代理人工智能正在医疗保健领域创造一场地震式的转变

creating a seismicshift in healthcare Contents The next wave ofhealthcare transformation The next wave ofhealthcare transformation change is underway. The emergence of agentic AI marks a new era of business transformation, where AI autonomy andinnovation converge to redefine how people and technology interact. discuss its risks and limitations, and how you can safely make a start on your agentic AI journey. Agentic AI is a system consisting of multiple software entities (or agents) operating as building blocks or “cogs in the machine”. These agents are capable of handling specific tasks independently. They collaborate with and learnfrom each other to make decisions and take actions in a broader context. •Plan tasks and break down goals into actionable steps •Access internal and external tools autonomously •Utilise knowledge•Use different models and capabilities to complete tasks How can agentic AItransform healthcare? How can agentic AItransform healthcare? healthcare we need AI systems that understand and can adapt to dynamic and complexenvironments, and make accurate decisions in real time. This transformative technology is reshaping industries, redefining market dynamics and heralding a future where machines not only assist humans but also autonomously enable innovation and greater productivity.Agentic AI is the framework that allows autonomous software entities (agents) to collaborate with other agents and with humans on an enterprise-wide scale. Within this framework, AI agents can specialise incertain tasks and have access to controlled knowledge bases for increased accuracy and autonomy. Thismakes agentic AI an incredibly powerful tool for solving complex healthcare challenges. that impact is primarily in two areas: Improving patient outcomes Agentic AI offers numerous benefits to both patients and staff, and can contribute significantly to staff Reducing the administrative burden Agentic AI can reduce the amount of time healthcare staff need to spend onadministrative tasks, allowing a larger percentage of their time to be spent withpatients and providing primary care. taken for patients to be transferred from an ambulance stretcher and intoa hospital emergency department (known as “patient off stretcher time”, or Care, an initiative launched in August 2023,1working with frontline clinicians inNSW Health to identify and reduce the impact of administrative responsibilitiesso that clinicians could instead spend more time on delivering safe, reliable, healthcare, there are few, if any, resources more precious and closely managedthan time. As many healthcare providers worldwide will tell you, there is simplynot enough time to care for all their patients with quality and compassionand, simultaneously, complete mandatory tasks, such as record reviews,documentation and insurance paperwork. Globally, the source of this problem Improving patient outcomes By analysing a patient’s data, medical history andcurrent symptoms, AI agents in healthcare canrecommend a course of action to the patient’sdoctor, who still has the final say. This helps healthcare notes: “Beyond diagnostics, AI agentsadvance predictive analytics and personalizedcare by analyzing EHRs [electronic health records],imaging, genomics and wearable device data.They identify subtle disease indicators, such asearly signs of diabetes, cardiovascular conditions, AI has to be safe approaches to AI. These include:•The European Union (EU) taking a risk-based approach to comprehensive AI regulation succinctly in the Australian Government’s legislative review on AI in healthcare: •The UK setting up an AI Security Institute to enable governance and ensure advanced of AI in health and care settings to ensure thatall Australians realise the potential benefits.”4 •Canada working towards a more structured regulatory framework with its proposed In a report commissioned by NSW Health,CSIRO identified augmented care and AI asone of the six megatrends shaping the future Effectively, healthcare organisationsneed to trust AI. Central to this trust is the explainability andtransparency of AI systems. “If a patient or a cliniciandoesn’t trust what a system is showing them, if itfeels like a black box, then the promise of efficiencyof healthcare.However, CSIRO emphasised that ensuring AI is safe,trustworthy, effective and responsible is a critical requirement of current and future applications inhealthcare: “With only 44% of Australians feeling thatthe benefits of AI will outweigh the risks, AI applicationsin healthcare need to be implemented in a way thatbuilds trust and acceptance among health consumers, constrained industries like healthcare. In Australia,pressure on healthcare systems due to issues suchas aged-care bed availability, hospital rampingand staffing shortages can be alleviated by agentic cautions: “The current developmental stage of agenticAI also brings challenges, particularly its ‘blackbox’nature and unpredictabl