The answer, the review finds, is both promising and sobering. The researchers identified three core dimensions along which LLM-driven health systems personalize their behavior, eight distinct technical architectures that developers use to adapt these models to individual users, and four categories of implementation challenges that consistently undermine effectiveness. That structure matters because personalization in health is not a single feature you can bolt onto a chatbot. It is a design problem that spans what the system knows about you, how it reasons about that knowledge, and how it changes its outputs over time as your needs evolve.

To understand why this review lands at such a pivotal moment, consider what makes large language models different from the health apps that came before them. Earlier generations of digital health tools relied on rigid rule engines and decision trees: if a user logged three workouts in a week, the app sent a congratulatory notification; if a diabetic patient’s glucose reading crossed a threshold, an alert fired. These systems were transparent but brittle. They could not hold a conversation, could not interpret the messy, ambiguous way people actually describe their lives, and could not adjust their tone or content to a user’s literacy level, culture, or emotional state. LLMs, by contrast, are trained on vast corpora of text and can generate fluent, contextually sensitive responses to essentially any prompt, which makes them natural candidates for conversational health coaching, mental wellness support, medication guidance, and lifestyle intervention.

But fluency is not the same as personalization. A model that answers every question beautifully but identically for a teenager in Nairobi, a retiree in Halifax, and a shift nurse in Mumbai is not adaptive; it is merely articulate. The Dalhousie team’s synthesis of the literature shows that researchers have attacked this problem from multiple technical directions. Among the eight adaptation architectures catalogued in the review are approaches such as fine-tuning, where a base model is retrained on domain-specific health data to sharpen its medical reasoning; retrieval-augmented generation, where the model consults external knowledge sources or a user’s personal records before responding; prompt engineering and in-context learning, where user profiles, preferences, and history are packed into the prompt itself so the model conditions its answers on them; and agent-based designs, where multiple specialized model components collaborate, with one handling medical accuracy, another managing conversational tone, and a third tracking long-term user goals.

Each architecture carries distinct trade-offs that the review helps clarify for the field. Fine-tuning can produce deep domain competence but requires curated datasets, computational resources, and careful validation to avoid degrading a model’s general capabilities or introducing subtle biases. Retrieval-augmented approaches keep personal data outside the model weights, which eases privacy concerns and allows information to be updated in real time, but they depend on the quality and currency of the underlying data stores. Prompt-based personalization is cheap and flexible, yet it consumes context window space and can be fragile when a user’s profile grows complex. Agentic pipelines offer modularity and auditability but add engineering complexity and latency. The review’s contribution is to map these options against the health domains where they have been deployed, giving designers a conceptual framework for choosing the right tool for the right intervention.

The health domains covered in the synthesized literature are strikingly broad. LLM-driven systems have been explored for mental health support, where conversational agents can offer around-the-clock availability and nonjudgmental interaction that some users find easier than talking to a human; for chronic disease management, where models can interpret self-reported symptoms and medication logs; for physical activity and nutrition coaching, where adaptive feedback loops respond to a user’s progress; and for health information access, where models translate clinical jargon into plain language tailored to a reader’s background. Across these domains, the review highlights evidence that LLM-driven applications can be effective, while emphasizing that their success hinges on how well they meet individual needs and preferences rather than on raw model capability alone.

Here is where the review delivers its most uncomfortable finding: inclusivity and contextual sensitivity remain underexplored across the entire field. Most systems studied to date are built and evaluated with narrow populations in mind, often English-speaking, digitally literate, and from high-income settings. A personalization engine that models a user’s preferences but ignores their cultural context, language, disability status, or socioeconomic reality risks producing interventions that work beautifully in a pilot study and fail, or even cause harm, in the real world. The authors frame this as a central gap for the human-computer interaction community, arguing that future systems must be designed to be adaptive, personalized, inclusive, effective, and responsive to evolving individual characteristics and needs, a five-part standard that raises the bar well beyond current practice.

The four categories of implementation challenges identified in the review compound that concern. Technical hurdles include the tendency of LLMs to generate plausible but incorrect health information, a failure mode that is dangerous when the subject is medication dosing or symptom triage. Data challenges revolve around the scarcity of high-quality, representative datasets for training and evaluation, particularly for underrepresented groups. Ethical and privacy challenges loom largest for deployment: health conversations are among the most sensitive data a person can share, and questions about consent, data retention, and secondary use remain unsettled in most jurisdictions. Finally, evaluation challenges persist because there is no consensus on how to measure personalization quality itself; a system can score well on generic benchmarks while failing catastrophically at adapting to any single real user.

The ethical dimension receives particular attention in the review, which situates technical choices inside questions of fairness, autonomy, and global impact. If LLM health assistants become a mainstream interface to care, the populations excluded from their training data could face systematically worse automated advice, deepening existing health disparities rather than closing them. Conversely, the authors point to genuine opportunities for global reach: conversational agents that speak low-resource languages, that operate on inexpensive hardware, and that deliver evidence-informed guidance to communities with limited access to clinicians could extend the frontier of wellness support to billions of people. Whether that promise materializes depends on the design decisions being made in laboratories right now, which is precisely why a structured map of the field arrives at such a useful moment.

The Dalhousie team’s conceptual framework, built from the three personalization dimensions, eight adaptation architectures, and four challenge categories, is offered as a shared vocabulary for the next generation of research. Rather than each lab reinventing its own taxonomy of what personalization means, future studies can locate their contributions within a common structure, compare results across domains, and identify which combinations of architecture and dimension remain untested. The review’s dataset, including all coding records from the 149 articles, is available in supplementary material, and the article is open access, lowering the barrier for researchers in resource-constrained settings to build on it. The work was published on 30 September 2026 in Artificial Intelligence Review, with no conflicts of interest declared by the authors.

What emerges from this synthesis is a field in transition. The first wave of LLM health applications proved that the technology could hold convincing, helpful conversations about well-being. The second wave, which this review both documents and aims to accelerate, must prove something harder: that these systems can know who they are talking to, respect who that person is, and change what they say accordingly, safely, and fairly. The 149 studies analyzed here suggest the ingredients are in place, from mature adaptation architectures to growing awareness of ethical pitfalls. What remains is the disciplined, inclusive engineering work of turning a chatty generalist into a genuinely personal health companion, one carefully validated system at a time.

Subject of Research: Systematic review of large language model-driven personalized and adaptive systems for health and wellness

Article Title: LLM-driven personalized and adaptive systems for health and wellness: a systematic review

Article References: Kimeu, J. M., Chan, G., Orji, R., & Oyebode, O. (2026). LLM-driven personalized and adaptive systems for health and wellness: a systematic review. Artificial Intelligence Review. https://doi.org/10.1007/s10462-026-11703-6

Image Credits: AI Generated

DOI: 10.1007/s10462-026-11703-6

Keywords: large language models, personalization, adaptive systems, digital health, wellness, human-computer interaction, systematic review, health informatics, artificial intelligence, machine learning, e-health, conceptual framework

The scale of the shortage is not in dispute. Workforce surveys from the American College of Radiology and the Radiology Business Management Association have documented hiring pressures across the specialty, and pediatric radiology has been singled out as particularly vulnerable, with fewer trainees choosing the subspecialty even as children’s hospitals expand their imaging services. A previous analysis in the Journal of the American College of Radiology laid out action steps to address the pediatric radiologist deficit, warning that without intervention the pipeline of qualified imagers would narrow dangerously. Remote and hybrid work emerged as an unexpected lever. By removing geographic constraints, departments can recruit radiologists who live elsewhere, retain experienced staff facing caregiving responsibilities, and compete for talent against private practices and teleradiology companies offering location independence as a core benefit.

Some of the strongest evidence for the viability of remote radiology came from the COVID-19 pandemic, which functioned as an unplanned, worldwide stress test of distributed imaging practice. Surveys conducted by the Society for Pediatric Radiology and the Society of Chiefs of Radiology at Children’s Hospitals documented how rapidly pediatric imaging departments shifted to home-based picture archiving and communication system, or PACS, workstations during the crisis. Far from collapsing, diagnostic operations continued, and a follow-up study of pediatric neuroradiologists at a quaternary pediatric academic hospital found that subspecialists working from home during the pandemic maintained solid interpretive performance. That finding matters because it speaks to a central technical question: whether the cognitive work of pediatric image interpretation, which demands meticulous attention to subtle findings across growing, complex study volumes, can be performed reliably outside the walls of the hospital. The accumulating answer, at least for many read-oriented tasks, has been yes.

Broader research beyond radiology reinforces the point. A randomized controlled trial published in Nature in 2024 found that hybrid working from home improved employee retention without damaging performance, upending assumptions that physical presence is a prerequisite for high-quality output. Studies of work-from-anywhere arrangements have documented productivity gains when geographic flexibility is paired with clear accountability, while analyses of fully remote arrangements show more mixed results, particularly for tasks dependent on dense collaboration. Radiology-specific investigations echo this pattern. A study comparing remote and on-site radiologist productivity found comparable output between the two environments, and an academic radiology review concluded that work-from-home programs, when thoughtfully structured, offer real advantages in recruitment and well-being without measurable penalty to clinical throughput. Surveys within radiology departments also report that faculty and trainees largely perceive remote workstation technology as adequate for clinical work, even as they flag concerns about the learning experience it provides.

Those concerns point to the darker side of the distributed model. Organizational psychologists have long studied the power of proximity: coworkers who are physically present benefit from spontaneous interactions, serendipitous consultation, and the passive face time that shapes how colleagues and leaders perceive their commitment. Research on spontaneous trait inference shows that employees who are simply seen more often are unconsciously credited with positive qualities, a phenomenon that translates directly into proximity bias in hybrid workplaces. In radiology departments split between on-site and remote staff, this bias can generate real or perceived inequities in assignment of desirable work, access to leadership opportunities, and credit for team contributions. The Seattle authors identify this interpersonal friction as one of the principal hazards of hybrid practice, noting that perceived unfairness between the two cohorts can corrode morale in ways that no productivity metric will capture until valuable radiologists begin to leave.

Visibility poses a second, arguably existential, challenge. Radiology’s professional literature has warned for years about the vanishing radiologist, the clinician whose expertise becomes invisible to patients, referring physicians, and hospital administrators alike. The danger of being unseen has acquired new urgency as proponents of artificial intelligence argue that diagnostic algorithms could absorb radiologists’ interpretive work. If radiologists are also physically absent from the decision-making arenas where clinical priorities are set, protocols designed, and resources allocated, their institutional influence diminishes precisely when the specialty needs champions most. The authors argue that off-site radiologists must remain deliberately present in these spheres, both within radiology and across the broader hospital, because the credibility that protects the profession from being reduced to a commodity image-reading service is built through relationships, committee work, and face-to-face clinical engagement.

The third major casualty of remote practice is the informal curriculum of radiology training. Radiology education has traditionally relied on the live readout, in which a trainee sits beside an attending radiologist and watches cases unfold in real time, absorbing search patterns, diagnostic reasoning, and the defensive habits of safe practice. When faculty work from home, those side-by-side sessions shrink, and with them the mentorship conversations that occur in hallways and over lunch. A systematic assessment of faculty and trainee perceptions of the current state of radiology readouts documents tension over how teaching is delivered in the modern department. The Seattle group warns that degraded informal mentorship and reduced trainee contact could narrow the pipeline of future pediatric radiologists, compounding the very workforce shortage that remote work was meant to relieve. A specialty that recruits flexibly but fails to train the next generation solves one crisis by manufacturing another.

The mitigation strategies the authors describe are refreshingly concrete, and fairness sits at the top of the list. In their practice, hybrid eligibility, work distribution, and coverage obligations are structured so that on-site and remote radiologists feel the rules apply equally to everyone, rather than according to seniority, personality, or chance. Fairness also extends to the metrics used to evaluate performance. The authors describe deliberately avoiding siloing, in which remote radiologists are permanently relegated to isolated reading queues, and resisting an overemphasis on individual work relative value units, the wRVU benchmarks that quantify billed interpretive volume. Heavy wRVU-driven accounting, they argue, reduces a radiologist’s contribution to a single number and discourages the unbillable but essential activities, such as teaching, protocol development, multidisciplinary conferences, and peer review, that make an imaging department function. Departments that measure only throughput will get throughput, and lose everything else.

Communication emerges as the connective tissue holding the hybrid department together, and the authors advocate for it to be frequent and multidirectional rather than occasional and top-down. Structured check-ins, recurring video conferences, intentional inclusion of remote staff in departmental meetings, and deliberate rotation of on-site responsibilities all help distributed teams maintain the informal awareness that co-located teams get for free. Research on designing hybrid offices supports this engineering mindset: the most successful hybrid organizations treat collaboration as something to be actively designed, not left to chance encounters. The same principle applies to quality improvement. Peer feedback and learning, identified by radiology safety researchers as central to reducing diagnostic error, depend on cultures of open communication that must be consciously maintained when colleagues rarely share a physical room.

What emerges from the review is neither a celebration of remote work nor a call to retreat to the reading room, but a pragmatic framework for a specialty with little room to maneuver. Pediatric radiology cannot afford to forfeit the recruitment and retention advantages of flexibility, and the evidence suggests remote practice can deliver clinical quality. Yet the intangible assets of an academic imaging department, its visibility, its mentorship, its collective culture, are depleted by distance in ways that are easy to ignore and costly to rebuild. The Seattle authors’ prescription is essentially a bargain: hybrid work for pediatric radiologists, purchased with deliberate investments in fairness, communication, and presence in the rooms where decisions are made. Whether children’s hospitals strike that bargain well may determine not just how comfortable their radiologists are, but whether the subspecialty has a workforce a generation from now.

Subject of Research: Strategies for maintaining a healthy work environment among hybrid and remote pediatric radiologists facing a workforce shortage

Article Title: Maintaining a healthy work environment with hybrid and remote pediatric radiologists

Article References: Noda, S., Oztek, M. A., Ward, J., Pai, V., & Iyer, R. S. (2026). Maintaining a healthy work environment with hybrid and remote pediatric radiologists. Pediatric Radiology. https://doi.org/10.1007/s00247-026-06782-0

Image Credits: AI Generated

DOI: 10.1007/s00247-026-06782-0

Keywords: pediatric radiology, hybrid work, remote work, workforce shortage, radiologist retention, mentorship, trainee education, proximity bias, wRVU, teleworking, burnout, wellness