HANKS OUYANG

About Me

Interest Of Research

Sport Life

Life Science: How TCR-T Kill Cancer Cells

Research Projects

I have been specializing in AI for health care and precision medicine since I worked at a biotech focusing on T cell therapies for tumors. After completing my master program of Information Systems at Hofstra University, I am pursuing MS Artificial Intelligence at Northeastern University with developing applications with multiple AI agents for healthcare. My motivation and adaptability drive my pursuit of advanced opportunities in related fields.

Languages: Python, R, SAS, SQL, JAVA, JS
Skills: Deep Learning, LLMs Fine-tuning & RAG, AI Automation System DevOps, Cloud Computing(AWS)
Framework: PyTorch, LangChain, LangGraph, Unsloth, Pydentic AI, OpenCV, Node.JS
Tools: Android Studio, React Native, n8n, Ollama, Docker, WordPress, Figma, Tableau, Microsoft Project
Certificate: SAS Clinical Trials Programming Accelerated Version
Patent: A new type of temperature-controlled cell incubator

I created below video for my former company, Fineimmu, utilizing AI to succinctly explain the rationale behind TCR-T, a groundbreaking technology in cancer treatment. This underscores the growing significance of AI in the pharmaceutical and biopharmaceutical industries. For instance, AI technologies like machine learning and deep learning aid scientists in deciphering the mutation patterns of proteins. This advancement facilitates TCR's ability to detect and identify cancer cells more efficiently. Although there's still much progress to be made, it's important to remain optimistic about the potential benefits AI can bring to humanity.

Predictive diagnosis for ER(ongoing)

Conducted advanced feature engineering and trained models (DNN, RF, LightGBM, LR, TabTransformer), boosting model with AUC of 98.35% and F1-score 91.91% and created an innovative method for predicting meningitis and optimized machine learning models for imbalanced datasets (Advisor: Dr. Saeed Amal)

Stroke prediction from UK Biobank

Developed a sophisticated neural network model to predict stroke occurrence, achieving 97% accuracy. (Advisor: Jayon Lihm).

Glioblastoma subtyping via machine learning(ongoing)

Applied Ensemble Learning and Transfer Learning to classify Glioblastoma subtypes, achieving an 93.22% accuracy rate in distinguishing between four distinct subtypes.(Advisor: Dr. Shibiao Wan)

Analysis for PD-1 (Programmed cell death protein 1) efficacy

Leveraged machine learning models to learn patterns or representations from graphs with unlabeled data derived from tumor samples. (Advisor: Dr. Haiping Liu)

Health

Conduct machine learning to seek the best solution for healthcare institutes

Precision Medicine

Leverage Artificial Intelligence and Digital Pathology for disease diagnosis and patient care

Pharmaceuticals

Accelerate the discovery of new therapies or drugs via self-evolution or self-learning algorithms, and improves drug research, customizes treatment plans, and mitigates adverse reactions.