Zunliang Wang

Clinical pain management requires continuous and objective monitoring to support personalized analgesic decisions and treatment evaluations. However, existing subjective scales and intermittent observations often fail to capture rapid pain fluctuations in time. In this work, we propose a continuous pain assessment approach based on facial video sequences, which integrates multi-scale spatiotemporal feature fusion with a lightweight temporal attention mechanism. This design enables the model to capture both subtle facial micro-movements as well as head and limb behaviors for continuous pain estimation. To address the high computational cost and real-time inference challenges brought by spatiotemporal video data, we further design a plug-and-play module named Pain-Adapter based on parameter-efficient fine-tuning (PEFT). Unlike generic adapters that treat features uniformly, Pain-Adapter employs a decoupled design with specialized spatial and temporal adaptation pathways to explicitly model the dynamic evolution of pain. By keeping the pre-trained backbone frozen, Pain-Adapter introduces spatial and temporal adapters to enhance representation learning and temporal modeling, while optimizing only a small fraction of parameters to facilitate efficient transfer learning for pain assessment tasks. This significantly reduces training and deployment costs. The experimental results on the UNBC-McMaster dataset show that applying Pain-Adapter to Swin Transformer-Large backbone achieves an accuracy of 82.76%, with significant reductions in parameter count and computational cost, demonstrating strong potential for clinical deployment.

During seasonal influenza or emerging respiratory outbreaks, rapid home-based multiplex molecular point-of-care testing (POCT) for respiratory pathogens is crucial for early diagnosis and intervention, particularly in vulnerable populations. However, existing POCT systems, primarily designed for clinical settings, are often too complex, costly, and reliant on trained operators, limiting their suitability for home use. To overcome these barriers, we introduce a microfluidic cartridge-based system leveraging recombinase polymerase amplification (RPA) for multiplexed detection of respiratory pathogens in home environments. The microfluidic cartridge is designed with three parallel channels—each integrating a lysis chamber, an RPA chamber preloaded with lyophilized reagents, and an air storage chamber. Each detection channel enables extraction-free, single-channel 3-plex RPA assays, and by combining three-channel parallel detection, the system achieves simultaneous identification of eight respiratory pathogens and one internal control in under 25 min. A novel pneumatic pressure pumping strategy ensures precise flow control through dynamic bladder compression, paired with microchannel hydraulic resistance matching to guarantee uniform volumetric distribution and synchronized flow across all channels. Furthermore, a dynamic mixing method promotes homogeneous mixing of RPA reagents with lysed samples via a bidirectional flow between the lysis and RPA chambers, enhancing assay reliability. Our microfluidic design enables significant miniaturization, yielding a compact, lightweight system (97%) and specificity (>99%), ensuring reliable at-home diagnosis of respiratory co-infections without requiring professional operation.

We present a cost-effective and simple multiplex nucleic acid quantification method using droplet digital PCR (ddPCR) with digital melting curve analysis (MCA). This approach eliminates the need for complex fluorescent probe design, reducing both costs and dependence on fluorescence channels. We developed a convolutional neighborhood search algorithm to correct droplet displacement during heating, ensuring precise tracking and accurate extraction of melting curves. An experimental protocol for digital MCA on the ddPCR platform was established, enabling accurate quantification of six target pathogen genes using a single fluorescence channel, with an average accuracy of 85%. Our method overcomes the multiplexing limitations of ddPCR, facilitating its application in multi-target pathogen detection.

Molecular point‐of‐care testing (POCT) system is crucial for the timely prevention and control of infectious diseases. We recently proposed a gravity‐driven microfluidic cartridge for molecular POCT detection, without the need for external sources or actuators, demonstrating the advantages in terms of the reduced cartridge size and low development costs. How to achieve precise control of liquid flow behavior is challenging for the gravity‐driven cartridge. In this work, we explored the underlying mechanism of flow control in the cartridge and offered optimized solutions for our cartridge design to achieve precise control of dynamic flow rates and enhance pumping efficiency significantly. Through the computational fluid dynamics simulations, we demonstrated that adopting an asymptotic contraction chamber geometry design and a closed‐loop air flow channel design with the cartridge inlet can facilitate stable laminar flow of the liquid in our microfluidic cartridge, enabling precise control of flow velocity. We further optimized the microchannel diameter and the contact angle of the liquid with the microchannel wall. The effectiveness of the optimized cartridge for POCT detection was well validated by the accurate detection of the human papillomavirus type 16 virus in the 120 clinical swab samples.

Nonlinear effect in an ultrafast fiber laser has always been a major obstacle in realization of laser with narrow pulse width and high energy. To achieve higher peak power, we herein introduce pre-chirping management and pre-amplification before the chirped pulse amplification (CPA) system. The self-phase modulation effect is used to effectively stretch the spectrum when the pulse reaches a high peak power, and then the parabolic pulse and spectrum are obtained in a fiber amplifier. This parabolic pulse effectively reduces the impact of nonlinear effect in subsequent fiber amplifier and realizes the amplification of chirped pulse with a higher peak power compared with conventional CPA system. We finally achieved a mode-locked pulse with 143 fs pulse duration, 8.57 W average power, 85.7 μJ single pulse energy, M² < 1.10 beam quality, and without obvious pulse degeneration. The reported pulse energy is the highest among all-fiber CPA systems based on nonlinear amplification technology.

The pandemic of coronavirus disease 2019 (COVID-19), due to the novel coronavirus (SARS-CoV-2), has created an unprecedented threat to the global health system, especially in resource-limited areas. This challenge shines a spotlight on the urgent need for a point-of-care (POC) quantitative real-time PCR (qPCR) test for sensitive and rapid diagnosis of viral infections. In a POC system, a closed, single-use, microfluidic cartridge is commonly utilized for integration of nucleic acid preparation, PCR amplification and florescence detection. But, most current cartridge systems often involve complicated nucleic acid extraction via active pumping that relies on cumbersome external hardware, causing increases in system complexity and cost. In this work, we demonstrate a gravity-driven cartridge design for an integrated viral RNA/DNA diagnostic test that does not require auxiliary hardware for fluid pumping due to adopted extraction-free amplification. This microfluidic cartridge only contains two reaction chambers for nucleic acid lysis and amplification respectively, enabling a fast qPCR test in less than 30 min. This gravity-driven pumping strategy can help simplify and minimize the microfluidic cartridge, thus enabling high-throughput (up to 12 test cartridges per test) molecular detection via a small cartridge readout system. Thus, this work addresses the scalability limitation of POC molecular testing and can be run in any settings. We verified the analytical sensitivity and specificity of the cartridge testing for respiratory pathogens and sexually transmitted diseases using SARS-CoV-2, influenza A/B RNA samples, and human papillomavirus 16/18 DNA samples. Our cartridge system exhibited a comparable detection performance to the current gold standard qPCR instrument ABI 7500. Moreover, our system showed very high diagnostic accuracy for viral RNA/DNA detection that was well validated by ROC curve analysis. The sample-to-answer molecular testing system reported in this work has the advantages of simplicity, rapidity, and low cost, making it highly promising for prevention and control of infectious diseases in poor-resource areas.

Aptamers are a new kind of nano-probes for bioassays and drug delivery, etc. The good air-tightness is important for pipetting to achieve the purpose of the quantitative liquid distribution in the automated aptamer selection instrument. The pipette tip will be worn and torn after the long-term use, which may result in the leakage, reduction in liquid volume or even pipetting failure. The goal of this study is to constantly detect whether the gas path in the automated aptamer selection instrument leaks during pipetting. The air-tightness detection system is designed according to the change of the air pressure when sucking and discharging liquid. In this study, a hardware system was built to detect air pressure in the gas path of the instrument during the pipetting process, and data processing algorithm was designed to judge if the air pressure curve was normal which included differential processing and detection of extreme points within the threshold. The air-tightness detection model can accurately detect the air-tightness in the pipette during the suction and discharge process to ensure the stable operation of the aptamer selection. This system can also be used for air-tightness detection of other equipment.