Department of Computer Science  ·  University of Cincinnati  ·  Cincinnati, OH
AI Drug Discovery  ·  ISBRA 2025
AI-Powered Immune Checkpoint Inhibitor Discovery via Diffusion Models
De novo pocket-aware peptide generation using twin E(3)-invariant diffusion models — jointly producing 3D structures and amino-acid sequences tailored to checkpoint protein receptor pockets to boost cancer immunotherapy.Po-Yu Liang & Jun Bai  ·  Springer LNBI, ISBRA 2025
Spatial Genomics  ·  arXiv 2601.13331
MultiST: Cross-Attention Multimodal Model for Spatial Transcriptomics
A unified framework jointly modeling spatial topology, gene expression, and H&E tissue morphology via cross-attention fusion. Graph gene encoders with adversarial alignment + GAN–Fisher MMD latent refinement sharpen tumor microenvironment boundary identification.Wei Wang, Quoc-Toan Ly, Chong Yu, Jun Bai  ·  2026
Computational Biology  ·  Steered MD
Molecular Dynamics: vdW Energy Profiles of Protein–Inhibitor Binding
Steered MD simulations track van der Waals energy and center-of-mass distance over 500 ns — revealing the atomistic mechanisms of stable checkpoint inhibitor binding and validating generated peptide designs.
Medical Image Analysis  ·  Breast Cancer
3D Mammogram Cancer Prediction with Graph Neural Networks
DBT slices segmented via SLIC superpixels, connected by KNN into patient graphs, then classified by a self-attention GCN with multi-scale GAP+GMP pooling — achieving state-of-the-art cancer detection on 3D mammography.
Clinical AI  ·  Disease Prognosis
Semi-Supervised COVID-19 ICU Demand Forecasting via CGMNN
Chest radiography autoencoder features fused with structured clinical data into a patient graph, classified by a Markov neural network using EM-based semi-supervised learning — enabling accurate early ICU demand prediction at admission.
Computational Pathology  ·  Gleason Grading
Weakly-Supervised Prostate Cancer Gleason Grading
MIL-Transformer identifies discriminative WSI patches → patch graph → gated-attention GCN classifies Gleason grades without pixel-level annotation. Validated on TCGA-PRAD, PANDA & Gleason 2019 datasets.

Welcome to Jun BMI Computational Lab

We are a research group in the Department of Computer Science at the University of Cincinnati, using AI to push the boundaries of healthcare. Our work spans immune checkpoint inhibitor discovery, spatial transcriptomics, cancer medical image analysis, and molecular simulation of biologics.

Our mission: make smart healthcare more personal, effective, and accessible — by building rigorous computational methods bridging machine learning and biomedical science.

★ Lab News

Jan 2026MultiST preprint on arXiv (2601.13331) — multimodal spatial transcriptomics by Wei Wang & Jun Bai.
Aug 2025Congrats to Wei & Po-Yu! "DeepPH" accepted at ACM BCB 2025.
Aug 2025Congrats to Po-Yu! "E(3)-Invariant Diffusion Model" published at ISBRA 2025.
Jun 2025Congrats to Dang! Received the AI Bearcat Grant for medical image analysis.

👥 Team

Dr. Jun Bai
Principal Investigator
Dr. Jun Bai
Assistant Professor, CS Dept., University of Cincinnati. Ph.D. UConn 2023. AI for drug discovery, medical imaging & health informatics.
Po-Yu Liang
Ph.D. Student
Po-Yu Liang
E(3)-invariant diffusion models for pocket-aware peptide & immune checkpoint inhibitor generation.
Wei Wang
Ph.D. Student
Wei Wang
MultiST — multimodal spatial transcriptomics; multi-omics therapeutic target identification.
Jordan Shaheen
M.S. Student
Jordan Shaheen
RL-based decentralized federated medical image analysis — privacy-preserving diagnostics.
Austin Paulraj
M.S. Student
Austin Paulraj
Graph-based generative AI for diverse animation character design.
Anthony Brambilla
M.S. Student
Anthony Brambilla
Quantum Machine Learning applied to medical image analysis.
Nandini Goud Gadkol
M.S. Student
Nandini Goud Gadkol
Binding affinity estimation for drug-target interaction prediction.
?
Open Position
Recruiting
Interested in joining? Motivated Ph.D. applicants welcome — reach out to Dr. Bai.

🔬 Research

MultiST Spatial Transcriptomics

MultiST: Multimodal Spatial Transcriptomics for Cancer Target Identification

MultiST jointly models spatial topology, gene expression, and H&E histomorphology through cross-attention fusion. Graph gene encoders with adversarial alignment + GAN–Fisher MMD latent refinement achieve sharper tumor microenvironment boundaries and more accurate therapeutic target localization. Wei Wang, Quoc-Toan Ly, Chong Yu, Jun Bai · arXiv 2601.13331 · 2026

ICI Diffusion Model

AI-Powered Immune Checkpoint Inhibitor Discovery via Diffusion Models

Twin E(3)-invariant diffusion models generate pocket-aware peptide structures and amino-acid sequences de novo — leveraging geometric backbone representations and receptor pocket conditioning. Outperforms SOTA on binding success rate and structural similarity. Po-Yu Liang & Jun Bai · ISBRA 2025 (Springer LNBI)

Breast tomosynthesis GNN

3D Mammogram Cancer Prediction with Graph Neural Networks

DBT volumes represented as graphs via SLIC + KNN, then classified by self-attention GCN with multi-scale GAP+GMP pooling — achieving state-of-the-art cancer detection and localization accuracy on 3D mammography screening datasets.

COVID-19 prognosis

Semi-Supervised COVID-19 ICU Demand Forecasting via CGMNN

CGMNN fuses chest radiography autoencoder features with structured clinical data into a patient graph, classified by a Markov neural network using EM-based semi-supervised learning — providing actionable early ICU demand prognosis to optimize resource allocation.

Prostate cancer grading

Weakly-Supervised Prostate Cancer Gleason Grading of Histopathology WSIs

MIL-Transformer extracts discriminative patches from H&E WSIs, constructs a spatial patch graph, and applies gated-attention GCN for Gleason grade classification — no pixel-level annotation needed. Validated on TCGA-PRAD, PANDA & Gleason 2019 datasets.

📄 Publications

Preprints & Under Review
Hierarchical Enzyme Classification with Integrated Sequence and Folding-Derived Representations
Po-Yu Liang, Wei Wang, Jun Bai*
Submitted to ACM BCB 2026
PepEDiff: Out-of-Distribution Sampling Peptide Binder Design via Protein Embedding Diffusion
Po-Yu Liang, Jun Bai*
Submitted to ACM BCB 2026
Uncovering the Mechanism of Hepatotoxicity of PFAS Targeting L-FABP Using GCN and Computational Modeling
Lucas Jividen, Tibo Duran, Xi-Zhi Niu, Jun Bai*
arXiv:2409.10370 · 2024
A Tonic Signaling Code Predicts CAR-T Cell Efficacy in Diffuse Midline Glioma
Emily B Deng, Xiaowen Zhong, ..., Po-Yu Liang, Jun Bai, et al.
bioRxiv · 2025
MultiST: A Cross-Attention-Based Multimodal Model for Spatial Transcriptomics
Wei Wang, Chong Yu, Jun Bai*
bioRxiv · 2026
Refereed Journal Articles
Hybrid Transformer-based Model for Mammogram Classification by Integrating Prior and Current Images
Afsana Ahsan Jeny, Sahand Hamzehei, Annie Jin, Stephen Andrew Baker, Tucker Van Rathe, Jun Bai, Clifford Yang, Sheida Nabavi
Medical Physics · 2025  |  IF: 4.506
Advanced Feature Extraction and Outlier Detection for 3D Biological/Biomedical Image Registration
Sahand Hamzehei, Jun Bai, Gianna Raimondi, Rebecca Tripp, Linnaea Ostroff, Sheida Nabavi
IEEE Transactions on Computational Biology and Bioinformatics · 2025  |  IF: 3.71
Unsupervised Feature Correlation Model to Predict Breast Abnormal Variation Maps in Longitudinal Mammograms
Jun Bai, Annie Jin, Madison Adams, Clifford Yang, Sheida Nabavi
Computerized Medical Imaging and Graphics · Vol. 113 · 102341 · 2024  |  IF: 5.7
Weakly-Supervised Deep Learning Model for Prostate Cancer Diagnosis and Gleason Grading of Histopathology Images
Mohammad Mahdi Behzadi, Mohammad Madani, Hanzhang Wang, Jun Bai, et al.
Biomedical Signal Processing and Control · Vol. 95 · 106351 · 2024  |  IF: 4.9
Molecular Dynamics Modeling Based Investigation of the Effect of Freezing Rate on Lysozyme Stability
Tibo Duran, Bruna Minatovicz, Ryan Bellucci, Jun Bai, Bodhisattwa Chaudhuri
Pharmaceutical Research · Vol. 39(10) · 2585–2596 · 2022  |  IF: 4.914
Combining Multi-view Ensemble and Surrogate Lagrangian Relaxation for Real-Time 3D Biomedical Image Segmentation on the Edge
Shanglin Zhou, Xiaowei Xu, Jun Bai, Mikhail Bragin
Neurocomputing · Vol. 512 · 466–481 · 2022  |  IF: 5.719
Feature Fusion Siamese Network for Breast Cancer Detection Comparing Current and Prior Mammograms
Jun Bai, Annie Jin, Tianyu Wang, Clifford Yang, Sheida Nabavi
Medical Physics · Vol. 49(6) · 3654–3669 · 2022  |  IF: 4.506
Molecular Dynamics Simulation to Uncover the Mechanisms of Protein Instability During Freezing
Tibo Duran, Bruna Minatovicz, Jun Bai, Dongkwan Shin, Hossein Mohammadiarani, Bodhisattwa Chaudhuri
Journal of Pharmaceutical Sciences · Vol. 110(6) · 2457–2471 · 2021  |  IF: 4.064
Single-cell Classification Using Graph Convolutional Networks
Tianyu Wang, Jun Bai, Sheida Nabavi
BMC Bioinformatics · Vol. 22(1) · 1–23 · 2021  |  IF: 4.341
Applying Deep Learning in Digital Breast Tomosynthesis for Automatic Breast Cancer Detection: A Review
Jun Bai, Russell Posner, Tianyu Wang, Clifford Yang, Sheida Nabavi
Medical Image Analysis · Vol. 71 · 102049 · 2021  |  IF: 13.828
Conference Papers
DeepPH: A Multimodal Deep Learning Model for Predicting Enzyme Optimal pH Range
Wei Wang, Po-Yu Liang, Jun Bai*
16th ACM BCB · 2025  |  Acceptance rate: ~19%
Robust Training of Deep Learning Models for Mammogram Classification
Josue Martinez-Martinez, Olivia Brown, Jun Bai, Sheida Nabavi
IEEE 22nd ISBI · 2025  |  Acceptance rate: ~25%
E(3)-Invariant Diffusion Model for Pocket-Aware Peptide Generation
Po-Yu Liang, Jun Bai*
ISBRA 2025 · Springer LNBI vol. 15757 · pp 177–189  |  Acceptance rate: ~19%
Exploring Latent Space for Generating Peptide Analogs Using Protein Language Models
Po-Yu Liang, Xueting Huang, Tibo Duran, Andrew J. Wiemer, Jun Bai*
IEEE BIBM 2024 · pp 842–847  |  Acceptance rate: ~19%
3D Biological/Biomedical Image Registration with Enhanced Feature Extraction and Outlier Detection
Sahand Hamzehei, Jun Bai, Gianna Raimondi, Rebecca Tripp, Linnaea Ostroff, Sheida Nabavi
14th ACM BCB · 2023  |  Acceptance rate: ~19%
Applying Graph Convolution Neural Network in Digital Breast Tomosynthesis for Cancer Classification
Jun Bai, Annie Jin, Andre Jin, Tianyu Wang, Clifford Yang, Sheida Nabavi
13th ACM BCB · 2022  |  Acceptance rate: ~19%
Semi-supervised Classification of Disease Prognosis Using CR Images with Clinical Data Structured Graph
Jun Bai, Bingjun Li, Sheida Nabavi
13th ACM BCB · 2022  |  Acceptance rate: ~19%
A Deep Learning Approach for Ventricular Arrhythmias Classification using Microcontroller
Ya-Sine Agrignan, Shanglin Zhou, Jun Bai, Sheida Nabavi, Caiwen Ding
International Symposium on Quality Electronic Design · 2023
Patent
US 12387477 — Conjoined Twin Network for Breast Cancer Treatment and Analysis
United States Patent

Full list: Google Scholar ↗  |  ResearchGate ↗

🤝 Service

Conference Organization

Editorial Service

Program Committee Membership

Peer Review

🎓 Teaching

Instructor of Record

CS 5137/6037 — Deep Learning · Fall 2023, Fall 2024, Fall 2025

Core fundamentals of deep learning: architectures, optimization, CNNs, RNNs, transformers, and applications in science and engineering.

CS 2023 — Python Programming · Spring 2025, Fall 2025, Spring 2026

Introduction to programming concepts and computational thinking using Python.

Lab Teaching

CSE 1010 — Introduction to Computing for Engineers · Spring 2020, Fall 2021

Computing logic, algorithmic thinking, programming languages, and computing environments.

CSE 1729 — Introduction to Principles of Programming · Spring 2021

Data abstraction, functional abstraction, and problem solving in structured programming.

Teaching Assistant

CSE 4102 — Programming Languages · Fall 2020

Data types, control structures, runtime environments, and programming paradigms.

CSE 3300 — Computer Networks and Data Communication · Fall 2019, Spring 2022

Network types, topology, protocol architecture, routing, and performance.

CSE 2102 — Introduction to Software Engineering · Spring 2021

Software lifecycle, development models, specification, design, and verification.

CPS 542 — Database Management Systems · Fall 2018

E/R modeling, relational theory, and practical DBMS experience.

CPS 562 — Advanced Database Management Systems · Spring 2019

Course Demos & Resources

Computer Vision — Image Processing Teaching Demo (Colab) Python — String Manipulation Teaching Demo (Colab)