About

Xian Du is an Associate Professor in the Department of Mechanical and Industrial Engineering at the University of Massachusetts Amherst, affiliated with the Riccio College of Engineering. His research focuses on the scale-up of flexible electronics printing processes from laboratory settings to industrial applications, utilizing high-precision in-line inspection and pattern recognition technologies for large surface quality control. He is involved in healthcare and biomedicine research areas, contributing to advancements in manufacturing engineering and related fields. Du has been recognized for his innovative work, including being awarded a U.S. patent for a breakthrough in high-speed autofocus control. He is also a senior member of IEEE and has received pilot grants from the Elaine Marieb Center for Nursing and Engineering Innovation. His educational background includes a PhD in Innovation in Manufacturing Systems and Technology from the Singapore-MIT Alliance, a master's degree in Mechatronics Engineering from Shanghai Jiaotong University, and a bachelor's degree in Mechanical Engineering from Tianjin University.

Research topics

• Computer Science
• Materials science
• Political Science
• Medicine
• Nanotechnology
• Data science
• Bioinformatics
• Organic chemistry
• Optics
• Embedded system
• Combinatorial chemistry
• Composite material
• Photochemistry
• Chemistry

Selected publications

• SILVER-Pain Dataset

  Mendeley Data · 2026-04-24

  datasetOpen accessSenior author

  This open-access record contains the lower-risk component of the dataset: four-channel physiological signals and associated self-reported pain labels. Physiological signals in the younger cohort were collected using the Empatica E4 wristband. Physiological signals in the older cohort were collected using the Empatica EmbracePlus wristband, a newer device in the Empatica wearable line. The broader SILVER-Pain dataset also includes additional modalities collected during the study, including thermal video, depth video, RGB-derived facial action units (FAUs), and selected screening-questionnaire variables. Because those files may present a non-negligible risk of participant re-identification or sensitive information disclosure, they are not included in this open-access Mendeley record. Instead, they are available separately under a controlled-access process requiring user registration, verifiable identity information, and a signed Data Usage Agreement (DUA), consistent with the controlled-access model described by Scientific Data. The controlled-access materials and request instructions are described here: https://websites.umass.edu/xiandu/data-sharing/silver-pain-restricted/. This record includes both original physiological files and processed, ML-ready derivatives for the open-access component. Baseline utilities for parsing, synchronization, and preprocessing are available at https://github.com/straybird16/SILVER-pain. RGB-derived FAUs in the controlled-access component were obtained from RGB video using the open-source ME-GraphAU library: https://github.com/CVI-SZU/ME-GraphAU. **Access Note** This Mendeley record contains ONLY the open-access portion of the SILVER-Pain dataset. Additional controlled-access files, including thermal video, depth video, RGB-derived FAUs, and screening-questionnaire variables, are available separately through a request process requiring signed agreement to the DUA: https://websites.umass.edu/xiandu/silver-dua.

  PublisherDOI

• Improved YOLOv8 for Gesture Recognition in Human-Machine Interaction Under Complex Backgrounds

  Communications in computer and information science · 2026-01-01

  book-chapterCorresponding
  PublisherDOI

• Physics-informed displacement control for variable pattern printing with V-shaped PDMS stamps in roll-to-roll microcontact printing

  Communications Engineering · 2025-12-11

  articleOpen accessSenior author

  Roll-to-roll microcontact printing enables high-throughput production of flexible electronic devices by continuously transferring inks onto substrates via polydimethylsiloxane (PDMS) stamps. Traditional rectangular or cylindrical PDMS stamps yield uniform pattern sizes, limiting manufacturing versatility. This study introduces V-shaped PDMS stamps for variable pattern printing using a single stamp geometry. A physics-based deformation model was developed by combining finite element simulations and experiments to characterize the out-of-plane behavior of V-shaped PDMS under displacement. Leveraging this model, we implemented a neural network-based model predictive control system to precisely regulate vertical displacement and achieve desired pattern dimensions. Experimental results demonstrate that a single V-shaped PDMS stamp can reliably produce variable pattern sizes with high repeatability, improving the adaptability and process efficiency of roll-to-roll microcontact printing for flexible electronics manufacturing. Jingyang Yan and colleagues report the use of V-shaped PDMS stamps for roll-to-roll microcontact printing. With physics-informed displacement control, variable pattern sizes can be produced using a single stamp.

  PublisherOA PDFDOI

• Towards Stealthy and Effective Backdoor Attacks on Lane Detection: A Naturalistic Data Poisoning Approach

  ArXiv.org · 2025-08-04

  preprintOpen access

  Deep learning-based lane detection (LD) plays a critical role in autonomous driving and advanced driver assistance systems. However, its vulnerability to backdoor attacks presents a significant security concern. Existing backdoor attack methods on LD often exhibit limited practical utility due to the artificial and conspicuous nature of their triggers. To address this limitation and investigate the impact of more ecologically valid backdoor attacks on LD models, we examine the common data poisoning attack and introduce DBALD, a novel diffusion-based data poisoning framework for generating naturalistic backdoor triggers. DBALD comprises two key components: optimal trigger position finding and stealthy trigger generation. Given the insight that attack performance varies depending on the trigger position, we propose a heatmap-based method to identify the optimal trigger location, with gradient analysis to generate attack-specific heatmaps. A region-based editing diffusion process is then applied to synthesize visually plausible triggers within the most susceptible regions identified previously. Furthermore, to ensure scene integrity and stealthy attacks, we introduce two loss strategies: one for preserving lane structure and another for maintaining the consistency of the driving scene. Consequently, compared to existing attack methods, DBALD achieves both a high attack success rate and superior stealthiness. Extensive experiments on 4 mainstream LD models show that DBALD exceeds state-of-the-art methods, with an average success rate improvement of +10.87% and significantly enhanced stealthiness. The experimental results highlight significant practical challenges in ensuring model robustness against real-world backdoor threats in LD.

  PublisherOA PDFDOI

• Three-Dimensional Printing of Braille-Integrated Drug-Loaded Materials: Advancing Accessibility for Visually Impaired Patients

  ACS Applied Polymer Materials · 2025-05-05 · 4 citations

  articleSenior authorCorresponding

  3D printing presents a transformative pathway for personalized medicine, particularly in designing accessible medical devices for individuals with visual impairments. Millions worldwide face significant challenges in managing their healthcare routines due to nonstandardized Braille labeling and inadequately designed devices. This study utilizes digital light processing (DLP)-based 3D printing combined with reversible addition–fragmentation chain transfer (RAFT) polymerization to fabricate high-resolution, Braille-integrated materials tailored for visually impaired patients. Investigations focused on optimizing surface quality, structural integrity, and tactile readability to enhance device accessibility and usability. Comprehensive analyses examined the interplay between photopolymer formulations, mechanical performance, and tactile features. Furthermore, this study assessed the integration of pharmaceutical elements, such as drug loading and release characteristics, to evaluate the potential for multifunctional applications. These findings highlight the challenges and strategies involved in integrating tactile accessibility with biomedical functionality for visually impaired individuals.

  PublisherDOI

• Intermediate Layer in Titanium/Steel Dissimilar Welding: A Review

  Advanced Engineering Materials · 2025-06-11 · 9 citations

  review

  Dissimilar metal joints comprising titanium and steel are widely utilized in industrial fields. To mitigate the potential deterioration of welded joints resulting from the formation of brittle intermetallic compounds (IMCs) due to the direct contact between titanium and steel, an intermediary metal layer is typically employed during the welding process to prevent or restrict the generation of IMCs and thereby realize joints with superior mechanical properties. This study presents a comprehensive analysis of literature studies and technical processes pertaining to titanium/steel welding over the last decade. It provides a systematic review of the research progress in the dissimilar metal welding of titanium/steel systems with a particular focus on the utilization of intermediate layer metals. In addition, it discusses the existing challenges in the field of titanium/steel dissimilar metal welding in this area. Finally, this review critically analyzes emerging strategies for dissimilar material joining and proposes that in situ high‐entropy alloying of welded joints via fusion welding processes, guided by entropy design principles, represents a transformative pathway for next‐generation heterogeneous material welding technologies.

  PublisherDOI

• Spatial-terminal iterative learning control for registration error elimination in roll-to-roll gravure printing systems

  Journal of Manufacturing Processes · 2025-06-23 · 2 citations

  article
  PublisherDOI

• HO-Based Auto-disturbances Rejection Control of Mode Transition Processes in Turbine-Based Combined Cycle Engine

  Lecture notes in mechanical engineering · 2025-01-01

  book-chapter
  PublisherDOI

• Incorporation of clinical features into a multivariate logistic regression model for the differential diagnosis of benign and malignant TI-RADS 4 thyroid nodules

  Frontiers in Endocrinology · 2025-05-29 · 3 citations

  articleOpen access

  Objective: This study aimed to explore the diagnostic value of clinical features in the assessment of malignant thyroid Imaging Reporting and Data System (TIRADS) category 4 thyroid nodules and to provide a more effective reference for clinical diagnostic practices. Methods: A total of 998 patients with 1,103 TIRADS 4 thyroid nodules underwent conventional ultrasound (US) and clinical information assessment at the Shanghai Health and Medical Center from January 1, 2012, to June 30, 2024. A qualitative assessment of clinical and US features was performed, followed by univariable and multivariable logistic regression analyses using a training cohort, which contributed to the construction of the clinical TIRADS model. A receiver-operating characteristic (ROC) curve, a Hosmer-Lemeshow (HL) test and a decision curve analysis (DCA) were employed to further validate this model in the validation cohort. Results: Patient age, body mass index, sex, family history of thyroid carcinoma, and US features-such as vertical orientation, ill-defined or irregular margins or extrathyroidal extensions, microcalcifications, blood flow signals of central or peripheral vessels, and swollen cervical lymph nodes-were identified as independent risk factors in the clinical scoring model for TI-RADS 4 nodules. This diagnostic model achieved an area under the curve (AUC) of 0.943 [0.928, 0.959], with a sensitivity of 82.33%, specificity of 94.44%, diagnostic threshold of 5 points, accuracy of 87.42%, positive predictive value of 95.34%, and negative predictive value of 79.48% in the validation cohort. The HL tests and DCA also demonstrated excellent predictive performances. Conclusions: The integration of clinical and US features in the construction of the diagnostic model can significantly enhance the diagnosis of TIRADS 4 thyroid nodules and provide a reliable evaluation tool for clinical practice.

  PublisherOA PDFDOI

• Physics-Informed Displacement Control for Variable Pattern Printing with V-shaped PDMS Stamps in Roll-to-Roll Microcontact Printing

  Research Square · 2025-06-25

  preprintOpen access1st authorCorresponding
  PublisherOA PDFDOI

Recent grants

• CAREER: Modeling the Roll-to-Roll Soft Lithography Printing Process Through Deep Learning and Real-time Sensing

  NSF · $643k · 2020–2027

• GOALI: Monitoring and Control of Roll-to-Roll Printing of Flexible Electronics through Multiscale In-Line Metrology

  NSF · $591k · 2019–2023

Frequent coauthors

• C. Timossi

  Lawrence Berkeley National Laboratory

  12 shared

• C. Cork

  Lawrence Livermore National Laboratory

  12 shared

• A. Robb

  12 shared

• Sumeet Dua

  Louisiana Tech University

  10 shared

• James Shin Young

  University of California, Berkeley

  8 shared

• S. Magyary

  8 shared

• Yong Luo

  Chongqing University

  8 shared

• M. Fahmie

  Lawrence Berkeley National Laboratory

  8 shared

Education

• Ph.D., Singapore-MIT Alliance

  National University of Singapore

Awards & honors

• U.S. Patent for Breakthrough in High-speed Autofocus Control
• IEEE Senior Member