News & Views

Healthcare communication faces unprecedented challenges as the healthcare workforce is increasingly faced with increased administrative burdens and reduced time with patients. Conversational agents powered by generative AI may offer a potential solution by collecting information, answering questions, documenting encounters, and supporting clinical decision-making through fluid, contextual dialogue. However, realizing their potential requires rigorous validation, careful implementation, and a strong commitment to safety, equity, and preserving human-centered care.

The integration of physics-based digital twins with data-driven artificial intelligence—termed “Big AI”—can advance truly personalised medicine. While digital twins offer individual ‘healthcasts,’ accuracy and interpretability, and AI delivers speed and flexibility, each has limitations. Big AI combines their strengths, enabling faster, more reliable and individualised predictions, with applications from diagnostics to drug discovery. Above all, Big AI restores mechanistic insights to AI and complies with the scientific method.

Artificial intelligence (AI) is already having a significant impact on healthcare. For example, AI-guided imaging can improve the diagnosis/treatment of vascular diseases, which affect over 200 million people globally. Recently, Chiu and colleagues (2024) developed an AI algorithm that supports nurses with no ultrasound training in diagnosing abdominal aortic aneurysms (AAA) with similar accuracy as ultrasound-trained physicians. This technology can therefore improve AAA screening; however, achieving clinical impact with new AI technologies requires careful consideration of commercialization strategies, including funding, compliance with safety and regulatory frameworks, health technology assessment, regulatory approval, reimbursement, and clinical guideline integration.

Cognitive bias accounts for a significant portion of preventable errors in healthcare, contributing to significant patient morbidity and mortality each year. As large language models (LLMs) are introduced into healthcare and clinical decision-making, these systems are at risk of inheriting – and even amplifying – these existing biases. This article explores both the cognitive biases impacting LLM-assisted medicine and the countervailing strengths these technologies bring to addressing these limitations.

Neurological conditions, including dementia, pose a major public health challenge, contributing to a significant and growing clinical, economic, and societal burden. Traditionally, research and clinical practice have focused on diseases like dementia in isolation. However, in an ageing, multimorbid population, this approach is becoming increasingly inadequate. Recognising brain health as a lifelong attribute influenced by various health determinants, this paper explores the concept of brain health, identifies key challenges in assessing it effectively, and examines how digital biomarkers could provide a versatile measurement framework to enhance monitoring and facilitate earlier intervention. Finally, we outline future directions to help advance definitions of meaningful aspects of brain health integration, and practical adoption of digital biomarkers, enhancing our capacity to measure and preserve ‘brain health capital’ or ‘brain span’ across the lifecourse.

Wearable medical devices are becoming increasingly common in everyday life, and thus is the reliance on them, the data they generate, and the resulting treatment plans. However, recent events in the supply chains of other devices have shown the catastrophic outcomes manipulation of them can have. In this article we showcase the importance of supply chain cybersecurity for medical devices and describe measures that can mitigate these risks.

Despite significant advances, the prevention and management of cardiovascular disease remain challenging, especially for ischemic heart disease (IHD). Current clinical decision-making relies heavily on physician expertise, guideline-directed therapies, and static risk scores, which often inadequately accommodate individual patient complexity. Machine learning (ML) and artificial intelligence (AI), particularly reinforcement learning (RL), may augment current physician-driven approaches and provide enhanced cardiovascular disease prevention and management. Indeed, offline RL refers to a class of ML algorithms that learn optimal decision-making policies from a fixed dataset of previously collected experiences—such as electronic health records or registries—without the need for active, real-time interaction with the clinical environment. This approach enables the safe development of treatment strategies in high-stakes domains where experimentation on live patients could be unethical or impractical. Notably, offline RL models hold the promise of optimizing decision-making in complex clinical settings, such as revascularization strategies for coronary artery disease. However, challenges remain in integrating AI into practice, ensuring interpretability, maintaining performance, and proving cost-effectiveness. Ultimately, validation, integration, and collaboration among clinicians, researchers, and policymakers are crucial for transforming AI-driven solutions into practical, patient-centered cardiovascular care improvements, pending prospective (and hopefully randomized) validation.

The rapid increase in real-time health information collected from wearable devices has allowed digital biomarkers to emerge as a promising tool to support perioperative care, including surgical prehabilitation, intra-operative guidance, and post-operative monitoring. Important challenges include the accuracy of generated information, data security risks, and slow adoption of new technologies. Active stakeholder engagement and following existing digital biomarker development/implementation frameworks may support using this technology to improve surgical outcomes.

Recently, a genomic language model (gLM) with 40 billion parameters known as Evo2 has reached the same scale as the most powerful text large language models (LLMs). gLMs have been emerging as powerful tools to decode DNA sequences over the last five years. This article examines the emergence of gLMs and highlights Evo2 as a milestone in genomic language modeling, assessing both the scientific promise of gLMs and the practical challenges facing their implementation in medicine.

Wearable artificial intelligence (AI) technologies show promise in healthcare, with early applications demonstrating diverse benefits for patient safety. These systems go beyond traditional data collection, using advanced algorithms to provide real-time clinical guidance. From infectious disease monitoring to AI-powered surgical assistance, these technologies enable proactive, personalized care while addressing critical safety gaps. However, successful implementation requires careful consideration of technical, operational, and ethical challenges.

We live in interesting regulatory times. In January, a bill was introduced to the US Congress proposing that AI “can qualify as a practitioner eligible to prescribe drugs” if overseen by the States and FDA. This a bold and contentious move. Even proponents of AI’s swift integration into medicine must recognize the deep paradox: this proposal emerges even as the FDA’s world-leading infrastructure for AI oversight faces dismantling.

Generalist AI systems in healthcare can handle multiple complex clinical tasks, unlike narrow AI tools that perform isolated functions. However, current payment systems struggle to capture the value of these integrated capabilities. We examine potential solutions, including value-based and tiered structures, balancing innovation, equitable access, continuous performance evaluation, and cost-effectiveness to realize generalist AI’s transformative potential.

We place ‘Model Cards’ and graphical ‘nutrition labels’ for health AI in context with the information needs of patients, health care providers and deployers. We discuss the applicability of Model Cards for General Purpose AI (GPAI) models. If these approaches are to be useful and safe they need to be integrated with regulatory approaches and linked to deeper layers of open and detailed model information and optimized through user testing.

Enabled by the rapid rise in data collected by technologies, Digital Biomarkers (DBx) have emerged as a novel mechanism for assessment, diagnosis, and monitoring. However, the exponential growth and ability to generate new data has also raised questions about ways of ensuring the authenticity and accuracy of digital data. A recent study highlights how Large Language Models (LLMs) generating human-like content amplify these risks, and propose watermarking as a scalable solution to ensure data integrity. This article examines the potential of digital watermarking to help safeguard the reliability and provenance of DBx data, whilst also addressing broader challenges in health systems.

Large language models (LLMs) are increasingly applied in medical documentation and have been proposed for clinical decision support. We argue that the future for LLMs in medicine must be based on transparent and controllable open-source models. Openness enables medical tool developers to control the safety and quality of underlying AI models, while also allowing healthcare professionals to hold these models accountable. For these reasons, the future is open.

John J. Hopfield and Geoffrey E. Hinton were awarded the 2024 Nobel Prize in Physics for developing machine learning technology using artificial neural networks. In Chemistry it was awarded to Demis Hassabis and John M. Jumper for developing an AI algorithm that solved the 50-year protein structure prediction challenge. This highlights AI’s impact on science, medicine and society; however, the winners acknowledge ethical aspects of AI that must be considered.

The integration of artificial intelligence (AI) in sports medicine is opening new frontiers for athlete health and performance, aligning with the spirit of the Paris 2024 Olympic Games slogan, “Games Wide Open.”

Traditional healthcare delivery models face mounting pressure from rising costs, increasing demand, and a growing environmental footprint. Hospital at Home (HaH) has been proposed as a potential solution, offering care at home through in-person, virtual, or hybrid approaches. Despite focus on expanding HaH provision and capacity, research has primarily explored patient care outcomes, patient satisfaction economic costs with a key gap in its environmental impact. By reducing this evidence gap, HaH may be better placed as a positive enabler in delivering healthier planet and population. This article explores the environmental opportunities and challenges associated with HaH compared to traditional hospital care and reinforces the case for further research to comprehensively quantify the environmental impact including any co-benefits. Our aim for this article is to spark conversation, and begin to help prioritise future research and analysis.

Healthcare AI faces an ethical dilemma between selective and equitable deployment, exacerbated by flawed performance metrics. These metrics inadequately capture real-world complexities and biases, leading to premature assertions of effectiveness. Improved evaluation practices, including continuous monitoring and silent evaluation periods, are crucial. To address these fundamental shortcomings, a paradigm shift in AI assessment is needed, prioritizing actual patient outcomes over conventional benchmarking.

A highly ambitious FDA initiative will explore, through a hub and ideas lab, how equitable healthcare at home can be delivered, recognizing that this is unlikely to come about without intervention. Market forces, as shaped by current regulations, are leading to digital health tools developed and operating in islands rather than enabling integrated digital care. Can the initiative, which adopts system-level regulatory thinking, solve this issue?