JMIR AI

No preview text available.

In the ongoing AI boom—now supercharged by large language models and generative AI tools—the rapid integration of AI into health care and public health has created an urgent need for professionals who combine deep domain expertise (clinical, epidemiological, ethical, and contextual expertise in health and public health) with AI fluency (the ability to understand, critically evaluate, use, and oversee AI systems effectively). Without domain guidance, AI risks producing biased, unsafe, or contextually irrelevant outputs that compromise patient safety and exacerbate disparities. Conversely, domain experts lacking AI fluency may underuse or misuse these technologies, missing opportunities to improve care quality, operational efficiency, and population health. A foundational understanding of the modern AI system stack—comprising energy, accelerators, infrastructure, models, and applications—empowers health professionals to make informed decisions on adoption, evaluation, governance, and ethical use without needing to code or build models themselves. This stack framework highlights critical interdependencies and health care–specific challenges, including regulatory compliance, data privacy, sustainability, and resource constraints, particularly in low- and middle-income settings. This paper argues that cultivating domain-AI bilingualism is essential for safe, effective, and equitable AI integration worldwide. It outlines the 5-layer AI system stack applied to health care, key reasons for hybrid expertise, and a practical 4-step pathway: awakening (recognizing AI’s transformative role), intrepidity (courageously engaging foundational AI concepts), bilingualism (translating between clinical needs and AI capabilities), and visionary leadership (driving organizational and policy readiness). It outlines major barriers, targeted strategies, and recommendations for educators, institutions, and policymakers. By bridging domain knowledge with AI fluency, health professionals can direct—rather than be directed by—AI technologies, including emerging agentic systems, toward more precise, sustainable, and human-centered outcomes while mitigating risks. This hybrid capability represents a nonnegotiable competency for the health workforce for thriving in the post-2026 AI era.

While AI has existed for decades, the introduction of generative AI tools and subsequent advancements in their capability and public use resulted in a shift in public awareness and perceptions of AI. In response, the research community has grown increasingly interested in understanding public perceptions of AI use in various contexts, including health care. Public perceptions shape AI acceptance and consequently integration into health care. It is thus crucial to understand public trust in AI and perceptions of risk as we consider its implementation moving forward.

Clinical practice guidelines (CPGs) translate research evidence into actionable recommendations, but their development faces mounting challenges. With dozens of trials and systematic reviews published daily, the traditional 18- to 24-month development cycle struggles to incorporate emerging evidence, and manual screening, synthesis, and quality assessment consume thousands of person-hours per guideline. The resulting knowledge translation pipeline is slow, inconsistent, and quickly outdated.

Hypothyroidism is a prevalent endocrine disorder characterized by an insidious onset and nonspecific symptoms, often leading to significant diagnostic delays. While biochemical thyroid function tests (TFTs) are the gold standard, their invasive nature and cost make them impractical for continuous, proactive screening of at-risk populations. Thyroid hormones exert direct chronotropic effects on the cardiovascular system, and their levels are closely correlated with heart rate (HR). Wearable devices capable of continuous HR monitoring offer a novel opportunity for noninvasive, digital phenotype–based detection.

Large language model (LLM) benchmarking in clinical medicine is increasingly reliant on LLM-based assessment frameworks; in these systems, one AI scores the outputs of another. This practice relies on the assumptions of evaluator neutrality and rubric consistency. These assumptions have not been empirically tested in clinical fields.

Clinical AI operates in a health care environment heterogeneous in biology, clinical practice, and documentation, so its benefit-harm balance is deployment-specific: an intervention can help in one setting and harm in another when the determinant of response is not routinely observed.

Large language models (LLMs) show promise for mental health counseling applications. However, their deployment faces significant challenges, including data privacy concerns requiring on-premise solutions, high operational costs, and Western-centric biases that limit cultural appropriateness for non–English-speaking populations.

The increasing availability of machine-readable research data has created a growing need for efficient large-scale data harmonization (DH). Although large language models (LLMs) show promise for reducing the time and labor required for DH, their effective integration into harmonization workflows remains an important methodological challenge.

Mentalization is a core human capacity involving interpretation of one’s own and others’ behavior in terms of underlying mental states. Within the mentalization-based treatment (MBT) framework, it is described along multiple dimensions, integrating cognitive, affective, relational, and regulatory processes. Large language models (LLMs) can generate linguistically reflective discourse, raising the question of whether the formal linguistic structure of mentalization can be reproduced without experiential and affective processes.

Depression is a common mental disorder that seriously affects mental health and quality of life. Traditional depression diagnosis mainly relies on clinical interviews and subjective assessments, which may lead to variability in diagnostic outcomes. Electroencephalogram (EEG)-based computer-aided diagnosis has emerged as a promising, objective approach for depression recognition. However, effectively modeling the complex nonlinear characteristics of EEG signals remains a major challenge for existing deep learning methods.

Identifying individuals on parole or probation can help inform care pathways and ensure continuity of care in health care systems. However, this sensitive justice-related information is often absent from structured electronic health record fields and is instead documented in narrative clinical notes. Developing automated parole and probation detection (PPD) systems typically requires expert-annotated clinical data, but manual annotation is costly, time-consuming, and constrained by data sparsity and privacy concerns.