London, England, United Kingdom
I am a PhD graduate from Imperial College London with a diverse background in the fields of AI-assisted sensor development, medical devices, signal processing and analysis, and wearable technology. Throughout my academic and professional journey, I have cultivated a unique blend of technical expertise, creative problem-solving skills, and a deep commitment to making a meaningful impact on the future bio/healthcare or general industry. Whether it's developing novel wearable devices/sensors, or implementing advanced engineering or data science skills for cutting-edge technologies, I am passionate about applying my skills and knowledge to drive positive change in the industry. I am continually seeking opportunities to connect with like-minded professionals, explore new ventures, and contribute my expertise to initiatives that align with my values and goals.
Project: Algorithm Design for Optical Signal Processing and Modeling The aim of this project is to design signal post processing algorithm for the analysis of the signals obtained by the self-developed optical fiber sensors to provide a robust and accurate readout of 6 brain biomarkers simultaneously. • Employed multivariate and bayesian inference (regression) models to analyze optical spectra from six sensors, implementing filtering, baseline correction, and environmental compensation to mitigate cross-talk interference. • Developed HMM to inference the hidden states of brain conditions based on the current sensor signals. • Achieved exceptional accuracy (R²≥ 0.95) in predicting biomarker concentrations using CNN-based multitask learning models (PyTorch) and regression models. • Designed and implemented a user-friendly concentration readout interface using Tkinter, and built a prototype system using Nvidia Jetson Xavier for brain monitoring.
• Supervised 9 BSc and MSc student projects, honing mentoring and communication skills. • Worked with students to develop project ideas aligned with their timeframes. • Organised regular meetings to motivate, discuss, and track students’ progress. • Led weekly lectures, reviews, and discussions for a class of 120 freshers (Matlab, Foundation Lab).
Project: Multiplexed Optical Fiber Sensors for Brain Biomarker Analysis The aim of this project is to develop a multiplexed optical fiber bundle sensing system for the monitoring of 6 brain biomarkers simultaneously. The system is miniaturized and has good reversibility, sensitivity, specificity, and stability. • Designed and established an optical sensing setup for portable and multimodal brain monitoring. • Designed and fabricated an optical fiber bundle for multiplexed monitoring of pH, dissolved oxygen, temperature, glucose, and ions with a total diameter of less than 1 mm. • Validated the sensors under simulated disease conditions and optimized sensor fabrication methods for better sensing performance (high sensitivity, selectivity, stability, and reversibility).
Wearable Device for Impaired Gait Prediction • Designed and executed experiments to collect walking IMU data from neurological patients and healthy individuals using wearable IMUs in clinical settings. • Pre-processed signals using butterworth filter and Extended Kalman Filter. Extracted gait and balance features from corrected signals, conducting in-depth statistical analyses to explore the relationship between these features and clinical behaviors. • Developed and fine-tuned prediction models using a combination of deep learning algorithms (CNN/LSTM/CNN+LSTM) and shallow machine learning models (Bayesian&boosting). • Enhanced model algorithms by adopting novel features and models for accurate prediction of impaired gait, proactive fall prevention, and medical guidance. • Conceptualized and constructed a prototype wearable device and validated the system on patients.
• Pre-processed the dataset(6 axes IMU) for signal filtering, denoising and normalization. Extracted features from both time and frequency domains and performed PCA for feature selection. • Tracked hand movement and position using Kalman Filter corrected imu signals for clinal reference. • Built and optimized machine learning models for tremor degree analysis and classification. • Evaluated the model results and their correlations with the extracted features, movement types, and daily tasks using statistical methods for future and clinical evaluation guidance.
Edison Engineering Development Program • Participated in courses and training focused on public speaking, entrepreneurship, and leadership. • Enhanced image qualities using filtering and histogram equalization techniques. Utilized edge detection, region growing, and opening and closing operations for object segmentation. • Utilized and optimized CNN-UNet models (Keras) for metal segmentation through optimizing loss function and hyperparameters. • Achieved a pixel accuracy of 94% by enhancing model performance and convergence speed through transfer learning and backbone techniques.
• Developed valve control software for user interface using C#. • Optimized algorithms for HPLC spectra analysis and chemical concentration calculation(more than 80,000 lines), creating user interface for both Windows and Linux using C/C++. • Designed a website for data readout using JavaScript, PHP, and HTML. • Collaborated with team members using version control systems such as Git to organize modifications and assign tasks.