arXiv:2511.18272v1 Announce Type: new 
Abstract: Large vision-language models (VLMs) are increasingly deployed for optical character recognition (OCR) in healthcare settings, raising critical concerns about protected health information (PHI) exposure during document processing. This work presents the first systematic evaluation of inference-time vision token masking as a privacy-preserving mechanism for medical document OCR using DeepSeek-OCR. We introduce seven masking strategies (V3-V9) targeting different architectural layers (SAM encoder blocks, compression layers, dual vision encoders, projector fusion) and evaluate PHI reduction across HIPAA-defined categories using 100 synthetic medical billing statements (drawn from a corpus of 38,517 annotated documents) with perfect ground-truth annotations. All masking strategies converge to 42.9% PHI reduction, successfully suppressing long-form spatially-distributed identifiers (patient names, dates of birth, physical addresses at 100% effectiveness) while failing to prevent short structured identifiers (medical record numbers, social security numbers, email addresses, account numbers at 0% effectiveness). Ablation studies varying mask expansion radius (r=1,2,3) demonstrate that increased spatial coverage does not improve reduction beyond this ceiling, indicating that language model contextual inference - not insufficient visual masking - drives structured identifier leakage. A simulated hybrid architecture combining vision masking with NLP post-processing achieves 88.6% total PHI reduction (assuming 80% NLP accuracy on remaining identifiers). This negative result establishes boundaries for vision-only privacy interventions in VLMs, provides guidance distinguishing PHI types amenable to vision-level versus language-level redaction, and redirects future research toward decoder-level fine-tuning and hybrid defense-in-depth architectures for HIPAA-compliant medical document processing.

أظهرت تقييم منهجي لتقنية إخفاء الرموز البصرية في التعرف الضوئي على الحروف (OCR) للوثائق الطبية أنه على الرغم من قدرتها على تقليل تسرب المعلومات الصحية المحمية (PHI)، إلا أنها ليست كافية بمفردها. استخدمت الدراسة DeepSeek-OCR واختبرت سبع استراتيجيات إخفاء، محققةً تقليصًا بنسبة 42.9% في PHI عبر فئات مختلفة محددة بواسطة HIPAA، باستخدام بيانات فواتير طبية اصطناعية للتحليل.

Una evaluación sistemática del enmascaramiento de tokens visuales en el OCR de documentos médicos ha revelado que, aunque puede reducir la filtración de información de salud protegida (PHI), no es suficiente por sí solo. El estudio utilizó DeepSeek-OCR y probó siete estrategias de enmascaramiento, logrando una reducción del 42.9% en PHI en varias categorías definidas por HIPAA, utilizando estados de facturación médica sintéticos para el análisis.

Une évaluation systématique du masquage des jetons visuels dans l'OCR de documents médicaux a révélé que, bien qu'il puisse réduire les fuites d'informations de santé protégées (PHI), il n'est pas suffisant à lui seul. L'étude a utilisé DeepSeek-OCR et testé sept stratégies de masquage, atteignant une réduction de 42,9 % des PHI dans diverses catégories définies par la HIPAA, en utilisant des états de facturation médicale synthétiques pour l'analyse.

A systematic evaluation of vision token masking in medical document OCR has revealed that while it can reduce protected health information (PHI) leakage, it is not sufficient on its own. The study utilized DeepSeek-OCR and tested seven masking strategies, achieving a 42.9% reduction in PHI across various categories defined by HIPAA, using synthetic medical billing statements for analysis.

Vision Token Masking Alone Cannot Prevent PHI Leakage in Medical Document OCR: A Systematic Evaluation

arXiv:2511.20983v1 Announce Type: new 
Abstract: Collaborative machine learning across healthcare institutions promises improved diagnostic accuracy by leveraging diverse datasets, yet privacy regulations such as HIPAA prohibit direct patient data sharing. While federated learning (FL) enables decentralized training without raw data exchange, recent studies show that model gradients in conventional FL remain vulnerable to reconstruction attacks, potentially exposing sensitive medical information. This paper presents a privacy-preserving federated learning framework combining Vision Transformers (ViT) with homomorphic encryption (HE) for secure multi-institutional histopathology classification. The approach leverages the ViT CLS token as a compact 768-dimensional feature representation for secure aggregation, encrypting these tokens using CKKS homomorphic encryption before transmission to the server. We demonstrate that encrypting CLS tokens achieves a 30-fold communication reduction compared to gradient encryption while maintaining strong privacy guarantees. Through evaluation on a three-client federated setup for lung cancer histopathology classification, we show that gradients are highly susceptible to model inversion attacks (PSNR: 52.26 dB, SSIM: 0.999, NMI: 0.741), enabling near-perfect image reconstruction. In contrast, the proposed CLS-protected HE approach prevents such attacks while enabling encrypted inference directly on ciphertexts, requiring only 326 KB of encrypted data transmission per aggregation round. The framework achieves 96.12 percent global classification accuracy in the unencrypted domain and 90.02 percent in the encrypted domain.

تم تقديم إطار جديد للتعلم الفيدرالي الذي يحافظ على الخصوصية، يجمع بين المحولات البصرية والتشفير المتجانس الخفيف لتحسين تصنيف علم الأمراض النسيجية عبر مؤسسات الرعاية الصحية المتعددة. تتناول هذه الطريقة التحديات التي تفرضها لوائح الخصوصية مثل HIPAA، التي تحظر تبادل بيانات المرضى المباشرة، بينما تتيح التعلم التعاوني.

Se ha introducido un nuevo marco para el aprendizaje federado que preserva la privacidad, combinando Transformadores de Visión con cifrado homomórfico ligero para mejorar la clasificación en histopatología entre múltiples instituciones de salud. Este enfoque aborda los desafíos planteados por regulaciones de privacidad como HIPAA, que restringen el intercambio directo de datos de pacientes, mientras permite el aprendizaje colaborativo.

Un nouveau cadre pour l'apprentissage fédéré préservant la vie privée a été introduit, combinant des Transformers de Vision avec un chiffrement homomorphe léger pour améliorer la classification en histopathologie entre plusieurs institutions de santé. Cette approche répond aux défis posés par les réglementations sur la vie privée comme la HIPAA, qui interdisent le partage direct des données des patients, tout en permettant l'apprentissage collaboratif.

A new framework for privacy-preserving federated learning has been introduced, combining Vision Transformers with lightweight homomorphic encryption to enhance histopathology classification across multiple healthcare institutions. This approach addresses the challenges posed by privacy regulations like HIPAA, which restrict direct patient data sharing, while still enabling collaborative machine learning.

Privacy-Preserving Federated Vision Transformer Learning Leveraging Lightweight Homomorphic Encryption in Medical AI

<p>Protecting Customer Data Across Voice AI and Automation 2X Solutions, the performance infrastructure for outbound Voice AI and human-in-the-loop automation, today announced successful completion of its SOC 2 Type II certification and HIPAA compliance across its platform. The milestone underscores 2X’s commitment to helping customers connect, comply, and convert outbound and inbound AI calling...</p>
<p>The post <a href="https://ai-techpark.com/2x-solutions-achieves-soc-2-type-ii-and-hipaa-compliance/">2X Solutions Achieves SOC 2 Type II and HIPAA Compliance</a> first appeared on <a href="https://ai-techpark.com">AI-Tech Park</a>.</p>

حققت شركة 2X Solutions بنجاح شهادة SOC 2 Type II وتوافقت مع HIPAA عبر منصتها، مما يعزز التزامها بحماية بيانات العملاء في مجال الذكاء الاصطناعي الصوتي والأتمتة.

2X Solutions ha completado con éxito su certificación SOC 2 Tipo II y ha logrado la conformidad con HIPAA en su plataforma, lo que refuerza su compromiso de proteger los datos de los clientes en el ámbito de la IA de voz y la automatización.

2X Solutions a réussi à obtenir la certification SOC 2 Type II et à atteindre la conformité HIPAA sur sa plateforme, renforçant ainsi son engagement à protéger les données des clients dans le domaine de l'IA vocale et de l'automatisation.

2X Solutions has successfully completed its SOC 2 Type II certification and achieved HIPAA compliance across its platform, reinforcing its commitment to safeguarding customer data in the realm of Voice AI and automation.

Vision Token Masking Alone Cannot Prevent PHI Leakage in Medical Document OCR: A Systematic Evaluation

Was this article worth reading? Share it

GPTHuman

Humanize AI

Vaulteq