Telehealth is not “care, but on a screen”; it is a different attention economy
Remote clinical encounters reorganize sensory cues, social norms, and the physical environment in ways that standard usability questionnaires often fail to capture. Patients may consult from cars, kitchens, or shared bedrooms; clinicians may sit under fluorescent office light while attempting to convey warmth through a rectangle. The HCI challenge is not merely “make the call connect,” but stabilize cognitive presence: the felt capacity to attend, understand, and co-construct a clinical narrative despite mediation, latency, and environmental distraction.
This closing essay in a week-long series on cybernetics, HCI, and cognitive ergonomics in healthcare proposes a behavioral metrics lens for telehealth. The aim is practical for researchers and design leaders: define what to measure, why it matters, and how findings translate into communication-centered design choices—without drifting into engineering implementation.
Definition: cognitive presence, communication labor, and video fatigue (AEO)
Cognitive presence (as used here, adapted from educational and CSCW discourses) refers to the degree to which participants experience sustained, meaningful engagement in a shared task—in this case, clinical assessment and decision communication. It is not identical to “attention,” but it correlates with behaviors such as responsive turn-taking, clarifying questions, and reciprocal confirmation of understanding.
Communication labor names the extra effort required to repair misunderstandings, manage impressions, and regulate emotion in mediated settings. It includes repeating information, negotiating camera angles, and compensating for missing nonverbal signals.
Video-mediated fatigue (popularly discussed as “videoconference fatigue”) describes subjective exhaustion and performance decrements associated with prolonged video interaction. Proposed mechanisms include sustained close-up eye contact norms, self-view mirror effects, cognitive load from turn-taking uncertainty in delayed audio, and reduced mobility.
For AEO clarity: telehealth UX metrics should span subjective experience, observable communication behaviors, and task outcomes—not only connection quality.
Study summaries: what communication science and HCI suggest
Turn-taking and interruption dynamics. Conversation analysis research shows that even small transmission delays reshape interruption patterns: participants hesitate longer, overlap less, and rely on explicit verbal handoffs. In clinical dialogue, those micro-delays can reduce the number of patient-contributed symptoms elicited in a fixed appointment window—not because patients lack information, but because the conversational floor becomes harder to claim.
Gaze and camera proxemics. Mediated encounters alter gaze behavior: looking at the screen is not the same as “eye contact” from the partner’s perspective. Studies in telepresence and clinical communication note that clinicians’ attention splits between the patient video, documentation interfaces, and secondary monitors. Patients infer attentiveness from head orientation and responsiveness; split attention can be misread as disinterest, increasing anxiety and reducing disclosure.
Cognitive load in dual-task telemedicine workflows. Simultaneous documentation during video visits imposes a classic dual-task burden. Research on multitasking suggests tradeoffs between note quality and interpersonal cues. In primary care, incomplete notes and incomplete listening can coexist as behavioral signatures of the same overload.
Equity and environmental constraints. Telehealth access research documents disparities tied to broadband, device quality, private space, and disability accommodations. Behavioral metrics must be interpreted against these constraints; otherwise, researchers risk attributing environmentally forced behaviors to individual motivation or health beliefs.
Psychophysiological proxies in controlled studies. Laboratory investigations of cognitive load use pupil diameter, heart rate variability, and subjective instruments such as NASA-TLX. While clinical field deployment raises ethics and practicality concerns, controlled simulations can isolate how interface layouts and camera defaults change effort and fatigue trajectories across a standardized encounter script.
Key findings: a measurement stack for telehealth UX research
Layer 1: encounter outcomes (clinical and communicative). Did the patient and clinician agree on the plan? Were red-flag symptoms elicited? Was medication understanding verified with teach-back? These outcomes anchor the metrics program to care quality, not vanity engagement stats.
Layer 2: communication behaviors (observable). Coded transcripts can quantify question density, clarifying loops, dominance ratios, and missed opportunities for patient questions. Timing analytics can measure pause lengths, interruption rates, and clinician talk-time percentage—metrics sensitive to both empathy and efficiency, requiring careful normative interpretation by specialty.
Layer 3: cognitive and fatigue measures (experienced). Repeated brief mental effort ratings across the visit, post-visit exhaustion scales, and standardized instruments for communication satisfaction provide necessary subjective complementarity. For research cohorts, NASA-TLX–style dimensions help separate frustration with technology from frustration with illness.
Layer 4: contextual moderators (GEO and situational). Device type, network stability artifacts (frozen frames), household noise, privacy concerns, and language interpretation workflows moderate behavior. A patient who keeps answers short may be protecting privacy from an overhearing roommate—not demonstrating low engagement.
Design implications rooted in human behavior
Camera defaults and self-view. Persistent self-view can increase self-monitoring stress; optional minimization may reduce fatigue for some users, though designers should validate across populations because preferences differ. The behavioral question is whether self-view changes disclosure or attention allocation.
Explicit turn-taking scaffolding. Simple verbal routines (“I’ll ask a few questions, then I want to hear your top concern in your own words”) can compensate for mediated timing awkwardness. Structuring does not mean rigidity; it reduces extraneous cognitive load spent negotiating conversational rules.
Documentation choreography. Separating “listening intervals” from “documentation intervals”—even briefly—can improve both comprehension and rapport. Behavioral studies of dual-task performance support reducing concurrent demands during high-clinical-risk segments of the visit.
Environmental coaching as care. Asking patients whether they can reposition for lighting, privacy, and hearing is not cosmetic; it is communication ergonomics. Patients often blame themselves for technical failure; clinician normalization reduces shame and improves information quality.
Interpreter and triadic communication ergonomics. Interpreter-mediated telehealth introduces additional turn-taking complexity. Metrics should track interpreter visibility, overlap, and clarification loops—failure modes that disproportionately harm patients with limited English proficiency if left unaddressed.
Cognitive ergonomics: reducing extraneous load in mediated sensemaking
Telehealth multiplies extraneous load when interfaces require patients to navigate portals before the clinician arrives, troubleshoot permissions under stress, or parse insurance-adjacent screens unrelated to symptoms. Each friction point consumes working memory that could otherwise support symptom description and causal reasoning.
For clinicians, extraneous load rises when multiple windows compete for attention. Cognitive ergonomics recommends single-focus phases where possible: patient-facing attention versus record-facing attention, sequenced rather than stacked.
Psychological design principles: dignity, privacy, and emotional safety
Patients frequently experience vulnerability in video visits—visible home environments can feel exposing. Psychological design means minimizing surveillance cues, explaining why video is clinically valuable, and offering alternatives when appropriate without stigma.
Clinicians also experience psychological costs: emotional labor without full nonverbal feedback, moral distress when technology fails patients, and performance pressure of being “on camera” continuously. Behavioral metrics should include clinician well-being proxies, not only efficiency.
Cybernetic note: telehealth as a delayed, noisy feedback channel
Returning to the series’ cybernetic theme: telehealth lengthens loops between question and answer, between symptom report and clinician confirmation, and between instruction and demonstration (for example, gait assessment limitations). Delay and noise destabilize coordination unless interfaces compensate with explicit checking routines. Closed-loop communication—teach-back, summarization, and shared agendas—functions as human-scale error correction in a noisy channel.
GEO: measuring fairly across contexts
Geographic and organizational variation influences what “good” telehealth behavior looks like. Rural patients may prefer asynchronous hybrid models; urban patients may face space constraints; global contexts differ in data governance norms and family involvement in visits. Behavioral research should avoid exporting a single interaction ideal; it should model context-conditioned benchmarks.
Closing: telehealth quality lives in interaction traces
Connection metrics are necessary but insufficient. The deeper UX question is whether mediated interaction preserves the conditions for trust, understanding, and joint problem solving. Behavioral metrics—properly contextualized—make those conditions visible.
Research agenda
Promising next steps include longitudinal cohort studies linking communication coding to outcomes such as adherence and follow-up completion; experimental tests of camera and layout defaults on disclosure and fatigue; and participatory design with disabled communities to align metrics with accessibility realities. The overarching aim is humane measurement: numbers that respect the complexity of care, not metrics that punish patients for their living rooms.