PRML Lab · Morgan State University

Pattern Recognition & Machine Learning Laboratory

Advancing the foundations of pattern recognition and trustworthy machine learning, with applications in cardiovascular health, epidemiology, and cryptography

Focus Areas

The lab develops foundational methods in pattern recognition and machine learning and applies them in three areas.

Cardiovascular Health

Multimodal ECG and PPG analysis, cardiovascular stability, cuffless blood pressure estimation, and explainable, trustworthy models for cardiac signals, including the NIH-funded ORBIL-AI project.

Frameworks and status

Epidemiology & Public Health

Machine learning for population-health data: COVID-19 fatality classification and forecasting, ECG-based COVID-19 prediction, and natural language processing of public sentiment and vaccine hesitancy during the COVID-19 and mpox outbreaks.

NLP & Health AI publications

Cryptography & Data Security

Secure and privacy-preserving computation for data-driven systems, building on the group’s earlier work on homomorphic encryption and data security in the cloud. This is a developing direction of the lab.

Related publication

Lab Mission & Vision

Lab Mission

The mission of the Oladunni Lab is to advance the mathematical and computational foundations of pattern recognition, machine learning, and trustworthy artificial intelligence, with applications in cardiovascular health, epidemiology, and cryptography. We seek to discover fundamental principles governing how intelligent systems learn, represent, integrate, and interpret complex information. Through rigorous theoretical development, computational innovation, and experimental validation, we aim to establish generalizable scientific knowledge that enables reliable, interpretable, and equitable AI systems for biomedical applications.

We are committed to scientific excellence, original research, interdisciplinary collaboration, and the development of future researchers. Our mission extends beyond technological applications to advancing fundamental knowledge through high-impact publications, reproducible research, and meaningful contributions to computer science. We provide students with opportunities to engage in theoretical inquiry, mathematical modeling, algorithm development, and independent research, preparing them to become innovative, ethical, and intellectually independent scientists. As a research laboratory at a historically Black university, we are dedicated to strengthening research capacity, expanding opportunities for scientific discovery, and developing the next generation of leaders in computing and artificial intelligence.

Lab Overview

The Oladunni Lab at Morgan State University conducts foundational, theory-driven research in pattern recognition, machine learning, and trustworthy artificial intelligence, with applications in cardiovascular health, epidemiology, and cryptography. Our research seeks to uncover the mathematical and computational principles governing how information, including physiological information, is represented, preserved, integrated, and interpreted by intelligent systems. We develop and investigate nine complementary theoretical frameworks: Cardiac Stability Theory (CST), Complementary Feature Domains (CFD), Explainable-AI Trustworthiness (EAT), Physiologic Invariance Theory (PIT), Physiologic Energy Conservation Theory (PECT), Complementarity-Preserving Generative Theory (CPGT), Attractor Domain Theory (ADT), Attractor-Vascular Coupling Theory (AVCT), and Physiologic Trajectory Theory (PTT). These research directions address fundamental questions in representation learning, information theory, dynamical systems, physiological invariance, and computational trustworthiness, providing a theoretical foundation for advancing biomedical machine learning.

Our research translates these foundational principles into computational methods for cardiovascular monitoring, cuffless blood pressure estimation, physiological signal interpretation, and equitable wearable health technologies. By integrating mathematical modeling, deep learning, signal processing, and rigorous experimental validation, we investigate how physiological measurements can support reliable and interpretable predictions across individuals, sensing conditions, and demographic populations. The lab emphasizes reproducible research, theoretical analysis, and systematic evaluation, with the broader goal of establishing generalizable principles for trustworthy physiological intelligence. Through interdisciplinary collaboration and student mentorship, we aim to advance fundamental computer science research while developing computational technologies that address important challenges in health, epidemiology, and data security.

Current Research Projects

ORBIL-AI: Orthogonality-Regularized Biomedical Information-theoretic Learning

NIH R16 SuRE (National Library of Medicine), award 1R16LM015384-01 · PI: Dr. Timothy Oladunni · Students: Farouk Ganiyu Adewumi, Raymond Mbam, Oluwafemi Ifesanmi, Daniel Owolabi, Obaloluwa Arulogun, Aditya Raj Poudel

Investigates how learning systems represent the information shared across multiple biomedical observations and the information distinct to each. Current student work examines whether models that combine time, frequency, and time–frequency ECG representations learn complementary or largely redundant information, using the PTB-XL and Mendeley ECG datasets and a research plan specified in advance. Project page →

Attractor-Based Analysis of Cardiovascular Signals

Frameworks: CST, ADT, AVCT · Student: Farouk Ganiyu Adewumi

Develops and tests dynamical-systems descriptions of PPG and ECG signals, including stability indices, the partition of attractor information into domains, and calibrated cuffless blood pressure estimation. Results to date are reported in 2026 preprints; see the Research page for the status of each.

Concept Drift and Stability in Multimodal Physiological Signals

Frameworks: PECT, PIT · Student: Farouk Ganiyu Adewumi

Studies when a model’s prediction should change under physiological variation, using energy-based drift criteria and explainability methods such as SHAP and LIME. Work in 2026 developed a CFD-guided framework combining ECG, PPG, and respiration (PECS), reported in a 2026 preprint.

Lab Members

Lab Director

Graduate Researchers

Undergraduate Researchers

Alumni & Past Student Projects

Historical student projects. Select a name to read the project summary.

Student Opportunities

Join Our Lab

The Oladunni Lab provides hands-on research experiences in machine learning, biomedical AI, epidemiology, data security, and applied data science. Students work on real datasets, contribute to ongoing publications, and build practical skills that prepare them for graduate study and industry careers.

What We Look For

Curiosity & Initiative

Genuine interest in AI, healthcare, or signal processing

Technical Foundation

Solid programming skills in Python, experience with ML libraries

Discipline & Reliability

Commitment to rigorous research and meeting deadlines

Communication Skills

Ability to articulate ideas clearly in writing and presentations

What You'll Gain

Mentorship from experienced researchers

Hands-on experience with current machine learning methods and rigorous evaluation practice

Opportunity to co-author peer-reviewed publications

Experience with public and research biomedical datasets and reproducible research workflows

Professional development in academic writing and presentation

Interested? Contact Dr. Oladunni with your resume, CV, and a statement of research interests.

Research Areas for Students

Cardiac Stability Theory (CST) & Multimodal Signal Analysis

Implement and test the proposed Cardiac Stability Index (CSI) on ECG and PPG signals using dynamical systems methods. Learn Lyapunov exponents, recurrence analysis, and entropy computation, and study where these measures are and are not reliable.

Skills: Signal processing, nonlinear dynamics, Python/PyTorch

Complementary Feature Domains (CFD) & Multimodal Fusion

Design deep learning architectures that preserve complementarity across time, frequency, and time-frequency domains. Explore fusion strategies (early, late, hybrid) that enhance robustness without redundancy.

Skills: Signal representations, neural architecture design, TensorFlow/PyTorch

Attractor Domain Theory (ADT) & Cuffless Blood Pressure (AVCT)

Study how cardiac attractor geometry relates to hemodynamic measures. Build and evaluate PPG-based cuffless blood pressure estimators using the proposed single-calibration AVCT approach, on clinical datasets and, as data become available, wearable recordings.

Skills: PPG processing, machine learning, mobile app integration, validation study design

Physiologic Trajectory Theory (PTT) & Longitudinal Monitoring

Develop Kalman state-space models for per-individual cardiac trajectory monitoring. Implement CUSUM drift detection to distinguish benign excursions from true physiologic changes in longitudinal data.

Skills: Kalman filtering, state-space modeling, time series analysis

Explainable AI & Trustworthiness (EAT)

Study interpretability methods for cardiac AI models, including saliency maps, attribution analysis, and uncertainty quantification. Test whether model explanations align with clinical reasoning and physiologic principles.

Skills: Model interpretability, visualization, clinical domain knowledge

Fairness, Equity & Demographic Validation

Evaluate PPG-based algorithms across Fitzpatrick skin types I–VI. Contribute to a planned longitudinal smartphone PPG study (IRB protocol submitted). Study demographic bias in wearable health monitoring.

Skills: Fairness metrics, bias detection, IRB protocols, diverse dataset collection

Lab Values & Working Philosophy

Rigor & Excellence

We state assumptions explicitly, plan analyses in advance, and hold our experimental design and publication ethics to a high standard, so that results withstand independent scrutiny.

Collaboration

We work together across disciplines, institutions, and perspectives. We value diverse viewpoints and collaborative problem-solving.

Reproducibility

We document data, code, and evaluation protocols, guard against leakage and other sources of inflated performance, and report limitations and negative results.

Real-World Impact

We study problems that matter for human health, and we are careful to distinguish research findings from validated clinical practice.

Transparency & Interpretability

We believe AI systems should be understandable and trustworthy. We develop interpretable methods and communicate findings clearly to diverse audiences.

Ethical Practice

We conduct research ethically, respect data privacy, acknowledge limitations, and consider broader societal implications of our work.