Healthcare Ethical AI Lab (HEAL)

Summary

The Healthcare Ethical AI Lab (HEAL) is focused on the development and application of Transparent Algorithms and Ethical AI in Healthcare. Its research spans four main areas: data systems aligned with FAIR and CARE principles, development of ethical artificially intelligent systems, addressing healthcare challenges, and conducting analysis for integrated solutions focusing on health equity.

The research team, led by Mohammad Adibuzzaman, Ph.D., specializes in healthcare data curation, bioinformatics, and bioethics. This specialization supports the lab's goal of facilitating access to clinical data and enhancing the understanding of causal inference. The lab's research activities cover health equity, integration of multimodal data according to FAIR (Findability, Accessibility, Interoperability, Reusability) and CARE (Collective Benefit, Authority to Control, Responsibility, Ethics) principles, and the study of causal inference. These activities include both theoretical exploration and practical application, such as the evaluation of drug interventions with structural causal models.

In its approach to machine learning and artificial intelligence, the lab focuses on the use of interpretable models for Causal Inference to improve transparency in AI-driven decision-making processes, aiming to build trust between scientists, healthcare professionals and patients. Causal inference methodologies are utilized to identify relationships in healthcare outcomes, with counterfactual reasoning models supporting evidence-based decision-making. Efforts to address bias involve the implementation of strategies designed to prevent discriminatory outcomes in diverse patient populations, with a particular focus on the estimation and mitigation of bias.

The lab also emphasizes the importance of data privacy and security, developing data-sharing protocols that facilitate collaborative research while safeguarding patient information. Commitment to informed consent is key to the development of interactive tools designed to enhance patients' understanding of how their data is used and the implications of AI recommendations. Support for bioethics decision-making is integral, incorporating established guidelines and frameworks for ethical considerations in AI algorithms.

HEAL represents a collaborative research environment that combines the expertise of clinicians and computer scientists to promote ethical practices in the application of AI within healthcare.

Selected Publications and Projects

Adibuzzaman, Mohammad, Poching DeLaurentis, Jennifer Hill, and Brian D. Benneyworth. "Big data in healthcare–the promises, challenges and opportunities from a research perspective: A case study with a model database." In AMIA Annual Symposium Proceedings, vol. 2017, p. 384. American Medical Informatics Association, 2017.
Adibuzzaman, Mohammad, Ken Musselman, Alistair Johnson, Paul Brown, Zachary Pitluk, and Ananth Grama. "Closing the data loop: An integrated open access analysis platform for the mimic database." In 2016 Computing in Cardiology Conference (CinC), pp. 137-140. IEEE, 2016.
Adibuzzaman, Mohammad, and Paul M. Griffin. "Big data in health care delivery." (2020).
Gani, Md Osman, Shravan Kethireddy, Riddhiman Adib, Uzma Hasan, Paul Griffin, and Mohammad Adibuzzaman. "Structural causal model with expert augmented knowledge to estimate the effect of oxygen therapy on mortality in the icu." Artificial Intelligence in Medicine 137 (2023): 102493.
Adib, Riddhiman, Md Osman Gani, Sheikh Iqbal Ahamed, and Mohammad Adibuzzaman. "Causal Discovery on the Effect of Antipsychotic Drugs on Delirium Patients in the ICU using Large EHR Dataset." arXiv preprint arXiv:2205.01057 (2022).
Adib, Riddhiman, Paul Griffin, Sheikh Iqbal Ahamed, and Mohammad Adibuzzaman. "A causally formulated hazard ratio estimation through backdoor adjustment on structural causal model." In Machine Learning for Healthcare Conference, pp. 376-396. PMLR, 2020.
Adib, Riddhiman, Md Mobasshir Arshed Naved, Chih-Hao Fang, Md Osman Gani, Ananth Grama, Paul Griffin, Sheikh Iqbal Ahamed, and Mohammad Adibuzzaman. "Ckh: Causal knowledge hierarchy for estimating structural causal models from data and priors." arXiv preprint arXiv:2204.13775 (2022).
Adib, Riddhiman, Sheikh Iqbal Ahamed, and Mohammad Adibuzzaman. "Pragmatic Clinical Trials in the Rubric of Structural Causal Models." arXiv preprint arXiv:2204.13782 (2022).
Chen, Yao, Xiao Wang, Yonghan Jung, Vida Abedi, Ramin Zand, Marvi Bikak, and Mohammad Adibuzzaman. "Classification of short single-lead electrocardiograms (ECGs) for atrial fibrillation detection using piecewise linear spline and XGBoost." Physiological measurement 39, no. 10 (2018): 104006.
Bikak, M., R. Adib, W. Ingram, P. Griffin, and M. Adibuzzaman. "Outcomes of use of antipsychotic for delirium in the icu: A big data approach." In D50. CRITICAL CARE: THE METAMORPHOSIS-PAIN, SEDATION, DELIRIUM, ICU-ACQUIRED WEAKNESS, AND PALLIATIVE CARE, pp. A6672-A6672. American Thoracic Society, 2019.
Fang, Chih-Hao, Vikram Ravindra, Salma Akhter, Mohammad Adibuzzaman, Paul Griffin, Shankar Subramaniam, and Ananth Grama. "Identifying and analyzing sepsis states: A retrospective study on patients with sepsis in ICUs." PLOS Digital Health 1, no. 11 (2022): e0000130.
Patidar, Kavish R., Mobasshir A. Naved, Ananth Grama, Mohammad Adibuzzaman, Arzina Aziz Ali, James E. Slaven, Archita P. Desai et al. "Acute kidney disease is common and associated with poor outcomes in patients with cirrhosis and acute kidney injury." Journal of hepatology 77, no. 1 (2022): 108-115.
Patidar, Kavish R., Mohammad Adibuzzaman, Mobasshir A. Naved, Dylan Rodriquez, James E. Slaven, Ananth Grama, Archita P. Desai et al. "Practice patterns and outcomes associated with intravenous albumin in patients with cirrhosis and acute kidney injury." Liver International 42, no. 1 (2022): 187-198.

Selected Active Projects

Cyber-infrastructure for FAIR principle: Democratizing access to clinical data for research
Democratizing access to clinical data for research: Implementation and evaluation strategies in an Academic Medical Center
Radical Transparency for Healthcare Data: Policies, Technology and Implications
Causal Discovery on the Effect of Antipsychotic Drugs on Delirium Patients in the ICU using Large Observational EHR Dataset
Structural Causal Model with Expert Augmented Knowledge to Estimate the Effect of Oxygen Therapy on Mortality in the ICU
Risk Stratification of Atrial Fibrillation for Patients with Ischemic Stroke: External Validation of the HAVOC Score
Causal Knowledge Hierarchy for Estimating Structural Causal Models from Data with Priors in Clinical Research
Backdoor adjustment of structural causal models in estimation of the Hazard ratio for survival analysis
Tutorial on Fairness Analysis and Societal Bias Mitigation through Causal Inference
Causal Inference for Rare Disease Cohort from N3C

Research Team

Mohammad Adibuzzaman

Dr. Mohammad Adibuzzaman is a computational scientist with a focus on using his expertise to improve human well-being and contribute to a more fulfilling life for people through technology. His research is centered on developing models, methods, and systems for analyzing large-scale clinical data and enhancing research infrastructure with the use of high-performance computing systems. Dr. Adibuzzaman aims to combine technology and extensive data resources to augment clinical expertise, with the goal of influencing the future of medicine.

He is currently an Assistant Professor in the Department of Medical Informatics and Clinical Epidemiology and serves as the Director of the Informatics Program at the Oregon Clinical and Translational Research Institute (OCTRI) at Oregon Health and Science University in Portland, Oregon. His previous roles include Assistant Director of Data and Computing at the Regenstrief Center for Healthcare Engineering (RCHE) at Purdue University, Indiana, following a five-year period as a Research Scientist at the same institution.

Dr. Adibuzzaman obtained his Ph.D. in Computational Sciences from Marquette University in Milwaukee, Wisconsin. His doctoral research involved developing mathematical models using eigenvalue-based methods, such as the eigenface for facial image classification and algorithms for detecting hemorrhage from blood pressure readings based on mixing rates or the second largest eigenvalue. These models have been applied to clinical data at the US Food and Drug Administration (US FDA), with the mentorship of Dr. David Strauss, to enhance the use of high-performance computing systems in medical research.

Prior to his Ph.D., Dr. Adibuzzaman worked as a Junior Research Fellow at the Human Computer Interaction Lab, under Professor Shendong Zhao at the National University of Singapore (NUS), and as a Software Engineer at Structured Data Systems Ltd. (SDSL). He began his academic career with an undergraduate degree in Computer Science and Engineering (CSE) from Bangladesh University of Engineering and Technology (BUET), which set the stage for his contributions to computational science and healthcare innovation.

Riddhiman Adib

Riddhiman Adib (Riddhi) is currently an Sr. Research Engineer at Oregon Clinical & Translational Research Institute (OCTRI) in Oregon Health & Science University (OHSU), Portland, Oregon, U.S.A. He is also a Marquette alumnus, an ex-BUETian, and an ex-cadet.

His research area aims to apply data-driven causal explorations; his vision is to improve public health, health informatics, and overall healthcare research. AI in Healthcare is an evolving and ever-expanding discipline, and he aims to contribute to that with knowledge and ideas he has. Currently, he is actively working in the field of Causal Inference; his primary focus is to apply Causal Inference and related theories in exploring cause-and-effect relationships in experimental and observational datasets.