Yidong Zhu

Ph.D. in Computer Science

Healthcare AI, wearable sensing, and clinical data

Download Resume

About

I hold a Ph.D. in Computer Science from University of Massachusetts Lowell (May 2026). My research area is a dynamic intersection of deep learning, mobile and wearable computing, cloud computing, and distributed systems, all aimed at developing innovative, community-engaged solutions. My work is centered on leveraging scalable deep learning and fair machine learning techniques to integrate AI ethics into prediction models, fostering a community-trust loop that enhances the feasibility and sustainability of citizen science frameworks.

Recent Publications and Conference Papers

  • Sultana, S., Zhu, Y., Liu, S.-H., & Alam, M. A. U. (2024). "Wearable-Based Fair and Accurate Pain Assessment Using Multi-Attribute Fairness Loss in Convolutional Neural Networks." 21st EAI International Conference on Mobile and Ubiquitous Systems: Computing, Networking and Services (MobiQuitous 2024), Oslo, Norway, November 12-14, 2024.
  • Zhu, Y. (2024). "Enhancing Wearable based Real-Time Glucose Monitoring via Phasic Image Representation Learning based Deep Learning." 46th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC), Florida, US.
  • Zhu, Y. (2023). "Augmenting Deep Learning Adaptation for Wearable Sensor Data through Combined Temporal-Frequency Image Encoding." IEEE-EMBS International Conference on Body Sensor Networks.

Background

Education

Doctor of Philosophy, Computer Science

UML
2020 – May 2026

University of Massachusetts Lowell, Lowell, MA, USA

  • Research Focus: Deep Learning, Mobile and Wearable Computing, Cloud Computing, and Distributed Systems

Master in Analytics

Northeastern University
2018 – 2019

Northeastern University, Boston, MA, USA

  • Specialization: Data Analytics and Machine Learning
  • Focus: Statistical analysis, predictive modeling, and data visualization techniques

Master in Health Informatics

UML
2014 – 2017

University of Massachusetts Lowell, Lowell, MA, USA

  • Focus: Healthcare data management, clinical information systems, and health data analytics

Academic & Professional Experience

Teaching Assistant

UML
September 2021 – May 2026

University of Massachusetts Lowell

  • Assist in grading assignments, quizzes, exams, and projects. Provide constructive feedback to help students improve their understanding and performance.
  • Hold regular office hours to provide additional support to students, answer questions, and clarify course material.
  • Prepare lab materials, deliver lab lectures and supervise lab sessions.

Data Analyst

2016 – 2019

Various Organizations

  • Puma Group (Sep 2019 – Dec 2019): Data Analyst for sponsored school project
  • Boston Software Group (Jul 2019 – Sep 2019): Data Analyst position
  • Viacom (Apr 2019 – Jun 2019): Data Analyst for sponsored school project
  • Southern New Hampshire Health System (Nov 2016 – Mar 2017): Clinical Informatics Intern

Research Projects

Research on clinical AI, wearable sensing, community-engaged systems, and machine learning.

Opioid Use Disorder (OUD) Risk Prediction with Wearable and EHR Data

Interdisciplinary Research & Workflow Design | All of Us Researcher Workbench

Overview: This research focuses on developing a predictive framework to identify the risk of Opioid Use Disorder (OUD) onset and progression. By leveraging the All of Us Researcher Workbench, I integrate high-frequency longitudinal wearable data, including Fitbit heart rate and step data, with structured Electronic Health Records (EHR) to build robust machine learning models that can detect early physiological and clinical indicators of OUD.

Machine Learning & Predictive Modeling
  • Multimodal Integration: Developing models that fuse longitudinal wearable signals with diverse clinical features including medications, diagnoses, and demographics
  • Temporal Pattern Recognition: Identifying specific heart rate and activity patterns associated with OUD onset through deep learning workflows
  • LLM-based Encoding: Exploring Large Language Model (LLM) approaches to transform unstructured clinical narratives and complex EHR histories into high-dimensional embeddings for improved prediction
Data Pipeline & Cloud Infrastructure
  • Reproducible Workflows: Implementing secure, end-to-end data pipelines using Python and SQL within the All of Us cloud environment
  • Privacy-Preserving Analysis: Adhering to strict institutional policies for secure, in-platform data processing and aggregate-only data exportation
  • Scalable Architecture: Utilizing the All of Us Workbench to handle large-scale datasets while ensuring computational efficiency and research reproducibility

Research Impact: This project bridges the gap between traditional clinical informatics and modern wearable computing. By providing a more nuanced, real-time understanding of patient behavior and physiology, this work aims to support earlier intervention strategies and more personalized care pathways for individuals at risk of opioid-related complications.

NSF Smart & Connected Communities for Cleanwater

2022 - 2026 | NSF Award #2230180

Faculty Advisors: Dr. Pradeep U. Kurup (Distinguished University Professor, Francis College of Engineering) & Dr. Mohammad Arif Ul Alam (Assistant Professor, Kennedy College of Sciences)

Portable E-Tongue Based Distributed Computing Platform for Community-Engaged Water Contamination Prediction

Overview: Led the development of an integrated system combining a portable electronic tongue (E-Tongue) device with a secure, cloud-based distributed computing platform. This solution empowers both community members and researchers to monitor water quality in real-time through user-friendly software, cloud infrastructure, machine learning, and advanced data visualization.

🔧 System Development & Integration
  • Integrated a portable E-Tongue with a distributed cloud computing backend for real-time water quality prediction
  • Designed a secure, cloud-connected hardware-software interface
  • Developed a comprehensive, interactive dashboard for monitoring contamination data across communities
💻 Software & Application Development
  • Built cross-platform software enabling the E-Tongue device to conduct on-site water testing
  • Community App: Simplified UI with EPA-based, color-coded safety indicators (green: safe, red: unsafe)
  • Researcher App: Advanced features including sensor diagnostics, detailed plots, and system feedback
☁️ Cloud Infrastructure & Data Management
  • AWS Serverless Architecture: DynamoDB & S3 for real-time storage, SES for automated communications
  • Authentication Interface: Secure researcher portal for lab result uploads and automated participant updates
  • Address Normalization: Python-based system with 85.5% match threshold for accurate data linking
🌍 Data Visualization & Geocoding
  • ArcGIS Integration: Public dashboard with real-time spatial data syncing
  • Interactive Mapping: Color-coded visualization of water quality metrics (lead, copper, conductivity, pH)
  • Safety Monitoring: Automated threshold violation alerts and community notifications
🤖 Machine Learning & Predictive Modeling
  • Developed models to predict lead levels using Maximum Mean Discrepancy (MMD) feature extraction
  • Applied SMOTE for dataset balancing and improved model performance
  • Compared distance metrics (MMD, KL, JS divergences) across Random Forest and XGBoost models

Impact: This transdisciplinary project collaborates with community stakeholders across socio-economically diverse Massachusetts communities, addressing critical water contamination issues including lead, arsenic, copper, and PFAS through innovative IoT-enabled monitoring systems.

Smart Infant: Monitoring Tummy Time

Prototype Development & Pilot Data Collection | UMass Lowell

Faculty Advisors: Dr. Ainat Koren (Professor, Zuckerberg College of Health Sciences) & Dr. Mohammad Arif Ul Alam (Assistant Professor, Kennedy College of Sciences)

Overview: This interdisciplinary project addresses the early prevention of childhood obesity by developing a wearable system to monitor infant physical activity, with a focus on tummy time. The project is currently in the prototype testing and data collection phase. I am the lead technical contributor responsible for building the sensor-based wearable system and developing the mobile application used for pilot data collection.

👕 Sensor-Integrated Wearable System
  • Designed and built a wearable prototype by embedding a lightweight motion sensor into a soft, infant-safe cotton onesie
  • Focused on achieving sensor stability, comfort, and wearability for infants at 1, 2, and 4 months of age
  • Conducted hardware testing to ensure reliable data logging during up to 10 hours of daytime use
📱 Mobile Application Development
  • Developed a custom smartphone app to serve as the main interface for caregivers and research staff
  • Bluetooth Integration: Collect and store raw motion data from the wearable sensor
  • Caregiver UI: Simple interface enabling parents to monitor activity sessions and complete in-app surveys
  • Research Support: Built-in logging for timestamps and metadata essential to research protocol
🔬 Role in Pilot Deployment
  • Played a central role in preparing and supporting initial field deployments for pilot data collection
  • Created documentation and provided technical assistance to caregivers for correct system usage
  • Logged and debugged system performance across multiple test sessions to inform prototype iterations
📈 Future Work & Impact
  • Algorithm Development: Developing posture classification algorithms (prone vs. non-prone) from motion data
  • Automated Tracking: Enable real-time tummy time duration monitoring
  • Health Assessment: Evaluate impact on early motor development and healthy weight gain
  • Early Prevention: Contributing to childhood obesity prevention through evidence-based interventions

Technical Innovation: This project combines wearable technology, mobile app development, and machine learning to create a comprehensive infant health monitoring solution that bridges the gap between research and practical parental care.

Sleep App Research: Investigating Language Style and Habit Formation

2023-2025 | UMass Lowell

Faculty Advisors: Dr. Ann Kronrod (Associate Professor, Manning School of Business) & Dr. Mohammad Arif Ul Alam (Assistant Professor, Kennedy College of Sciences)

Overview: This research project, led in collaboration with Professor Ann Kronrod, investigates how figurative vs. literal language in mobile app design affects user engagement and habit formation. The broader goal is to determine whether the use of metaphorical and expressive language in sleep-related apps can promote healthier sleep behaviors—an innovative angle on behavioral health interventions. I am the lead technical contributor, responsible for the development of the app infrastructure, participant data flow, and system integrity.

📱 App Development
  • Figurative Version: Uses metaphorical and playful language (e.g., "Time to tuck in!")
  • Literal Version: Uses straightforward, instructional prompts
  • Designed both versions to support real-time data logging and survey collection
  • Implemented A/B testing framework to compare language style effectiveness
☁️ Backend & Data Pipeline
  • AWS Cognito: Secure user authentication and management
  • Lambda Functions: Serverless event processing and data validation
  • DynamoDB: Scalable storage for sleep logs and survey responses
  • Data Flow: Seamless collection of daily logs, pre-surveys, and post-surveys
📊 Data Monitoring & Analysis
  • Developed tools for tracking participant engagement across two groups:
  • Figurative app users
  • Literal app users
  • Prepared datasets for final study analysis to determine language-style effects on sleep habits
🛡️ Fraud Prevention & Data Integrity
  • Implemented fraud detection feature to flag irregular or duplicate entries
  • Ensured high-quality, trustworthy participant data throughout the pilot study
  • Developed automated data validation and quality assurance protocols

Research Impact: This interdisciplinary collaboration between computer science and business psychology explores novel approaches to behavioral health interventions, potentially informing the design of more effective health apps through evidence-based language strategies.

Side Projects

MassUniHub — Verified Student Marketplace

Side project, 2025–2026 · Live

A bilingual (EN/中文) marketplace where Massachusetts college students buy and sell used items, gated by .edu email verification. Listing and contact only: no payments, escrow, or delivery. Built through an AI-agent workflow (Claude Code, Cursor, a custom agent): I owned product scope, data-model decisions, test-first requirements and PR review across 36 merged PRs.

  • Privacy by design: public listing endpoints were returning every seller's contact ID. Contact details now sit behind a login- and .edu-gated endpoint, rate-limited to 30 requests per user per hour, and seller email is no longer included in message threads.
  • Trust and safety: report queue, listing takedown or restore with a reason shown to the seller, account suspension that revokes existing sessions immediately, and an append-only admin audit log.
  • Production hardening: server-side image validation and WebP re-encoding (sharp to Cloudflare R2), CSP and HSTS headers, throttled cryptographic verification codes, CI with 47 test files, and 0 production npm vulnerabilities.
  • Performance and accessibility: in a local load test (1,000 listings, 50 concurrent users), p99 was 76 ms. Lighthouse accessibility went from 87/89 to 100/98/100.

Stack: React · TypeScript · Tailwind · Node/Express · MongoDB · Cloudflare R2 · Railway · GitHub Actions

Skills & Technical Expertise

A comprehensive overview of technical skills, programming languages, and tools that support my research and professional work.

  • Languages: English and Chinese (Fluent)
  • Programming & Development: SQL, Python, R, Software development (Xamarin, Flutter)
  • Data Analytics & Visualization: Tableau, MS Office, Statistical analysis and modeling
  • Cloud & Infrastructure: AWS Cloud services, distributed systems
  • Specialized Software: Website design and management, ArcGIS
  • AI & Development Tools: AI IDE (Cursor, Trae), machine learning frameworks
  • Research Areas: Deep Learning, Mobile Computing, Wearable Computing, Health Informatics

Contact

I am looking for Healthcare AI, Clinical Data Scientist, and ML Engineer roles in Greater Boston. Feel free to reach out.