AI – PI-Union Medical Science Ltd.

ChatGPT Health and the Future of AI Healthcare: The Risks and Opportunities of Sharing Electronic Health Records with AI

PI-Union Medical Science — Fri, 12 Jun 2026 03:30:36 +0000

JAMA Viewpoint Commentary: When Patients Share Their Entire Medical Records with AI—The Opportunities and Risks of ChatGPT Health

From the Democratization of Medical Knowledge to the Democratization of Medical Data

As large language models (LLMs) rapidly enter the healthcare sector, artificial intelligence is evolving beyond providing general health information and is beginning to interact directly with patients’ most sensitive personal health data.

A recent Viewpoint article in JAMA, “When Patients Share Everything With an AI Chatbot: Risks and Opportunities of Large Language Models,” examines the opportunities and challenges posed by a new generation of health-focused AI platforms such as ChatGPT Health. The authors argue that as patients gain the ability to synchronize their complete electronic health records (EHRs) with AI systems, healthcare is moving from the democratization of medical knowledge toward the democratization of medical data.

While this transformation has the potential to advance personalized medicine, it also raises unprecedented concerns regarding privacy, bias, and regulatory oversight.

What Is the Potential Value of AI Access to Complete Medical Records?

In theory, if AI systems can integrate patients’ medical histories, health monitoring data, wearable device information, and lifestyle records, they may provide several important benefits:

*More personalized health recommendations

*Earlier identification of rare diseases

*Enhanced epidemic and public health surveillance

*Greater utilization of Real-World Data (RWD)

*Accelerated clinical research and drug development

For researchers, large-scale and real-time integration of health data may become a critical foundation for the future of precision medicine.

Medical Records Are Not Entirely Objective

However, the authors highlight an important reality:

Electronic health records are not neutral repositories of facts.

In addition to objective laboratory and diagnostic findings, medical records often contain subjective assessments and interpretations made by healthcare professionals.

Examples include:

1. Behavioral descriptions of patients

2. Preliminary diagnostic impressions

3. Unconfirmed clinical assumptions

4. Potentially biased language or documentation

If AI systems learn directly from these records, they may reproduce—or even amplify—existing biases.

For example, a patient suffering from chronic pain may have previously been labeled as “drug-seeking.” Even if subsequent evaluations confirm a legitimate physiological cause for the pain, an AI system may still be influenced by earlier documentation and provide less appropriate recommendations.

In other words, AI systems may learn not only medical knowledge but also the biases embedded within healthcare systems.

AI May Reinforce Existing Health Disparities

The authors further note that healthcare systems already face significant health disparities.

Patients from different racial, ethnic, socioeconomic, and demographic backgrounds may experience unequal diagnosis and treatment.

If such disparities are reflected in EHR data and AI systems treat these records as objective truth, future applications may generate:

*Biased diagnoses

*Biased recommendations

*Biased risk assessments

As a result, existing healthcare inequities could become further entrenched.

Therefore, the risks associated with AI may stem not only from the model itself but also from the data used to train and inform it.

Can HIPAA Protect Patient Data Uploaded to AI Platforms?

Another key issue discussed in the article is data privacy.

Many patients assume that their medical information remains protected under the Health Insurance Portability and Accountability Act (HIPAA).

However, the authors point out that once patients voluntarily upload their medical records to a commercial AI platform, those data may no longer be fully protected under HIPAA.

The reason is that most AI platforms are not considered HIPAA-covered entities.

Consequently:

1. HIPAA restrictions on data use may not apply.

2. HIPAA security requirements may not apply.

3. HIPAA breach notification obligations may not apply.

Although AI companies may promise strong privacy protections, corporate privacy policies are fundamentally different from legally enforceable regulatory safeguards.

Lack of Transparency May Be the Greater Concern

According to the authors, the most significant challenge may not be data breaches, but rather the lack of transparency.

Currently, independent researchers have limited ability to evaluate:

*Whether AI systems exhibit bias against specific populations

*Whether safety incidents have occurred

*Whether inappropriate medical recommendations are being generated

*Whether AI is influencing patients’ healthcare-seeking behavior

Because these data remain under the control of platform developers, external validation is often impossible.

As a result, even when AI companies report strong performance, there may be insufficient independent evidence to verify such claims.

PI-Union Medical Science Commentary

Health-focused AI platforms such as ChatGPT Health represent an important milestone in the evolution of AI-powered healthcare. For the first time, patients may be able to provide AI systems with comprehensive health records for personalized analysis.

However, when AI begins reading entire medical records, it receives not only information about diseases and treatments but also decades of accumulated clinical assumptions, documentation biases, and systemic healthcare challenges.

For this reason, the future development of healthcare AI should not focus solely on technological innovation. Equal attention must be given to data governance, clinical evidence generation, regulatory oversight, and ongoing performance monitoring.

Only through robust safeguards can AI become a tool for improving healthcare outcomes rather than amplifying existing inequities and risks.

Reference: When Patients Share Everything With an AI Chatbot–Risks and Opportunities of Large Language Models

Reviewer: PI-Union Medical Science Ltd.

* E-mail: piunion@pi-union.com

* Official Website: https://pi-union.com/

* Facebook: www.facebook.com/piunion2020

* Youtube: www.youtube.com/@pi-union

* Instagram: www.instagram.com/piunion2020

* LINE: @654eukag

ChatGPT Health與AI醫療新時代：電子病歷上傳AI的風險與機會

PI-Union Medical Science — Fri, 12 Jun 2026 02:35:43 +0000

JAMA觀點導讀：當患者把完整病歷交給AI——ChatGPT Health帶來的機會與風險

從醫療知識民主化到醫療資料民主化

隨著大型語言模型（Large Language Models, LLMs）快速進入醫療領域，人工智慧不再只是提供一般健康資訊，而開始直接接觸患者最敏感的個人健康資料。

近期刊登於《JAMA》的 Viewpoint 文章〈When Patients Share Everything With an AI Chatbot: Risks and Opportunities of Large Language Models〉，探討了新一代健康型 AI 平台（如 ChatGPT Health）所帶來的機會與挑戰。作者指出，當患者能夠將完整電子病歷（Electronic Health Records, EHRs）直接同步至 AI 系統時，我們正從「醫療知識的民主化」邁向「醫療資料的民主化」。

這項變革可能促進個人化醫療發展，但同時也引發前所未有的隱私、偏見與法規問題。

AI讀取完整病歷：潛在價值何在？

理論上，如果 AI 能夠整合患者的病歷資料、健康監測數據、穿戴裝置資訊以及生活型態紀錄，將可能帶來多項效益：

* 提供更個人化的健康建議

* 協助罕見疾病的早期辨識

* 強化流行病監測能力

* 促進真實世界資料（Real-World Data, RWD）的運用

* 加速臨床研究與藥物開發

對於研究人員而言，大規模且即時的健康資料整合，更有機會成為未來精準醫療（Precision Medicine）的重要基礎。

醫療紀錄並非完全客觀

然而，作者提醒我們一個重要事實：

「電子病歷不是中立的事實倉庫。」

病歷中的內容除了客觀檢驗數據外，也包含醫護人員的主觀判斷與描述。

例如：

1. 對患者的行為評價

2. 初步診斷推測

3. 尚未證實的臨床印象

4. 可能帶有偏見的文字紀錄

如果 AI 系統直接學習這些內容，可能會將既有偏見複製甚至放大。

例如某位慢性疼痛患者曾被標註為「疑似尋求藥物（drug-seeking）」，即使後續證實其疼痛有明確生理原因，AI 仍可能受到早期紀錄影響，而提供較不適當的建議。

換言之，AI 不只是學習醫學知識，也可能學習醫療體系中的偏見。

AI可能複製醫療不平等

作者進一步指出，醫療體系本身就存在健康不平等（Health Disparities）問題。

不同種族、性別、社經背景的患者，在診斷與治療過程中可能面臨不同待遇。

若這些偏差已存在於電子病歷中，而 AI 又將其視為「真實資料」進行學習，未來可能形成：

* 偏差診斷（Biased Diagnosis）

* 偏差建議（Biased Recommendations）

* 偏差風險評估（Biased Risk Assessment）

最終使既有醫療不平等被進一步強化。

因此，AI 的風險不一定來自模型本身，而可能來自模型所學習的資料。

HIPAA保護不了患者主動上傳的資料？

文章另一項值得關注的議題是資料隱私。

許多患者認為醫療資料受到 HIPAA（Health Insurance Portability and Accountability Act）保護，因此相當安全。

然而作者指出：

當患者自行將病歷上傳至商業化 AI 平台後，這些資料未必仍受到 HIPAA 的完整保障。

原因在於：

AI 平台通常不是 HIPAA 所定義的醫療照護提供者（Covered Entity）。

因此：

1. HIPAA 的資料使用限制可能不適用

2. HIPAA 的資訊安全規範可能不適用

3. HIPAA 的資料外洩通報義務可能不適用

雖然企業可能承諾保護使用者隱私，但企業政策與法律保障仍有本質上的差異。

缺乏透明度是更大的問題

作者認為，目前最大的隱憂或許不是資料外洩，而是缺乏透明度。

目前外界難以得知：

* AI 是否對特定族群產生偏見

* 是否曾發生安全事件

* 是否提供不當醫療建議

* 是否影響患者就醫行為

由於相關資料掌握在平台開發者手中，獨立研究人員往往無法進行驗證。

因此，即使 AI 系統宣稱具有良好效能，也缺乏足夠的第三方證據支持。

百聯醫學評論

ChatGPT Health 類型的產品代表醫療 AI 發展的重要里程碑，也讓患者首次有機會將自己的完整健康資料交由 AI 分析。

然而，當 AI 開始閱讀完整病歷時，它接收到的不只是疾病資訊，也包括醫療體系長期累積的偏見、誤判與制度性問題。

因此，未來醫療 AI 的發展不應只關注技術創新，更需要建立完善的資料管理、臨床證據與法規監管機制，才能真正讓 AI 成為改善健康照護的工具，而非放大既有問題的新風險來源。

Reference: When Patients Share Everything With an AI Chatbot–Risks and Opportunities of Large Language Models

Reviewer: PI-Union Medical Science Ltd.

* E-mail: piunion@pi-union.com

* Official Website: https://pi-union.com/

* Facebook: www.facebook.com/piunion2020

* Youtube: www.youtube.com/@pi-union

* Instagram: www.instagram.com/piunion2020

* LINE: @654eukag

Key Turning Points in Cancer Vaccines

PI-Union Medical Science — Sat, 07 Mar 2026 15:28:37 +0000

Key Turning Points in Cancer Vaccines: Rethinking the Global Strategy from Technological Breakthroughs to Strategic Implementation

A Review and Commentary on the Emerging Era of Cancer Vaccines

Abstract

In recent years, rapid advances in vaccine technology and tumor immunology have brought cancer vaccines back into the global spotlight. The United Kingdom’s Cancer Vaccine Advance program, launched in 2023, aims to provide personalized mRNA cancer vaccines to 10,000 patients by 2030, marking a critical milestone in transitioning cancer vaccines from proof-of-concept to large-scale clinical application. This review summarizes the development history of cancer vaccines, recent technological breakthroughs, and clinical evidence, while analyzing how artificial intelligence (AI), genomics, and vaccine platform technologies are reshaping cancer immunotherapy. Furthermore, the review incorporates global policy and research investment trends to discuss the strategic significance of cancer vaccines in future public health and precision medicine.

I. Cancer Vaccines: From Concept to Clinical Turning Points

Vaccines have long been regarded as a public health tool for preventing infectious diseases by priming the immune system to defend against pathogens before infection occurs. Cancer vaccines, however, differ in both design and purpose. They aim not only to prevent cancer but also to serve as therapeutic tools that enable the immune system to recognize and eliminate tumor cells.

Cancer vaccines can generally be classified into two categories:

Preventive vaccines
Therapeutic vaccines

Currently, preventive vaccines remain the most clinically impactful, with notable examples including:

Hepatitis B vaccines (preventing liver cancer)
Human papillomavirus (HPV) vaccines (preventing cervical cancer)

By contrast, therapeutic cancer vaccines have developed more slowly. To date, only a limited number of therapeutic cancer vaccines or vaccine-based immunotherapies have gained regulatory approval:

BCG (Bacillus Calmette–Guérin) for bladder cancer
Talimogene laherparepvec for melanoma
Sipuleucel-T for prostate cancer

While these vaccines demonstrate the feasibility of immunotherapy, their clinical use remains limited.

II. COVID-19 Pandemic: An Unexpected Catalyst for Cancer Vaccine Development

Since 2020, the success of COVID-19 vaccines has transformed the trajectory of global vaccine technology. The rapid development and large-scale production of mRNA vaccines dramatically shortened the vaccine development cycle, which previously took decades.

Key infrastructures established during the pandemic include:

mRNA vaccine manufacturing technologies
Global cold-chain logistics
Real-time genomic surveillance
Large-scale clinical trial platforms
Cross-national government–industry collaboration models

After the pandemic, these infrastructures were quickly repurposed for other medical research fields, most notably cancer vaccines.

The UK’s 2023 Cancer Vaccine Advance program is a representative example. By integrating government, academic, and industry resources and conducting large-scale clinical trials across multiple tumor types, this program aims to evaluate the safety and efficacy of personalized mRNA cancer vaccines.

This national-level strategy demonstrates that cancer vaccines are no longer merely an academic research topic but have become a key indicator of national healthcare and biotech competitiveness.

III. Technological Breakthroughs: Personalized Cancer Vaccines and the Neoantigen Revolution

The most significant breakthrough in cancer vaccine research has come from neoantigen vaccines.

Tumor cells acquire genetic mutations that produce tumor-specific protein variants—neoantigens—that are absent in normal tissues, making them ideal immunotherapy targets.

Through next-generation sequencing (NGS) and bioinformatic analysis, researchers can:

Characterize the mutational landscape of a patient’s tumor
Predict immunogenic antigens
Design personalized vaccines

The mRNA vaccine platform provides the speed and flexibility required for personalized manufacturing.

In a phase 2 clinical trial, the personalized mRNA neoantigen vaccine mRNA-4157, combined with the immune checkpoint inhibitor pembrolizumab, showed a substantial reduction in recurrence in melanoma patients, with a recurrence-free survival rate of 79%, demonstrating the clinical potential of personalized vaccines.

Another approach involves shared antigen vaccines, which target common tumor-driving genes such as HER2 (ERBB2), and have demonstrated long-term immune responses in multiple clinical trials.

IV. Artificial Intelligence and Cancer Vaccine Design

One of the greatest challenges in cancer vaccine development is antigen selection. Tumors are highly heterogeneous, and many are immunologically “cold”, meaning their microenvironments lack immune cell infiltration, particularly T cells, limiting effective immune responses.

Artificial intelligence and high-performance computing are transforming this landscape. By training generative AI models, researchers can predict the most immunogenic antigen combinations and design multi-antigen vaccines.

The UK’s Cancer Vaccines AI & Supercompute Project trains AI models on real tumor datasets to accelerate antigen discovery and vaccine design. These technologies have the potential to shorten development timelines and improve treatment precision.

V. Global Scientific Competition and Policy Implications

From a global perspective, the United States remains a leader in cancer vaccine research, producing approximately half of all related publications. However, recent reductions in US funding for vaccine and related technology research may hinder future innovation.

In contrast, Europe and the UK are actively repurposing biotechnology infrastructure developed during the pandemic for cancer research and accelerating clinical translation through national-level programs.

The cancer vaccine market is also experiencing rapid growth:

2023: approximately $10.12 billion
2032 (projected): approximately $42.58 billion

This trend indicates that cancer vaccines are not only a medical breakthrough but also an emerging arena of global biopharmaceutical competition.

VI. Future Directions: From Treatment to Prevention

Cancer vaccine strategies are undergoing a major shift. Historically focused on treating advanced tumors, emerging research suggests that vaccination during minimal residual disease (MRD) stages may be more effective in preventing recurrence.

Future applications may include:

Postoperative recurrence prevention
Immunoprevention of premalignant lesions
Preventive vaccines for high-risk populations

This approach will increasingly integrate cancer vaccines into public health and preventive medicine frameworks.

Conclusion

The development of cancer vaccines has undergone decades of exploration, with early results limited. However, advances in mRNA technology, genomics, and artificial intelligence mark a critical turning point. The UK’s Cancer Vaccine Advance program and multiple global clinical trials indicate that cancer vaccines are transitioning from conceptual research to practical clinical application.

In the context of rising cancer incidence worldwide, cancer vaccines not only have the potential to transform tumor treatment paradigms but also to serve as a key tool in cancer prevention and public health. Continued investment in research, integration of AI, and application of precision medicine technologies may enable cancer vaccines to become the next major breakthrough in medical history.

Reference: The Time for Cancer Vaccines Is Now Advancing Toward Lasting Cancer Immunity

中文版: 癌症疫苗的關鍵轉折點：從技術突破到全球戰略的再思考

Reviewer: PI-Union Medical Science Ltd.

E-mail: piunion@pi-union.com
Official Website: https://pi-union.com/
Facebook: www.facebook.com/piunion2020
Youtube: www.youtube.com/@pi-union
Instagram: www.instagram.com/piunion2020
LINE: @654eukag

AI在醫學中的應用: GPT-4作為醫學AI聊天機器人的優點、限制和風險

PI-Union Medical Science — Wed, 14 May 2025 14:09:24 +0000

AI在醫學中的應用: GPT-4作為醫學AI聊天機器人的優點、限制和風險(NEJM)

Peter Lee, Ph.D., Sebastien Bubeck, Ph.D., and Joseph Petro, M.S., M.Eng.

本文探討了GPT-4作為醫學AI聊天機器人的應用，介紹了其潛在好處、限制和風險。
聊天機器人包括通用AI系統和聊天介面，GPT-4是一種通用AI系統，具備自然語言聊天介面。
使用聊天機器人時，使用者通過自然語言輸入查詢，機器人會快速生成相關的回應，模擬了人與人之間的對話。
聊天機器人對提示的措辭和形式敏感，需要精心設計的提示以獲得最佳結果。
GPT-4並非專門為醫療任務程式設計，但可以執行各種醫學和健康保健任務，包括醫學檔處理、診斷、研究和教育。
儘管GPT-4僅在互聯網上開放的資料基礎上進行訓練，但在醫學領域的回答正確率超過90%。
GPT-4的醫學知識可用於諮詢、診斷和教育，對醫療專業人員和研究人員提供有用的資訊。
GPT-4是一個不斷發展的工具，具有巨大潛力，但也存在限制和錯誤。在使用中需要小心謹慎，驗證其輸出的準確性。
作者認為醫學界和大眾將討論有關GPT-4和類似AI工具的性能和可信度問題，這將是未來的重要話題。
總之，GPT-4代表了醫學領域新型AI的潛在應用和風險，如果謹慎使用，這些工具有助於提供更好的醫療護理。

資料整理:百聯醫學編譯

Translator: PI-Union Medical Science

Source:

大型語言模型(LLMs)對神經科護理品質與效率的影響

PI-Union Medical Science — Sun, 11 May 2025 11:09:18 +0000

大型語言模型(LLMs)對神經科護理品質與效率的影響：神經科學中的新興議題

Lidia Moura, et al. Neurology. 2024 Jun 11;102(11):e209497.

大型語言模型（Large Language Models, LLMs）是先進的人工智慧（Artificial Intelligence, AI）系統，擅長識別和生成類似人類的語言，可能成為與神經科學相關資訊任務的寶貴工具。儘管LLMs在多個領域顯示出顯著潛力，但它們在日常臨床實踐中動態環境下的表現仍然不確定。

本文概述了在臨床環境中使用LLMs的多種局限性和挑戰，包括有限的臨床推理、可靠性和準確性不一、再現性偏差、自我服務偏差、贊助偏差，以及可能加劇醫療差距的風險。這些挑戰進一步受到實際業務考量和基礎設施需求（包括相關成本）的影響。為了克服這些障礙並有效發揮LLMs的潛力，本文為考慮在臨床實踐中使用LLMs的醫療機構、研究人員和神經科醫生提供了一些建議。

醫療機構必須培養接受AI解決方案的文化，並將其無縫融入醫療運營中。明確的目標和商業計劃應引導AI解決方案的選擇，確保其符合組織需求和預算考量。讓臨床和非臨床的相關單位參與其中，有助於確保必要的資源、建立信任，並確保AI應用的長期可持續性。測試、驗證、培訓和持續監測是成功整合的關鍵。

對於神經科醫生來說，保護患者數據隱私至關重要。尋求機構資訊技術單位的指導，以做出知情且合規的決定，並保持警惕，避免LLMs輸出中的偏見，是負責任且公正使用AI工具的重要實踐。

在研究方面，即使是已去識別的患者數據，也應獲得機構審查委員會的批准，以確保其倫理使用。遵守如SPIRIT-AI、MI-CLAIM和CONSORT-AI等既定指南，對於在AI研究中保持一致性並減少偏見至關重要。

總之，LLMs在臨床神經學中的整合提供了巨大的前景，但也帶來了巨大的挑戰。了解這些考量對於有效利用AI提升神經科護理的品質和安全性至關重要。本文為醫療機構、研究人員和神經科醫生在這個變革性領域中提供了指導。

評論：大型語言模型在神經科護理的潛力與挑戰——從創新工具到責任實踐

在神經醫療正朝向數位化與精準化加速前進的當下，Lidia Moura 等人的文章為我們帶來一份極具時代意義的觀察報告。他們不僅指出了大型語言模型（LLMs）在神經科學臨床工作中的潛能，也誠實揭示了在導入過程中所面臨的倫理、技術與制度性挑戰。

1. LLMs的潛能：神經科資訊處理與溝通的加速器

神經科醫療常涉及大量複雜資訊，包括影像診斷、病史資料、藥物紀錄與功能性評估報告等。LLMs 作為一種具自然語言處理與生成能力的人工智慧模型，其在以下方面展現高度潛力：

自動化文書處理與摘要生成：可協助減輕醫護人員記錄與報告的負擔。
病患溝通與健康教育：LLMs 可生成淺顯易懂的醫療說明文字，促進病患參與。
臨床決策支援初步草稿：可整合醫療指南與文獻資訊，提供診斷與處置方向參考。

這些應用若能妥善整合，有望提高醫療效率，釋放醫師時間，專注於更具人文關懷的臨床互動。

2. 風險與挑戰：人工智慧也會「腦霧」

作者精確點出了 LLMs 在神經科臨床實踐中所面臨的六大挑戰：

有限的臨床推理能力：LLMs 對語言的掌握並不等於醫學邏輯或因果關係的理解。
準確性與再現性不穩定：不同輸入可能產出風格一致但意義不一致的結果。
偏見風險與資料不透明性：包括訓練數據的社會文化偏見、自我強化偏誤、產業贊助影響等。
基礎設施與成本障礙：系統整合、資安維護、人力再訓練等皆需投入資源。
隱私風險：醫療資料的機密性在與第三方AI模型互動時可能受到威脅。
健康不平等的放大器：若設計不良，LLMs 可能忽視少數族群、語言障礙者與邊緣化社群的需求。

3. 系統性整合：不只是導入科技，更是改變文化

本研究最具貢獻的地方，在於其對醫療機構與臨床人員提出具體可行的實務建議：

建立AI文化與治理架構：不僅要有技術，更需跨部門溝通與價值共識。
設定清晰目標與效益評估指標：避免因追求科技感而忽略實際需求與可用性。
臨床參與設計流程：避免「工具與人分離」的疏離現象。
持續測試與教育訓練：保持模型透明、可監測、可溝通，並強化用戶的數位素養。

4. 研究倫理：透明、可重現與公平為未來AI醫學研究核心

在研究方面，文章特別強調需遵循 SPIRIT-AI、MI-CLAIM、CONSORT-AI 等新興AI研究指引，避免「科技炫耀性研究」（techno-optimism without accountability）。即便使用去識別化資料，也應由倫理委員會審查，反映出 AI 醫學研究不只是科技挑戰，更是價值與規範的實踐。

總結：將AI作為提升神經科照護的同盟，而非替代者

LLMs 在神經醫學的未來角色，不應是取代醫師的智能代理人，而應是與人類臨床智慧共舞的輔助工具。Moura 等人提醒我們，有效導入AI不只是「科技問題」，更是「倫理問題」、「制度問題」與「文化問題」。唯有透過多方參與與謹慎規劃，我們才能避免在擁抱創新時，遺落了醫療人文的初心。

Source:

Implications of Large Language Models for Quality and Efficiency of Neurologic Care

編譯與評論: 百聯醫學編譯

Translator and Reviewer: PI-Union Medical Science Ltd.

E-mail: piunion@pi-union.com

Online Courses: course.pi-union.com

Official Website: pi-union.com

Youtube: www.youtube.com/@pi-union

Instagram: www.instagram.com/piunion2020

Facebook: www.facebook.com/piunion2020

神經倫理學、隱性意識與殘障權利：當人工智慧遇上認知運動分離會發生什麼？

PI-Union Medical Science — Sun, 11 May 2025 10:58:19 +0000

神經倫理學、隱性意識與殘障權利：當人工智慧遇上認知運動分離會發生什麼？

Joseph J Fins,Kaiulani S Shulman.

J Cogn Neurosci. 2024 Jul 1;36(8):1667-1674.

在本文中，我們探討了認知運動分離（cognitive motor dissociation, CMD）與人工智慧（artificial intelligence, AI）之間的交集，也就是當CMD遇見AI時的情境。

在隱性意識狀態中，觀察到的行為、傳統的床邊評估模式以及神經影像或腦電圖（EEG）研究所揭示的對意志性指令的反應之間存在不一致。這一系列縮寫代表了隱性意識中新興技術的承諾與風險。

在診斷方面，識別認知活動與外在行為之間的不一致性具有複雜性和不確定性。在治療方面，當AI被用於生成語言時，可能會錯誤傳達那些無法表達自己想法和意圖的人的觀點。這些因素的和諧結合使AI在CMD中的應用值得深入探討。

我們以預先管理的精神提供這項分析，這是一個審慎的過程，旨在預防或減輕新技術帶來的意外後果。首先，我們考慮了CMD在臨床實踐、神經倫理學和法律領域中的規範性挑戰。

接著，我們回顧了隱性意識的歷史，探討嚴重腦損傷與安樂死運動之間的關係，並介紹了三個與腦損傷相關的傳記，突顯殘障偏見或能力主義在臨床實踐、輔助技術和轉化研究中的潛在影響。

隨後，我們探討AI如何為無法溝通的有意識個體賦予表達能力，以及這項技術必須克服的倫理挑戰，以促進人類的幸福，並引用Nussbaum和Sen所提出的“能力方法”來促進規範性推理。

評論：當AI為沉默者發聲——CMD、神經倫理學與人權的交匯點

在人工智慧快速滲透醫療場域的今天，Fins 與 Shulman 提出一個極具挑戰性卻又迫切的倫理議題：當AI技術介入無法動作、無法表達卻仍具意識的人（cognitive motor dissociation, CMD）時，誰來界定「真正的意志」？

這篇文章深刻地挑戰了我們對於「表達」、「意識」與「人性尊嚴」的基本理解，並指出科技若缺乏倫理自覺，可能在幫助某些人發聲的同時，也可能替他們說了他們從未想說的話。

一、CMD 與 AI：科技的希望與倫理的懸問

認知運動分離（CMD）患者常被誤判為無意識狀態，實則其大腦對語言與外界刺激有高度反應。AI結合fMRI或EEG等腦機介面技術，的確開啟了為這群沉默者建構表達平台的希望之窗。但這個過程並非中立或無風險：

AI是否能準確地轉譯「意志」？
當語言由演算法生成，我們怎麼知道這是患者的聲音而非模型的投射？這一點對於法律、臨床決策甚至生命維持選擇至關重要。
科技能不能成為壓迫的延伸？
在缺乏病患授權的情況下，由AI代表其「說話」，是否反而侵犯了其沉默的尊嚴？

Fins 與 Shulman 以極為清晰的倫理視角提醒我們：科技不是中性的，每一個應用場景都需要先被道德辯證所照亮。

二、能力主義偏見與殘障倫理：人性的再定義

本研究特別令人稱道的是將 CMD 案例置於更宏觀的社會文化背景下，對抗「能力主義」（ableism）的隱性假設。

CMD 不等於「生命質量低下」：
作者透過腦損傷者的傳記案例，批判當前醫療系統經常將無法互動視為「不可逆的悲劇」，而不是一種需要不同理解與支持的存在狀態。
Nussbaum 與 Sen 的能力方法（Capability Approach）：
本文引用此框架來重申，每個人都應擁有實現其潛能與尊嚴生活的機會，而非僅以生產力或溝通能力為衡量標準。

這種觀點使得AI的發展必須從「賦權」而非「替代」的邏輯出發——AI不應幫人說話，而是幫助他們自己說話。

三、預先管理精神（Anticipatory Governance）：AI醫療倫理的新典範

在科技發展總是快於倫理反思的背景下，作者提出「預先管理精神」的概念格外重要：

慎重評估技術影響而非事後補救
主動思考潛在偏誤與不平等的再製風險
將倫理納入設計初期，而非事後修補

這種態度值得作為AI醫療發展的「預防倫理」模型。

結語：當代AI醫學的倫理試煉場

Fins 與 Shulman 的文章可謂當代神經倫理學的代表作之一，它不僅關注科技本身，更強調科技背後的人性、價值與制度框架。

在CMD與AI交會的領域裡，我們所面對的，不只是能否「讓他們說話」，而是能否讓他們真正被理解、被尊重、被作為有意志的人來對待。這是一場不僅關乎技術準確度的挑戰，更是一場深刻的倫理覺醒。

Source:

Neuroethics, Covert Consciousness, and Disability Rights: What Happens When Artificial Intelligence Meets Cognitive Motor Dissociation?

Translator and Reviewer: PI-Union Medical Science Ltd.

編譯與評論: 百聯醫學編譯

E-mail: piunion@pi-union.com
Online Courses: course.pi-union.com
Official Website: https://pi-union.com/
Youtube: www.youtube.com/@pi-union
Instagram: www.instagram.com/piunion2020
Facebook: www.facebook.com/piunion2020