid interruptions, fear, and trade-offs.
For clinicians, time is the scarcest resource and safety the overriding value. They are often operating inside complex socio-technical systems: team handovers, multidisciplinary meetings, pagers, and electronic records that are themselves a patchwork of modules. Research that respects clinical tempo—ward rounds, clinic slots, off-duty patterns—and documents the invisible glue of coordination (who prints, who calls, who double-checks) will pay dividends. Observe what “good” looks like in the current workflow, catalogue the sources of error and delay, and design around them. Your product is only as good as its fit with the last mile of care.
Carers and family members are the third leg of the stool, frequently overlooked. They fill documentation gaps, translate medical language, and implement care plans at home. The usability of home equipment, the burden of logging, and the clarity of alerts are as much their problem as the patient’s. Including carers in research sessions is therefore not optional when home management is in scope; it is the difference between a hopeful pilot and enduring adoption.
UX research in healthcare has to clear a higher bar, not because teams are distrusted but because patient safety demands evidence and traceability. That does not mean endless paperwork. With a bit of forethought, you can design a research programme that is compliant, proportionate to risk, and still fast enough to drive iterative delivery. Start with a research plan that nests inside your overall quality management and clinical safety frameworks. Describe user groups, objectives, methods, data handling, consent and withdrawal options, and how findings will influence design and risk controls. Keep it short, precise, and versioned.
Ethics and consent come next. Even if your research is not formal clinical research requiring institutional review, you must handle personal and sometimes special category data with care. Write consent materials in plain language that explains why you are doing the study, what will be recorded, how confidentiality is protected, and the right to stop at any time without affecting care. Where feasible, use role-play and simulated data for early explorations, moving to real-world data only when you have tight controls. If your product is likely to be a medical device or borderline device, ensure your usability engineering work can feed the relevant safety case and risk documentation; your research artefacts will be evidence later.
Recruitment is the practical bottleneck. Clinicians are time-poor and patients are not a market panel. Build relationships with clinical champions who understand your goals and can act as gatekeepers. Offer flexible scheduling and short, well-run sessions. Compensate patients for their time within local policies. For rare conditions or sensitive topics, work with patient advocacy groups and use asynchronous methods (e.g., diary studies) that reduce burden. When you cannot recruit enough representative users, prioritise depth over breadth and make your sample limitations explicit in the analysis; it is better to be honest and precise than to overgeneralise.
Finally, design your data minimisation strategy. Capture only what you need and store it only as long as necessary. Anonymise recordings swiftly, redact screens that reveal identifiers, and keep research notes separate from product databases. In healthcare settings, trust is the currency that buys you permission to continue learning. Small operational habits—arriving on time, using secure hardware, sending debriefs that close the loop—do more for your long-term access than any legal document.
Qualitative research is necessary to reveal the “why” behind clinical behaviour and patient choices. It is also the quickest way to uncover safety hazards, misaligned incentives, and the work-as-imagined versus work-as-done gap that plagues many deployments. Choose methods that let you see the environment, the artefacts in use, and the coping strategies people employ when systems do not align.
Contextual inquiry—observing tasks in situ while asking short, clarifying questions—should be your default method in clinical settings. Shadow ward rounds, sit in on triage, or observe pre-op assessments. Pay attention to what people check, who they consult, how they document, and which alarms they ignore. Capture the topology of the workspace: whiteboards, stickers on monitors, laminated quick guides. These are evidence of the tacit system your product must integrate with. When in-person access is impossible, ask clinicians to record short videos or screen-captures of typical workflows using de-identified cases; it is less ideal but often sufficient to understand sequencing and bottlenecks.
Interviews with patients and carers work best when they are short, frequent, and scaffolded by artefacts. Bring visual prompts: a service blueprint of their last clinic visit, a printed mock-up of a screen, or a timeline of symptom flares. Ask for stories rather than opinions: “Tell me about the last time you adjusted your insulin” beats “How would you like to adjust insulin?”. When topics are sensitive—sexual health, mental health—consider asynchronous methods like text-based interviews or diary studies that give participants control over pace and setting. For long-term conditions, diary studies capture the ebb and flow of symptoms and engagement far better than a single interview.
To make qualitative research efficient and cumulative, structure your outputs so they are easy to reuse. Write short “finding cards” that pair a quote or observation with the implication for design and risk. Maintain a living service map that shows where your product touches clinical pathways and home life. Keep a decision log that links design changes to research evidence; it will become part of your usability engineering file and help new team members ramp up fast.
When to choose a qualitative method:
Quantitative methods help you validate at scale, monitor change over time, and build the case for safety and effectiveness. They do not replace qualitative work; they complement it by showing how often patterns occur and whether design changes have the intended effect. In a regulated environment, choose metrics that are meaningful for clinical risk and user burden rather than vanity numbers.
Surveys are a common starting point. Standard instruments such as the System Usability Scale (SUS), UMUX-Lite, or single-item measures like the SEQ can give you a signal without adding heavy respondent burden. Resist the temptation to over-survey patients; keep instruments short, accessible, and available in the languages relevant to your population. For clinicians, tie survey cadence to release cycles or training events rather than spamming them during busy periods. Always link survey responses to contextual variables—role, experience, setting—so you can stratify findings. A SUS score of 72 means very different things in an emergency department and in a community clinic.
Product analytics in healthcare need to be privacy-preserving and clinically relevant. Track task-level events (e.g., “entered dose”, “acknowledged alert”, “completed triage”) rather than just generic page views. Time-on-task, completion rates, and error rates can be diagnostic if you have trustworthy task definitions. Be transparent about what you track and why; clinicians and patients are rightly wary of telemetry that feels intrusive. Where possible, use on-device processing and configurable logging levels to respect local policies. In enterprise deployments, align your dashboards to service outcomes that matter—reduced time to escalate, fewer incomplete referrals—so your metrics speak the language of the organisation.
Remote unmoderated testing is tempting when recruitment is hard, but treat it carefully in healthcare. Unmoderated tasks with simulated patient data are useful for basic navigation and labelling decisions. Anything that touches clinical judgement, dosing, or triage should be moderated with clear scenario boundaries and a facilitator trained to halt the session if unsafe mental models emerge. If you run A/B tests, pre-define stopping rules that consider not only click-through but also potential safety signals, and secure sign-off from your clinical safety lead before live experiments.
Quantitative signals to consider:
Research only matters if it changes the product. In digital health, that means translating insights into design decisions that are traceable, testable, and aligned with clinical safety and regulatory goals. Start by framing findings in terms of risks and controls. If clinicians consistently misinterpret an abbreviation in your UI, the problem is not merely “confusing copy”; it is a potential use error with a severity level. That framing forces clarity on the design response: change terminology, add confirmation, restructure the flow, or remove the risky option. Document the rationale and link it to the evidence.
Prototyping strategies should match risk. Use low-fidelity sketches for early alignment and to explore the shape of workflows without anchoring on visual polish. Shift to high-fidelity, data-realistic prototypes for anything involving calculations, alerts, or clinical decision support—this is where micro-copy, units, and default values surface real hazards. Build in “safety scaffolding” as you iterate: progressive disclosure for advanced settings, constrained inputs for numeric fields, contextual explanations tied to clinician mental models, and clear, consistent error recoveries that never erase data without consent.
Close the loop with a cadence of formative and summative tests. Formative tests happen early and often; they help you learn. Summative tests are structured to demonstrate that the final design supports safe and effective use by intended users in intended use environments. In practice, you will be doing both across the life of a product, but the shift to summative thinking—controlled scenarios, clear success criteria, representative samples—matters before wide deployment. Pair this with a training and onboarding plan that your research has validated; a good design can still fail if the organisation introduces it badly.
Finally, remember that adoption in healthcare is a team sport. Share artefacts that help colleagues make informed decisions: short video clips that show a usability risk in action; annotated screenshots that explain a control; a one-page brief that summarises research evidence for a change request. When teams can see what you saw, they are more likely to support the right trade-offs. Over time, this builds a culture where UX research is not an external checkpoint but a core mechanism for delivering safer, kinder, and more effective digital health care.
Is your team looking for help with digital health design? Click the button below.
Get in touch