Dialysis Machine UX: What the Interface Gap Is Costing Your Organisation

This article draws on Creative Navy's project work in medtech UX, spanning practice management software, surgical equipment, ventilators, blood pumps, infusion systems, and patient monitoring devices, including Class II and Class III regulated products. Our work in this sector covers clinical environments including the ICU and operating theatre, designing for surgeons, nurses, and biomedical engineers. Dennis Lenard, who leads this work at Creative Navy, is the author of User Interface Design For Medical Devices And Software, the practitioner reference on UX design for medical devices and software. Our approach integrates IEC 62366 usability engineering requirements and FDA Human Factors guidance as structural inputs to the design process, not post-hoc compliance activities.

Six to eight weeks. That is the standard training period for home haemodialysis, reported consistently across programmes in the United States and Europe (PMC, 2024). Not because kidney replacement therapy is complex to understand. Because the interfaces in this product category were designed around software module logic rather than the sequences clinicians and patients actually follow, and nobody has absorbed the cost of that decision directly enough to fix it.

The gap between that figure and what purpose-built machines now achieve is measurable and public. Outset Medical's Tablo is vendor-reported to bring new clinical staff to competent operation in under four hours. The Fresenius 6008 CareSystem, compared against its predecessor in a peer-reviewed time-motion study, reduced handling steps by 26% and saved 2.83 minutes per treatment session (Gründler et al., 2021). These are not projections from a redesign proposal. They are compounding results from organisations that treated the interface as an infrastructure decision five years ago and are now signing multiyear network agreements.

This review benchmarks four established dialysis machine platforms against a common evaluation framework and sets them alongside the entrants now gaining ground. The audience is product directors and senior PMs. The conclusion is that interface design in this product category is no longer a UX project. It is a competitive position, and for most incumbent platforms, the position has already weakened.

The Sense Decay Audit is the evaluation framework applied in this review. It asks a single question of each platform: has the interface remained coherent with how clinicians and patients actually use the machine, or has it drifted toward serving internal product logic at the cost of operational clarity?

Three criteria drive the assessment.

Operational coherence examines whether the interface supports the clinical task sequence as it runs in a real unit, including under alarm conditions, shift handover, and patient movement. Designed-to-ideal and designed-to-operational are rarely the same.

Learning cost treats training duration as a design output rather than a staffing input. When onboarding runs to six weeks, the curriculum is compensating for interface ambiguity. The question is how much of that curriculum the design created.

Safety signal integrity asks whether the alerts, warnings, and state changes the interface generates support the clinical decisions they are supposed to prompt, or produce noise that clinicians learn to discount before the harm event occurs.

The Sense Decay Audit identifies interfaces that have drifted from the operational reality they were designed to serve. Applied to dialysis machines, it assesses operational coherence with actual clinical task sequences rather than idealised ones; learning cost as a variable the interface controls rather than one training absorbs; and safety signal integrity, meaning whether alarm behaviour prompts the right clinical decisions under the conditions of a real shift. An interface showing sense decay is one whose design assumptions no longer match the environment in which it runs.

The platforms reviewed, along with their market status as of Q1 2026:

The reviews below also assess two entrants whose interface decisions now define the competitive reference for this product category.

The Dialog+ is still in production. B Braun's most recent publicly available instructions for use (Rev. 2.18.02, July 2021) reference the same icon-heavy touchscreen with physical hard keys that characterised the platform in 2015. The company's own marketing copy continues to lead with "innovative use of icons on a colour touch screen."

The icon problem is not aesthetic. When icons carry no labels, every new staff member must memorise a separate symbol vocabulary before they can operate the machine with any confidence. That vocabulary cannot be built through task interaction alone; it requires explicit instruction. This is the mechanism by which interface design creates training curricula: the machine does not explain itself, so the training programme must.

The modal window behaviour compounds the problem. Input confirmations on the Dialog+ inconsistently require tapping outside an overlay to save, while other modal interactions behave conventionally. The result is two competing mental models for interactions that look identical. Clinicians working across multiple machine types in a unit, or returning from leave, carry a non-trivial cognitive reset at the start of each session.

What the Dialog+ reveals about the broader problem: an interface that has not been rebuilt since 2015 is not just frozen. It is encoding staffing and training assumptions from a previous period into every new deployment. The clinical environments where it now runs have changed. The interface has not.

B Braun has introduced the Dialog iQ as a next-generation platform with cassette-based circuit automation. The iQ may resolve some of these problems. No independent interface evaluation confirms that it does.

The AK 98 received FDA 510(k) clearance for a redesigned version in March 2021. Baxter, now Vantive following its kidney care spin-off in 2024, marketed the update on interface grounds: an app-like interface with customisable parameters and two-way EMR connectivity. Both additions are substantive.

Direct EMR prescription pull removes a manual transcription step. Automatic Alert Resolution, which self-clears pressure alarms triggered by patient movement rather than clinical events, addresses a documented harm pathway. A 2015 clinical analysis in the Clinical Journal of the American Society of Nephrology reconstructed a patient adverse event in which normalised alarm silencing had become the unit default. Designing out the class of alert that produces this pattern is not a cosmetic improvement.

The AK 98 is the closest among the established platforms to treating interface design as an operational system rather than a control panel.

What remains uncertain is whether the structural navigation logic has been rebuilt alongside these additions, or whether the update layers new functionality onto an older architectural base. In clinical device projects, the persistent finding is that the visual layer rarely accounts for more than a third of the usability problem. The navigation architecture, and specifically whether screen transitions match the procedural logic the operator is executing, is where training burden and error risk actually live. A medical device interface project we completed for a surgical tools manufacturer showed this directly: reorganising screens by procedural relevance rather than software modules was the intervention that reduced operating theatre errors, not the visual refresh that preceded it. Details are documented in our IEC 62366 case study for deSoutter Medical. Whether the redesigned AK 98 made the equivalent structural change is not confirmed by the available evidence.

What the AK 98 reveals about the broader problem: surface improvements are documentable and marketable. Structural improvements are harder to verify from outside. The distinction matters because training burden and error rate are functions of structure, not surface.

The DBB-EXA's D-FAS (Dialysis Full Assist System) is the strongest procedural automation feature in the benchmarked group. It handles bloodline preparation steps that conventional machines require operators to sequence manually, reducing both time and working memory load on clinicians. This is genuine functional progress.

The information architecture has not moved. The "Setting 2" menu label, which opens a navigation path with no signal of what category of settings it contains, has not changed in any publicly available documentation. This is a structural failure in information scent: operators cannot predict what they will find by following a navigation path, so they cannot plan their interaction before executing it. Exploratory navigation in a time-pressured clinical environment is not an acceptable operating model.

Nikkiso launched the DBB-06 PRO in the US in November 2025, positioning it as the first US haemodialysis machine with Full Assist functionality. This is a market entry, not an interface redesign. The automation capability moves to a new geography. The labelling problem moves with it.

What the DBB-EXA reveals about the broader problem: a valuable functional layer sitting on an information architecture that was not designed to support it creates a specific kind of sense decay. The automation reduces procedural steps; the navigation model still requires expert knowledge to traverse. These are not equivalent improvements, and they do not cancel each other out.

The Surdial X presents two problems that compound across a treatment session. Interactive values do not signal their interactivity; elements that can be tapped look identical to display-only fields. Operators learn the distinction through trial and error. In equipment handling critically ill patients, trial-and-error interaction design is a training cost that never appears on the product line budget.

The per-digit plus/minus input mechanism for numerical parameter changes requires a disproportionate number of interactions when the target value differs significantly from the current one. Standard numeric entry, where operators select individual digits directly, is measurably faster across the majority of parameter adjustment scenarios. The interaction cost is not hypothetical. It accumulates across every session.

Nipro's response to the US market has been the Surdial DX, commercially launched in 2022 with step-by-step on-screen guidance and reduced error risk as explicit selling points. The DX's existence is the market signal: Nipro concluded the Surdial X was not competitive on interface grounds and built a new product rather than updating the existing one. That is a decision with a price that does not appear in the current product line's balance sheet.

What the Surdial X reveals about the broader problem: when interaction cost accumulates invisibly across sessions, organisations absorb it as staff time without ever attributing it to the interface. The cost is real. The attribution is absent.

Two entrants are operating from a shared premise: the interface is the product, not a feature of the product.

Outset Medical Tablo uses 3D animations and conversational step-by-step setup instructions to guide both clinical staff and patients through operation. Vendor-reported training times for clinical staff are under four hours, against the six-to-eight-week industry standard. In April 2024, US Renal Care signed a multiyear agreement to deploy Tablo for home dialysis across 33 states. In January 2026, the next-generation Tablo received FDA cybersecurity clearance as the first dialysis system to meet the agency's most recent cybersecurity guidance.

Fresenius Medical Care 6008 CareSystem, launched in 2018, offers the most rigorously evidenced interface improvement figure in this product category. A time-motion study published in Medical Devices: Evidence and Research (Gründler et al., November 2021) compared the 6008 against its predecessor. Results: handling steps reduced by 26% for fistula access, total treatment process time reduced by 2.83 minutes per session, hand grips and thumb presses reduced by approximately 50%. The study was funded by Fresenius Medical Care with declared independent data collection by third-party researchers. The methodology is peer-reviewed. At a unit running three sessions per bay per day, 2.83 minutes per treatment across a quarter compounds into measurable staff capacity.

Neither organisation improved an existing interface. Both started from a different premise about what the interface is for.

The dominant response to UX-related incidents in dialysis units is training reinforcement. When nurses silence alarms habitually, the intervention is protocol review. When home HD uptake is low among eligible patients, the response is better patient education. This framing is wrong, and the published evidence says so directly.

A 2023 study using a Fresenius 2008K machine found that over 70% of patients with arteriovenous fistulas had vascular access pressures too low to trigger the standard needle dislodgement alarm (Marcén et al., PMC10278846, 2023). The threshold was not misconfigured by staff. It does not match the physiological reality of the patient population. Competent, well-trained clinicians cannot respond to a safety signal the machine does not generate.

A participatory design study conducted in a Swedish dialysis department observed nurses walking into corridors to locate which machine was alarming, because the corridor speaker was louder than the bedside indicator. Three nurses responded to a single alarm because the signal origin was ambiguous in the ward environment. These are failures of spatial information architecture in a live clinical environment, of the kind that clinical field observation identifies and controlled laboratory usability studies do not replicate. IEC 62366 summative testing does not require reproduction of these conditions.

Emergo by UL, a human factors consultancy, lists as a completed case study a haemodialysis machine interface that induced clinical staff to overlook over-ultrafiltration, resulting in patient death or injury. The case exists because the harm occurred first. The investigation followed. Training the staff who were operating that machine would not have changed the interface condition that produced the harm.

The cross-platform pattern is consistent. All four incumbents show the same structural decay. A layer of procedural or connectivity improvement has been applied to interface architectures from an earlier period. The automation in the DBB-EXA does not address the navigation problem. The EMR connectivity in the AK 98 may not address the screen sequence logic. The Surdial DX exists because the Surdial X could not hold a US market position where interface quality is now a procurement criterion.

The entrants gaining ground have not improved an existing model. They have replaced the premise.

Across the clinical device projects we have worked on, the finding that consistently surprises product teams is where training burden actually originates. The assumption going in is that clinicians need time to learn the clinical procedure. What the evidence consistently shows is that the procedure is understood quickly. It is interface inconsistencies that extend training: the modals that save unexpectedly, the icon sets that require separate memorisation, the navigation paths with no information scent. These are the actual training curriculum. Design them out and onboarding duration falls accordingly.

The argument that the dialysis machine market moves slowly because of regulatory and reimbursement friction is partly accurate and largely used to avoid a sharper question: how much ground is a product losing while defending in-centre positions?

The business case for dialysis machine interface investment rests on a peer-reviewed time-motion study. Gründler et al. (2021) compared the Fresenius 5008 CorDiax and 6008 CareSystem: handling steps fell 26%, treatment session time dropped 2.83 minutes. At a unit running three sessions per bay per day, this compounds into measurable staff capacity. The larger variable is training: moving from six to eight weeks of onboarding to under four hours is an order-of-magnitude shift in deployment economics.

Interface drift is an economic problem with a known trajectory. Each software release that adds a feature without rebuilding the structure it sits in extends the training curve and raises the cognitive overhead of competent operation incrementally. At some point the cost of restructuring exceeds the cost of a new platform, which is exactly the decision Nipro made when it built the Surdial DX rather than updating the Surdial X. That decision has a price. It does not appear in the product line budget for the year the decision was deferred.

The home dialysis market is expanding regardless of what incumbent platforms do next. The IM-HOME national survey (AJKD, June 2024) found that provider-perceived barriers to home HD are dominated by machine complexity and patient education burden. Interface complexity is the primary reported barrier to home HD uptake among both providers and patients. Outset, Nipro with the Surdial DX, and NxStage with the Versi HD are building install bases in that market. The incumbents whose platforms remain in their 2015 interface generation are not losing ground slowly.

Three principles follow from this analysis.

Treat interface reconstruction as a capital decision. Every product cycle in which the interface architecture is not rebuilt is a cycle in which the training cost compounds and the competitive gap widens. The cost of deferral is real; it accumulates off-budget.

Validate in conditions that replicate actual clinical use. The signals that precede structural redesign are present in the environment long before they appear in post-market surveillance data. Alarm density, fatigue conditions, multi-machine environments, and corridor acoustics are not reproduced in controlled usability lab studies. An organisation that validates only in lab conditions will learn about its interface's failure modes later than an organisation that validates in the field.

Distinguish surface from structure. EMR connectivity and alert resolution are measurable, documentable improvements. They are also visible to procurement teams and marketable to clinical directors. The harder question, and the more consequential one, is whether the screen sequence logic has been rebuilt. Training burden and error rate are functions of structure. Surface improvements do not reach them.

Several conclusions in this review warrant qualification.

Training time reductions reported for Outset Medical Tablo (under four hours) and the Surdial DX are vendor-reported figures. Both are commercially plausible given the scope of the interface redesign, but neither has been independently replicated in peer-reviewed literature as of March 2026.

The Gründler et al. time-motion study was funded by Fresenius Medical Care. The authors declare this, and data collection was carried out by independent third-party researchers. The methodology is published. It is the strongest business-impact figure available in this product category, but the funding relationship warrants care in interpretation.

The Dialog iQ, the DBB-06 PRO, and the Surdial DX have not been independently benchmarked for the interface problems identified in their predecessors. It is possible that B Braun's iQ resolves the icon ambiguity and modal inconsistency documented in the Dialog+. No public evidence base confirms this.

The most significant unresolved question is whether any of the 2021 to 2025 interface updates to incumbent platforms have addressed structural navigation logic, or have layered new functionality onto existing screen architectures. That distinction determines whether measurable improvements in EMR connectivity and alarm resolution translate into reduced training burden, or introduce further complexity into a structural model that has already exceeded its carrying capacity.

A dialysis machine that requires six to eight weeks of training to operate is not managing clinical complexity. It is managing design debt.

The gap between that figure and the four-hour training time now in the market is not a research gap. It is the accumulated consequence of treating the interface as a surface concern for years while competitor organisations treated it as an infrastructure decision. Those decisions were made in 2018, 2020, and 2021. The advantages are compounding now.

The in-centre market still belongs to the incumbents. The home HD market is being built by organisations that understood, five years ago, that training burden was the primary barrier to the modality that delivers better outcomes at lower system cost. Those organisations are now signing multiyear network agreements and receiving next-generation regulatory clearances. The incumbents whose platforms are still in their 2015 interface generation are not losing ground slowly.

The question for the next product cycle is not whether the current interface is IEC 62366 compliant. It is whether it is a competitive liability that has not yet been priced.

What is the Sense Decay Audit for dialysis machine UX? The Sense Decay Audit assesses whether a dialysis machine interface has remained coherent with how clinicians and patients actually use the device, or has drifted toward internal product logic. It evaluates three dimensions: operational coherence with real clinical task sequences, learning cost as a design output rather than a training problem, and safety signal integrity under actual working conditions rather than controlled lab settings.

Why does dialysis machine training take six to eight weeks? Up to 100 individual setup steps in a conventional machine (PMC, 2024), combined with modal interactions that behave inconsistently and icon sets without labels, mean procedural knowledge cannot be built through task interaction alone. Training compensates for interface ambiguity. Machines designed around guided task sequences, such as the Outset Medical Tablo with 3D animation and step-by-step setup instructions, have reduced clinical staff training to under four hours in vendor-reported data.

What changed in the Vantive AK 98 2021 redesign? The 2021 FDA 510(k) clearance for the updated AK 98 added two-way EMR connectivity (enabling direct prescription pull from Epic and Cerner) and Automatic Alert Resolution, which self-clears pressure alarms caused by patient movement rather than clinical events. The AK 98 is now sold by Vantive, spun off from Baxter in 2024. Whether the underlying navigation architecture was rebuilt alongside these additions is not confirmed by available documentation.

Does IEC 62366 compliance prevent dialysis machine UX failures? IEC 62366 provides the regulatory framework for usability engineering in medical devices. It does not require that validation environments replicate actual clinical conditions. Alarm density, interruption patterns, multi-machine ward environments, and extended-shift fatigue are not routinely reproduced in controlled usability lab studies. The gap between lab performance and field performance is where most documented harm in this product category occurs.

What is the business case for dialysis machine interface investment? The Gründler et al. time-motion study (Medical Devices: Evidence and Research, 2021) found a 26% reduction in handling steps and 2.83 minutes saved per treatment session when the Fresenius 6008 CareSystem replaced its predecessor. At a unit running three sessions per bay per day, this compounds into measurable staff capacity across a quarter. Training cost reduction is the larger variable: moving from six to eight weeks of onboarding to under four hours represents an order-of-magnitude shift in clinical deployment economics.

Which dialysis machine has the best interface for home use? The Outset Medical Tablo is the most distinct from conventional machine interaction logic among the platforms reviewed. It was designed for patient-operated home dialysis, with 3D animation and step-by-step guidance replacing mode-based interaction. FDA clearance for home use was granted in 2020, with a next-generation platform receiving cybersecurity-qualified clearance in January 2026. Independent comparative usability testing against conventional platforms in home settings has not been published as of March 2026.

Gründler, W., Schöning, L., Segelmark, M., Poulter, M., & Holmbeck, T. (2021). Comparison of the time-motion during hemodialysis treatment preparation between the Fresenius 5008 CorDiax and 6008 CareSystem. Medical Devices: Evidence and Research, 14, 411–419. https://doi.org/10.2147/MDER.S330044

Harasemiw, O., Hurton, S., MacKay, K., Murray, J., Tangri, N., & Komenda, P. (2019). Human factors testing of the Quanta SC+ home hemodialysis system. Hemodialysis International, 23(4), 494–505. https://doi.org/10.1111/hdi.12756

Kliger, A. S. (2015). Maintaining safety in the dialysis facility. Clinical Journal of the American Society of Nephrology, 10(4), 688–695. https://doi.org/10.2215/CJN.08960914

Wish, J. B., Sedor, J., Dharmarajan, S., & Bhatt, U. (2024). Facilitators and barriers to home dialysis: The IM-HOME national survey. American Journal of Kidney Diseases. https://doi.org/10.1053/j.ajkd.2024.02.007

Marcén, R., et al. (2023). Venous needle dislodgement detection in hemodialysis: Clinical vulnerability analysis. PMC10278846. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10278846/

Patient training and patient safety in home hemodialysis. (2024). PMC. https://www.ncbi.nlm.nih.gov/pmc/

Baxter International Inc. (2021, March 12). Baxter receives FDA clearance for next-generation AK 98 hemodialysis machine. BusinessWire. https://www.businesswire.com/news/home/20210312005185/en/

Emergo by UL. Human factors engineering services: Hemodialysis machine case study. Emergo by UL. https://www.emergobyul.com/services/human-factors-engineering

B Braun. (2021). Dialog+ instructions for use (IFU 38910540AU, Rev. 2.18.02). ManualsLib. https://www.manualslib.com/