Hi, I'm James Leo
I'm a Machine Learning Researcher/Cheminformatics Scientist at Centre for Eye and Vision Research (CEVR, Hong Kong Science Park.
👋 These are the projects I have completed and currently working on at CEVR.
June 2021 -Dec 2021
Project-1 -Using Machine Learning To Predict Partition Coefficient (Log P) and Distribution Coefficient (Log D) with Molecular Descriptors and Liquid Chromatography Retention Time
- Curated datasets from a variety of public databases, like PubChem and CheMBL
- Developed a machine learning models and deep learning models to rank molecules for the drug candidate.
- Patented the innovation at U.S. Patent and Trademark Office (USPTO) and CNIP.
- Published in American Chemical Society (ACS) recently. More details can be found in here https://pubs.acs.org/doi/10.1021/acs.jcim.2c01373
- GitHub Repository: https://github.com/jamesleocodes/p_chem_CEVR
On-going
Project-1.1 – Developing iMoleQ which is a cloud-based web application designed to predict molecular properties.
- The frontend development of iMoleQ has been completed, and you can access it at www.imoleq.com.
- Currently, we are in the process of integrating the backend with the API and server-side components.
- GitHub Repository: https://github.com/jamesleocodes/iMoleQ?tab=readme-ov-file
Jan 2022 -Dec 2022
Project-2 - AI-Enabled Diagnosis of Parkinson’s Disease from Eye Movement Data
- Fitted with analytical wavefunction to simplify the eye movement data.
- Developed and implemented signal processing algorithms to extract meaningful features from raw eye movement data (i.e. in waveforms).
- Built and fine-tuned machine learning models to differentiate Parkinson’s disease patients from healthy controls using eye movement data.
- The innovation from this research is preparing for patent to submit to the U.S. Patent and Trademark Office (USPTO).
- GitHub Repository: https://github.com/jamesleocodes/eye_movement
Jan 2023 -Aug 2023
Project-3- Diagnosing Diabetic Retinopathy: A Machine Learning Approach to CIL LC-MS Analysis of Tear Sample Metabolites
- Implemented a project on biomarker discovery for diabetic retinopathy disease using metabolomic data processing and machine learning.
- Managed data cleanup, feature engineering, and utilized tree-based algorithms (Gradient Boosting, XGBoost, Random Forest) for accurate disease classification, with potential for real-world clinical diagnostic application. The paper related to the research will be published in American Chemical Society (ACS) soon.
- GitHub link: https://github.com/jamesleocodes/Diabetic_Retinopathy_ML
Sep 2023 -Present
Project-4 Machine Learning-Assisted Identification of Potential Metabolite Biomarkers for Glaucoma Diagnosis through Serum Metabolomic Analysis
- Developed machine learning models to identify biomarkers for glaucoma from metabolomic data, enhancing diagnostic accuracy.
- Data acquired from analyzing serum samples of participants using high-resolution mass spectrometry.
- Utilized Random Forest, Gradient Boosting, and XGBoost algorithms, achieving up to 0.87 accuracy.
- Identified significant metabolite linked to glaucoma through SHAP analysis.
- The paper related to the research will be published in American Chemical Society (ACS) soon.
- GitHub link: https://github.com/jamesleocodes/glauSerum
👀 I’m interested in coding(code for fun) with python,R, C/C++, SQL and matlab.