As AI expands into healthcare at home, nurses say wound care still needs the human touch

When nurses arrive at a patient’s home to treat chronic wounds, they’re doing more than changing bandages. Caregivers build trust through repeated visits, use touch and smell to detect infection, and improvise when the kitchen becomes an examination room.

Now, as AI tools increasingly enter healthcare settings, a new study reveals concerns about whether such technologies can accommodate the complex, relationship-driven work that defines home healthcare nursing. The research, conducted with 14 registered nurses across two Swedish municipalities, explored how they perceive wound care and the potential role of AI in their practice.

AI promises efficiency, nurses see its limits

The European Commission has estimated that AI could help address workforce shortages and improve diagnostic accuracy. Yet the Swedish nurses’ experiences suggest that technical capability cannot itself solve all the challenges associated with healthcare.

“Nurses emphasized the importance of relational work in wound care, highlighting the trust and continuity necessary for effective wound care, which AI-driven automation might overlook,” the researchers wrote.

Home healthcare represents one of the toughest environments for automation. Unlike hospitals with standardized equipment, nurses work in patients’ living spaces, where lighting may be poor, surfaces are not sterile, and every encounter requires some amount of improvisation.

“You may have nowhere to put your bag or set up your stuff in a home environment,” one nurse told the study’s researchers. “Sometimes it’s a bad working environment and bad lighting, and it’s hard to get through. You might pick up a kitchen chair to use as a table, so you simply have to improvise.”

The nurses identified three aspects of their work that complicate AI integration: relational, embodied, and adaptive practices. Relational care refers to the interpersonal and trust-based connections nurses build with patients over time, forming the emotional and communicative foundation for effective healing. Nurses said that continuity of visits builds trust, improves adherence to treatment, and encourages patients to discuss issues beyond the wound itself.

“It becomes routine. It becomes security,” one nurse said. “And you have this little moment to talk to each other as well. These are the kinds of things that are important, even though they are not so visible.”

Embodied knowledge, the sensory and experiential skills nurses develop through direct, hands-on care, presents another barrier for AI-powered automation. Wound assessment depends on sensory engagement (seeing color gradients, feeling tissue texture, detecting infection by smell) that no algorithm can reproduce.

“A picture might make fibrin look yellow, but in fact, it’s gray-green,” one nurse noted. “Or if I suspect pseudomonas, I can sometimes tell by the slightly unpleasant smell and the way the fibrin shimmers green in the wound.” Although AI-based image recognition can assist with wound classification, the nurses doubted that such systems could capture the subtle, multisensory data that informs real-world judgment.

Adaptive practice shapes every home visit, as well. Nurses must tailor care to each person’s health goals, living conditions, and social supports. “You have to remember that all people are different,” another nurse said. “Two patients with the same health problem may not be looking for the same solution. They may not have the same goals at all.”

Philosophy of care challenges the push for standardization

The researchers drew on philosopher Annemarie Mol’s “logic of care,” which holds that good care develops through ongoing, context-specific interaction rather than fixed protocols. Mol contrasts this with the “logic of choice,” which treats healthcare as a series of individual decisions instead of a continuous, adaptive relationship. Viewed through this framework, the nurses’ skepticism reflects a broader worry that automation may replace responsiveness with rigidity.

Sweden’s current stage of AI adoption helps explain their caution. According to the study, most regional projects remain pilots, and municipalities rely mainly on basic digital documentation systems. Advanced machine-learning tools are still rare in everyday nursing practice.

The study authors acknowledged that nurses’ views were shaped more by expectations than by direct experience with AI systems. Yet their apprehension mirrors international findings that clinicians fear technology will narrow their professional discretion rather than support it.

Design flaws undermine digital health tools

They recognized potential advantages in applications such as wound-image analysis, diagnostic decision support, and care-team communication. What they questioned was whether these tools would support clinical reasoning or add new layers of complexity and constraint.

That tension reflects a recurring theme in nursing research; digital systems are often designed around administrative efficiency instead of clinical sense-making. Research has found that nurses frequently perceive new technologies as imposed from above, with limited opportunity to shape design or workflow.

When that happens, digital innovation can erode the elements (time, attention, and continuity) that make care effective. The Swedish findings echo those warnings, suggesting that AI could either amplify or undermine nursing practice depending on how it is implemented.

Nurses need a voice in designing AI tools

The study also highlights an enduring imbalance in who gets heard during technology rollouts. Across health systems, nurses are among the largest professional groups but are often excluded from early design discussions. As a result, tools may reflect managerial or engineering priorities more than clinical realities.

Research has shown that involving nurses in system design improves adoption rates and reduces workflow disruptions. The Swedish authors argue that similar inclusion will be crucial as AI expands into home healthcare.

The implications for policymakers and developers are clear. For AI to be effective in home healthcare, it must align with the embodied, relationship-driven nature of nursing. That means creating systems flexible enough to adapt to each patient’s needs, designed to support rather than replace clinical judgment, and structured to protect the time nurses need to build trust and connection.