AI医疗应用挑战与进展-优快云博客

本文链接：https://blog.youkuaiyun.com/linkequa/article/details/87169249

本文探讨了AI在医疗领域的应用面临的五大挑战，包括数据共享与隐私保护、算法透明度、数据标准化、跨平台互操作性和患者关怀。同时，文章介绍了AI领域的关键术语，并列举了FDA已批准的几款医疗AI产品，如Arterys、IDx-DR和GuardianConnect，这些产品在肺部CT扫描、糖尿病视网膜病变检测及血糖监测方面发挥了重要作用。

摘要生成于 C知道，由 DeepSeek-R1 满血版支持，前往体验 >

1, 将AI（Artificial Intelligence）运用到医疗上的问题：

1.1 Data sharing and privacy
目前个人健康数据的维度包括：demographics, healthcare provider notes, images, laboratory results, genetic testing data, recordings from medical devices or wearable sensors.

数据在共享过程中要去标签化，要在知情同意书中写清楚数据可能会有哪些用途。

1.2 Transparency of algorithms
监督学习模型需要数据透明来提高准确性；医生需要算法透明来保证诊断结果的准确性，这可能会与商业机构矛盾，商业机构可能会对算法的应用申请专利从而来保证利益。

1.3 Data standardization
将数据标准化，使得能存成计算机可以识别的格式，这是整个分析流程最关键的步骤之一。在前期数据收集步骤下功夫是最好的方法。

1.4 Interoperability across multiple platform
前期制定标准，如何整合不同算法的结果，如何保证算法在不同平台的可移植性。

1.5 Concern for patients
第一是AI诊断系统本身的准确性，程序是否会出现bug等等；第二是出问题后谁来承担责任，医生，仪器制造商，仪器提供商？

2，AI中常用术语解释：

• Artificial intelligence: A branch of applied computer science wherein computer algorithms are trained to perform tasks typically associated with human intelligence.
• Machine learning: Providing knowledge to computers through data without explicit programming. Attempts to optimize a ‘mapping’ between inputs and outputs.
• Representation learning: Learning effective representations of a data source algorithmically, as opposed to hand-crafting combinations of data features.
• Deep learning: Multiple processing layers are used to learn representations of data with multiple layers of abstraction.
• Supervised learning: Training is conducted with specific labels or annotations.
• Unsupervised learning: Training is conducted without any specific labels, and the algorithm clusters data to reveal underlying patterns.
• Natural language processing: The organization of unstruc- tured narrative text into a structured form that can be inter- preted by a machine and allows for automated information extraction.

3，FDA批准的医疗AI产品：

• Arterys: Aids in finding lesions within pulmonary computed tomography (CT) scans and liver CT and magnetic reso- nance imaging (MRI) scans using AI to segment lesions and nodules. This is the first FDA-approved deep learning clinical platform.
• IDx-DR:Provides automatic detection of more than mild dia- betic retinopathy in adults 22 years of age or older diagnosed with diabetes who have not been previously diagnosed with diabetic retinopathy. This is meant to be used in primary care settings with subsequent referral to an eye specialist if indicated and is the first autonomous AI diagnostic system, as no clinician interpretation is needed before a screening result is generated.
• Guardian Connect (Medtronic):Continuously monitors glucose and sends collected data to a smartphone app. This uses International Business Machines (IBM) Watson tech- nology to predict major fluctuations in blood glucose levels 10–60 minutes in advance.