##人工智能医疗应用测试混乱，专家呼吁加强监管和规范
**人工智能（AI）在医疗领域的应用正迅速发展，但其测试和评估却面临着巨大挑战。** 据《自然》杂志观点，目前医疗领域几乎没有自主的人工智能，测试方法混乱，缺乏标准化和规范，导致许多 AI 医疗算法在缺乏充分临床验证的情况下被批准使用，引发了安全和有效性方面的担忧。
**专家指出，AI 医疗应用的测试过程应该是一个复杂的多阶段过程，包括临床试验、安全评估和偏见评估等。** 然而，目前大多数 AI 医疗设备的测试数据不足，缺乏严格的审查和授权标准。例如，美国食品药品监督管理局（FDA）批准了数百种 AI 驱动的医疗设备，但其中许多设备并没有经过充分的临床验证。
**加利福尼亚州洛杉矶西达赛奈医疗中心的心脏病专家 David Ouyang 表示，一些医院选择自行测试 AI 设备，但测试方法缺乏统一标准，导致结果难以比较和评估。** 此外，AI 算法对不同人群的敏感性差异也需要引起重视，目前尚不清楚如何最好地向患者介绍这些技术，并征求他们同意使用其数据进行测试。
**专家呼吁加强对 AI 医疗应用的监管和规范，建立统一的测试标准和评估体系。** 同时，需要加大对 AI 医疗应用临床试验的投入，收集更多高质量的数据，并进行更深入的安全性、有效性和公平性评估。
**一些医院和医疗保健系统正在积极探索在医学中使用和评估 AI 系统的方法。** 随着越来越多的 AI 工具和公司进入市场，各组织正在努力达成共识，确定最有效、最严格的评估方法。
**专家认为，AI 医疗应用的未来发展需要政府、医疗机构、研究机构和企业共同努力，建立一个更加完善的监管和评估体系，确保 AI 技术在医疗领域的应用安全、有效和公平。** 只有这样，AI 才能真正发挥其在医疗领域的潜力，为患者带来更好的治疗效果。
英语如下：
##AI Medical Testing in Chaos: Limited Data, Accuracy Concerns
**Keywords:** AIhealthcare, chaotic testing, insufficient data
## AI Applications in Healthcare Face Testing Chaos, Experts Call for Enhanced Regulation and Standardization
**The application of artificial intelligence (AI) in healthcare is rapidly developing, but its testing and evaluation face significant challenges.** According to a perspective in *Nature*, there are virtually no autonomous AI systems in healthcare today. Testing methods are chaotic, lacking standardization and regulation, leading tothe approval of many AI medical algorithms for use without sufficient clinical validation, raising concerns about safety and efficacy.
**Experts emphasize that the testing process for AI healthcare applications should be a complex, multi-stage process, including clinical trials, safetyassessments, and bias evaluations.** However, most AI medical devices currently lack sufficient testing data and rigorous review and authorization standards. For example, the U.S. Food and Drug Administration (FDA) has approved hundreds of AI-powered medical devices, but many have not undergone sufficient clinical validation.
**David Ouyang, a cardiologist at Cedars-Sinai Medical Center in Los Angeles, California, notes that some hospitals choose to test AI devices themselves, but testing methods lack uniform standards, making results difficult to compare and evaluate.** Furthermore, the sensitivity differencesof AI algorithms across different populations need to be addressed. It remains unclear how best to introduce these technologies to patients and obtain their consent to use their data for testing.
**Experts call for enhanced regulation and standardization of AI healthcare applications, establishing unified testing standards and evaluation systems.** Simultaneously, increased investment in clinical trialsfor AI healthcare applications is necessary to collect more high-quality data and conduct deeper assessments of safety, efficacy, and fairness.
**Some hospitals and healthcare systems are actively exploring methods for using and evaluating AI systems in medicine.** As more AI tools and companies enter the market, organizations are striving to reach a consensus on themost effective and rigorous evaluation methods.
**Experts believe that the future development of AI healthcare applications requires joint efforts from governments, healthcare institutions, research institutions, and businesses to establish a more comprehensive regulatory and evaluation system, ensuring the safe, effective, and equitable application of AI technology in healthcare.** Only then can AI truly unleashits potential in healthcare, delivering better treatment outcomes for patients.
【来源】https://www.jiqizhixin.com/articles/2024-08-22-2
Views: 2