Welcome!
I’m Andy Tai, a researcher and educator working at the intersection of machine learning, public health, and data science education. As a Postdoctoral Teaching and Learning Fellow at the University of British Columbia’s Master of Data Science Program, I develop clinical decision support systems to address critical healthcare challenges while training the next generation of data scientists.
Research: Machine Learning for Vulnerable Populations
My research focuses on developing trustworthy AI systems for clinical decision support, particularly addressing the drug toxicity crisis and mental health outcomes in vulnerable populations. I specialize in translating complex machine learning models into practical tools that healthcare providers can use to improve patient outcomes.
Risk Assessment and Management Platform (RAMP): As Co-Investigator on this Health Canada-funded project ($1.4M CAD), I developed machine learning models for overdose risk prediction using the BC Provincial Overdose Cohort (36,679 cases, 2015-2019). My Random Forest and XGBoost models achieved 88.77% accuracy and 91.12% AUROC for general overdose prediction, analyzing 48 clinical features including medical conditions, substance use patterns, mental health history, and infectious complications. RAMP is currently in development and testing phase, with supervised implementation continuing under my guidance.
During my PhD in Neuroscience (completed April 2024), I conducted systematic reviews and meta-analyses of machine learning models for opioid-related outcomes, establishing the methodological foundations for this work. My dissertation, “A machine learning approach to overdose risk assessment,” bridges computational methods with clinical applications in addiction psychiatry.
This work demonstrates how data-driven approaches can transform treatment personalization and prevention strategies for populations facing complex health challenges. My research prioritizes not just predictive accuracy, but the ethical deployment of AI systems in real-world healthcare settings.
AI Applications Across Domains
I bring computational expertise to diverse problems requiring sophisticated data analysis:
Maritime Security: Consulting with Clause Technology, I developed vessel trajectory analysis systems using machine learning to detect illegal fishing activities and maritime criminal behavior, demonstrating the transferability of my methods beyond healthcare contexts.
Venture Analytics: Through Creative Destruction Lab, I applied predictive modeling to startup success assessment, helping investors make data-informed decisions about early-stage technology ventures.
Concussion Management: At Concussion RX, I developed AI systems analyzing patient data in real-time to provide personalized treatment recommendations, improving diagnosis accuracy and recovery outcomes.
Healthcare Communication: My work with Aggregate Intellect explored how large language models can transform audio transcripts into accessible health information, enhancing patient education and clinical communication efficiency.
Education: Training Data Scientists for Impact
Teaching is fundamental to my mission. I’ve taught over 550 graduate students across 10+ courses in UBC’s Master of Data Science and Graduate Neuroscience programs, earning multiple teaching awards including the MDS Teaching Assistant Award and Graduate Student Teaching Assistant Award.
Core Curriculum: DSCI 100 (Introduction to Data Science), DSCI 521 (Computing Platforms), DSCI 513 (Databases and Data Retrieval), DSCI 542 (Communication and Argumentation), STAT 302 (Probability and Statistics)
Specialized Methods: DSCI 531 (Data Visualization), DSCI 551 (Descriptive Statistics and Probability), SCIE 113 & 300 (Science Communication)
Applied Practice: DSCI 591 (Capstone Projects with industry partners including BC Cancer Agency, Terramera, and Best Buy Canada)
My teaching philosophy emphasizes hands-on learning, ethical AI deployment, and bridging theoretical knowledge with real-world application. I’ve co-authored “The Regression Cookbook: Machine Learning and Stats Flavours” and integrate innovative pedagogical approaches including labor-based grading systems and GenAI frameworks for data science education. Students learn not just algorithms, but how to deploy them responsibly in contexts affecting human lives.
Future Directions
My research trajectory expands into neurobiologically-inspired reward models for LLM fine-tuning, bridging my neuroscience background with current AI expertise. I’m also developing causal inference methodologies and real-time multimodal data integration approaches, moving beyond traditional machine learning to capture the complexity of clinical decision-making processes.
Bridging Research and Practice
What distinguishes my work is the commitment to operational implementation. From conducting rigorous statistical analyses through developing production-ready applications, I ensure machine learning models become practical tools that healthcare providers and decision-makers can use immediately. This end-to-end approach, spanning data preparation, model development, ethical validation, and web deployment, represents essential practice in applied machine learning for high-stakes domains.
I’ve secured substantial funding (over $1.4M CAD) and built collaborative networks spanning clinical researchers, policymakers, and industry partners. My interdisciplinary background in neuroscience, statistics, and computer science enables me to address complex problems requiring both technical sophistication and domain expertise.
This website showcases my scholarly contributions, interactive tools, and ongoing projects. Explore my work, test the live applications, and connect with me to discuss potential collaborations in healthcare AI, data science education, or clinical decision support systems.
Contact me to discuss projects, collaborations, or speaking opportunities.