About

Yingzi Lin is a Professor of Mechanical and Industrial Engineering and the director of the Intelligent Human-Machine Systems (IHMS) Laboratory at Northeastern University in Boston, MA. Her expertise encompasses intelligent human-machine systems, human factors and their applications in healthcare and transportation safety, smart structures and systems, sensors and sensing systems, multimodality information fusion, driver-vehicle systems, patient safety, human-machine interface design, human-robot interaction, robotics, and human-friendly mechatronics. Her research focuses on human-machine systems, biosensing, smart systems, human state and behavior modeling, transportation safety, healthcare, and patient safety. Dr. Lin's research has been funded by prominent agencies such as the National Science Foundation (NSF), National Institute of Standards and Technology (NIST), Office of Naval Research (ONR), National Institutes of Health (NIH), and the Natural Sciences and Engineering Research Council of Canada (NSERC), along with major industries including GM and BOSE. She has received several prestigious awards, including the NSF CAREER award, NSERC University Faculty Award, and other honors recognizing her contributions to research and mentoring. She has published over two hundred technical papers in refereed journals, conference proceedings, and book chapters. Her educational background includes a PhD in Mechanical Engineering from the University of Saskatchewan, obtained in 2004.

Research topics

• Computer Science
• Artificial Intelligence
• Machine Learning
• Psychology
• Human–computer interaction
• Mathematics
• Physical therapy
• Simulation
• Statistics
• Medicine

Selected publications

• The influence of in-car air quality on drivers’ brain states with hybrid fNIRS and EEG

  Building and Environment · 2025-09-08

  articleOpen access

  • Investigated the effects of CO 2 and body odor on driving brain states during driving. • Enrolled 25 participants in simulated highway driving under varying conditions. • Observed significant impact of CO 2 on very few EEG channels. • Altered EEG band power ratios found during cognition task due to body odor. • No significant impact of CO 2 or body odor on all fNIRS channels. This study investigates the effects of in-car carbon dioxide (CO 2 ) and body odor on cognitive performance during driving using advanced neuroimaging technologies. Prior literature on building environments suggested that CO 2 and body odor can negatively impact cognitive abilities, especially when building ventilation is limited. Various indoor environmental factors may hinder cognition and therefore driving performance, thereby raising concerns for transportation safety. In our study, we investigated the influence of elevated CO 2 and body odor on performance of driving and N-back tasks. We enrolled 25 participants in simulated highway driving scenarios for a two-factor experimental setup, varying the indoor CO 2 concentrations across three levels (800, 1800, and 3500 ppm) and two levels of body odor. CO 2 concentrations in the cabin were increased by introducing pure CO 2 and body odor was simulated by placing worn T-shirts in the cabin, while maintaining other environmental factors constant. Electroencephalography (EEG) and functional Near-Infrared Spectroscopy (fNIRS) were applied to monitor brain activities during driving. EEG data features included power spectral density (PSD) in delta, theta, alpha, and beta bands, and various band power ratios, while fNIRS data focused on the metrics of oxyhemoglobin (HbO) and deoxyhemoglobin (HbR). The findings indicated that body odor significantly impacts EEG band power ratios, especially during the secondary cognition task while driving. Specifically, the ratio index (α+θ)/β was lower in the condition with body odor from T-shirts, indicating increased alertness. Concurrently, with body odor, we found a lower θ/β ratio that was associated with an increase in stimulus-driven attention and an enhanced ability of the subjects to concentrate. In contrast, CO 2 levels exhibited a nuanced influence on cognitive functions, with insignificant impact on EEG band power or band power ratios observed. The results suggest a complex or trivial relationship between CO 2 exposure and cognitive responses that our neuroimaging modalities could not directly unravel. Moreover, fNIRS data did not indicate significant hemodynamic response changes attributable to CO 2 or body odor. The study contributes to the understanding of how CO 2 and body odor affect cognitive performance during driving, with implications for improving driving safety and designing better in-car environments.

  PublisherDOI

• Physiological Sensor Modality Sensitivity Test for Pain Intensity Classification in Quantitative Sensory Testing

  Sensors · 2025-03-26 · 4 citations

  articleOpen accessSenior authorCorresponding

  Chronic pain is prevalent and disproportionately impacts adults with a lower quality of life. Although subjective self-reporting is the "gold standard" for pain assessment, tools are needed to objectively monitor and account for inter-individual differences. This study introduced a novel framework to objectively classify pain intensity levels using physiological signals during Quantitative Sensory Testing sessions. Twenty-four participants participated in the study wearing physiological sensors (blood volume pulse (BVP), galvanic skin response (GSR), electromyography (EMG), respiration rate (RR), skin temperature (ST), and pupillometry). This study employed two analysis plans. Plan 1 utilized a grid search methodology with a 10-fold cross-validation framework to optimize time windows (1-5 s) and machine learning hyperparameters for pain classification tasks. The optimal time windows were identified as 3 s for the pressure session, 2 s for the pinprick session, and 1 s for the cuff session. Analysis Plan 2 implemented a leave-one-out design to evaluate the individual contribution of each sensor modality. By systematically excluding one sensor's features at a time, the performance of these sensor sets was compared to the full model using Wilcoxon signed-rank tests. BVP emerged as a critical sensor, significantly influencing performance in both pinprick and cuff sessions. Conversely, GSR, RR, and pupillometry demonstrated stimulus-specific sensitivity, significantly contributing to the cuff session but with limited influence in other sessions. EMG and ST showed minimal impact across all sessions, suggesting they are non-critical and suitable for reducing sensor redundancy. These findings advance the design of sensor configurations for personalized pain management. Future research will focus on refining sensor integration and addressing stimulus-specific physiological responses.

  PublisherOA PDFDOI

• Beyond Pain Intensity Scores: A Systematic Review of Multimodal Digital Neurophysiological Sensors for Objective Pain Assessment in the Clinical Application

  Proceedings of the Human Factors and Ergonomics Society Annual Meeting · 2025-09-01

  reviewSenior authorCorresponding

  People have been looking for an objective measure of chronic pain for decades. Current objective measures can be achieved by correlating patients’ self-report with physiological signal features, but such measures often yield inconsistent results due to individual variability. While existing approaches heavily rely on pain intensity, this study reviewed literatures in differentiating pain types and pain locations using neurophysiological sensors. Results showed that using modalities like EEG and fNIRS, high classification accuracy for specific stimuli can be achieved. However, most existing datasets lack integration of both neurological and peripheral physiological signals and do not label multimodal pain dimensions, such as type, intensity, and location. The review highlights an urgent need for datasets that combine these modalities to support objective pain assessment. Future research should also prioritize multidimensional labeling, population diversity, and validation frameworks to advance personalized pain management and rehabilitation workflows.

  PublisherDOI

• Impact of Visual Perception Mismatch Design on Response Time in Mixed Reality

  Proceedings of the Human Factors and Ergonomics Society Annual Meeting · 2025-09-01

  articleSenior authorCorresponding

  Mixed reality technology has provided more possibilities for users to interact with both the virtual and real worlds. By integrating virtual elements with real scenes, mixed reality offers a novel immersive experience. Mixed reality also introduces complex perceptual and cognitive challenges when processing conflicting information from the virtual and real worlds. This study investigates the impact of color mismatches on response time in a mixed reality scenario. We recruited 15 participants to identify the font color of words while ignoring their meaning. The experiment included three levels of perceptual-conceptual matching. We performed a global comparison across all conditions, a within-group comparison, and a between-group comparison. Results show that in the mixed reality scenario, the processing of visual information relies on low-level physical properties (e.g., brightness and hue) and is modulated by high-level cognitive factors (e.g., conflict processing and expectation). These findings indicate that visual mismatches in mixed reality can significantly affect response time, particularly when perceptual salience is low and different from cognitive expectations. The results have implications for the design of mixed reality interfaces in complex environments where users must process information from the virtual and real worlds, such as head-up displays, warehouse management systems, and industrial instructional overlays.

  PublisherDOI

• Font and background color combinations influence recognition efficiency: A novel method via primary color Euclidean distance and response surface analysis

  Displays · 2024-11-12 · 1 citations

  articleSenior authorCorresponding
  PublisherDOI

• A Systematic Review and Meta-Analysis of Older Adults’ Cognitive Impairments and Effects on Road Safety

  Proceedings of the Human Factors and Ergonomics Society Annual Meeting · 2024-09-01 · 2 citations

  reviewOpen accessSenior authorCorresponding

  Driving is challenging for older adults, especially those with cognitive impairments. This paper systematically reviewed 39 peer-reviewed studies to examine the impact of cognitive deficits on older adults’ driving abilities, focusing on the discussion of neuropsychological assessments including Trail Making Tests, Useful Field of View, Maze Test, and Mini-Mental Status Exam. The study proposed the concept of Road Safety Cognitive Health, which encompasses the cognitive processes crucial for safe driving, aiming to inform future driver training, licensing regulations, roadway designs, and vehicle technology innovations. Concussively, this research advances understanding of road safety challenges for cognitively impaired older adults and advocates for an integrative approach to ensure their driving safety.

  PublisherDOI

• Using Physiological Signals for Pain Assessment: An Evaluation of Deep Learning Models

  2024-10-03 · 2 citations

  articleSenior author

  Pain assessment is of major significance in clinical environments. The current gold standard is self-reporting of pain based on the patient’s subjective willingness. However, pain assessment based on physiological signals is developing rapidly due to the objectivity and real-time nature of physiological signals. This study aims to systematically compare the performance of convolutional neural networks (CNN) combined with long short-term memory networks (LSTM), bidirectional long short-term memory networks (BiLSTM), transformers, and gated recurrent units (GRU) models in pain classification tasks. We assessed these hybrid models’ performance experimentally using a variety of metrics, such as accuracy, precision, recall, F1 score, training time, and inference time. The experimental results show that the CNN+Transformer model best performs in most evaluation metrics with an accuracy of 0.795, while the CNN+GRU model performs the worst with an accuracy of only 0.559. In addition, we also analyze the computational efficiency of each model in terms of training and inference time. Overall, this paper provides important direction for future research and real-world applications by thoroughly evaluating the effectiveness of various deep learning models in pain classification tasks.

  PublisherDOI

• Temperature Influences Driving Performance, While Interior Ambient Lighting Does Not

  SSRN Electronic Journal · 2024-01-01

  preprintOpen access
  PublisherDOI

• Efficacy of Low‐Dose Scopolamine and Palonosetron in Reducing Immediate Post‐Gastrointestinal Endoscopy Nausea and Vomiting: A Prospective, Randomized, Controlled Study

  The Journal of Clinical Pharmacology · 2024-09-03

  article

  This study aimed to assess the incidence of post-discharge nausea and vomiting (PDNV) following sedation with nalbuphine and etomidate and to evaluate the prophylactic effects of scopolamine in reducing PDNV. A two-stage prospective clinical trial was conducted. The first part involved an observational study of 77 subjects to assess the PDNV incidence post-sedation with nalbuphine, etomidate, and propofol. The second part compared the effectiveness of palonosetron 0.075 mg (P group), scopolamine 0.1 mg (S group), and their combination (PS group) in reducing PDNV. The primary endpoint was the incidence of PDNV within 8 h post-sedation. Secondary outcomes included PDNV frequency and severity at 8-24, 0-24, and 24-48 h and side effects of medications. The incidence of PDNV within 8 h post-sedation was 37.66% (29/77). The PS group showed a significantly lower PDNV rate of 2.56% within 8 h, compared to the P group (35.71%, P < .001), S group (19.64%, P < .001), and control group (38.39%, P < .001), respectively. The S group (19.64%) also had a lower rate than the P group (35.71%, P = .007) and the control group (38.39%, P = .002). Subgroup analysis suggested a potential differential effect of palonosetron in reducing vomiting among male patients undergoing gastrointestinal procedures. The combination therapy was also associated with fewer cases of mild or no nausea and vomiting. In summary, the incidence of PDNV following sedation with nalbuphine and etomidate was notably high. The combination of scopolamine and palonosetron was more effective in preventing PDNV, with implications for improved post-sedation care.

  PublisherDOI

• A Novel Labeling Method of Physiological-based Pressure Pain Assessment Among Patients With and Without Chronic Low Back Pain

  Proceedings of the Human Factors and Ergonomics Society Annual Meeting · 2024-09-01 · 3 citations

  articleOpen accessSenior authorCorresponding

  Pain, especially chronic pain, is a complicated and subjective experience, threatening global healthcare as one of the most severe health problems. Traditionally, pain is assessed by Visual Analog Scale to indicate the pain intensity by the patient’s self-report, causing them to become biased by various psychosocial factors. In this study, we performed two distinct labeling methods to assess the pressure pain in Quantitative Sensory Testing and to differentiate healthy controls and chronic low back pain patients: time period labels and percentage timestamp labels. Physiological signals such as blood volume pulse and galvanic skin response were collected. The time period labeling method was to segment via fixed time windows. The percentage timestamp labeling method was to select the timestamp labels based on the percentage of the threshold or the tolerance time. Both methods demonstrate different advantages when visualizing the information of different pain states and different participant groups.

  PublisherOA PDFDOI

Recent grants

• Integrated Individualized Modeling towards Cognitive Control of Human-Machine Systems

  NSF · $235k · 2013–2017

• CAREER: Bridging Cognitive Science and Sensor Technology: Non-intrusive and Multi-modality Sensing in Human-Machine Interactions

  NSF · $400k · 2010–2016

• CNT-Integrated Sensing System for Driver State Detection

  NSF · $220k · 2008–2011

• I-Corps: Thin Film Cardiac Sensor

  NSF · $50k · 2016–2018

• SCH: INT: Collaborative Research: Novel Computational Methods for Continuous Objective Multimodal Pain Assessment Sensing System (COMPASS)

  NSF · $630k · 2018–2027

Frequent coauthors

• Wenjun Zhang

  Indiana University – Purdue University Indianapolis

  41 shared

• Mingxin Yu

  American Institute of Aeronautics and Astronautics

  22 shared

• Jing Du

  University of Florida

  13 shared

• Yikang Guo

  Nanjing Agricultural University

  12 shared

• Jiajun Zhou

  China University of Geosciences

  11 shared

• Hua Cai

  Peking University

  10 shared

• Xifan Yao

  South China University of Technology

  9 shared

• Xiaoying Tang

  9 shared

Awards & honors

• Distinguished Faculty Award (2026)
• Excellence in Mentoring Award (2023)
• Outstanding Paper Award (2013), IEEE Computational Intellige…
• Excellence in Research Award (2012), Northeastern University…
• Early Career Development Award (2011), College of Engineerin…