The audio-visual effects of rehabilitative landscapes on stress recovery in older adults: a case study of typical parks in Xi'an.

Integrating AI tools into clinical practice marks a paradigm shift in health care, with the potential for enhanced diagnostic accuracy, improved workflows, and personalized patient care. In fact, a 2019 survey of health professionals found that up to 71% of surveyed physicians opined that AI would improve and impact the field of medicine within the next decade []. The potential of AI tools in health care extends across various domains, including radiology [], pathology [], dermatology [], ophthalmology [], general practice [], surgery [], neurology [], cardiology [], and more. By leveraging vast amounts of medical data, AI algorithms can help identify patterns and correlations that may be more difficult for human clinicians to recognize due to the size, rarity, or complexity of certain clinical conditions. As a result, this can enable earlier and more accurate diagnoses and subsequent interventions. Machine learning models have shown promising task-specific performance in selected medical-imaging applications, although performance and clinical benefit vary substantially across tasks, datasets, and implementation settings []. In addition to aiding in clinical diagnostics, AI tools can enhance clinical decision-making []. AI-based decision support systems can synthesize plans from multimodal data and can continuously learn from new data, improving recommendations over time and adapting to the latest clinical guidelines and research findings []. The use of AI in clinical settings also addresses the growing concern of clinician burnout caused by an increase in workload in interpreting imaging examinations without a corresponding increase in manpower []. Yu et al [] further support the complexity of human-AI collaboration in diagnostic settings and indicate significant variability in how radiologists respond to AI assistance. They found that radiologists’ performance was easily influenced by AI errors in their large-scale study on AI-aided chest X-ray interpretation, but conventional factors such as experience or familiarity with AI did not predict the impact of AI support. There is also inherent variability in how radiologists use and interact with AI tools, which is influenced by personal biases, cognitive strategies, and the opacity of AI algorithms [-]. Integrating AI into clinical practice also raises concerns about data privacy [], algorithmic bias [], and the need for rigorous validation and regulation []. Ensuring that AI tools are reliable, explainable, and complement rather than replace clinical judgment is critical to ensuring their acceptance, since replacement is a significant fear in the field, for clinicians and patients receiving care alike [,]. Thus, a collaborative approach, where AI augments the clinician’s expertise instead of acting as a substitution, is essential for the successful adoption of these technologies.