Research Interest:

I am a postdoctoral researcher at Stanford University, Aghaeepour lab.

I completed my Ph.D. in Information at Florida State University. My Ph.D. was supervised by Dr. Zhe He at eHealth Lab with a research focus on medical informatics. I am mostly interested in predictive analysis using Electronic Health Records (EHRs) with machine learning and specifically deep learning methods. Such tasks include but are not limited to disease progression prediction, readmission prediction, and ICU mortality prediction. One of the most important concerns in this field is the lack of interpretability (comprehensibility) of deep learning methods. Although these methods result in promising performance and accuracy, physicians need to understand and interpret the generated output. On the other hand, the performance of a predictive model depends on the quality of EHR data that the model is built based on. Utilizing EHR data itself is challenging since it is heterogeneous, erroneous, and in the case of clinical notes, unstructured.

  • Ph.D. Dissertation Focus:

Since cardiovascular diseases represent around %31 of all global death (WHO, 2015), developing predictive models in this field seems essential. On average, there are 32.4 million myocardial infarctions yearly, around the world (WHO). My dissertation focus was on explainable AI (XAI). I worked on interpretability enhancement for deep learning models that are used for heart disease prognosis and diagnosis. I work closely with Dr. Zhe He and Dr. Xiuwen Liu From Florida State University and consult Dr. Pablo A. Rengifo-Moreno from the Tallahassee Memorial Hospital.

  • Research Assistantship:

I worked with Dr. Michael Killian and Dr. Zhe He at Florida State University College of Social Work, and College of Communication and Information, and Dr. Dipankar Gupta at the University of Florida Health Congenital Heart Center on an NIH supported grant for "Predictive Analysis of Post-transplant Health Outcomes in Pediatric Organ Transplantation" from the Clinical and Translational Science Institute.

Keywords: Medical Informatics, Deep Learning, Machine Learning, XAI (Explainable AI), Applied AI in Medicine

Publications:

  • Payrovnaziri, S. N., Xing, A., Salman, S., Liu, X., Bian, J., & He, Z. (2021). Assessing the Impact of Imputation on the Interpretations of Prediction Models: A Case Study on Mortality Prediction for Patients with Acute Myocardial Infarction. In AMIA Annual Symposium Proceedings (Vol. 2021, p. 465). American Medical Informatics Association. (PDF)

  • Payrovnaziri, S. N., Chen, Z., Rengifo-Moreno, P., Miller, T., Bian, J., Chen, J. H., ... & He, Z. (2020). Explainable artificial intelligence models using real-world electronic health record data: a systematic scoping review. Journal of the American Medical Informatics Association, 27(7), 1173-1185. (PDF)

  • Payrovnaziri, S. N., Barrett, L. A., Bis, D., Bian, J., & He, Z. (2019). Enhancing prediction models for one-year mortality in patients with acute myocardial infarction and post myocardial infarction syndrome. Studies in health technology and informatics, 264, 273. (PDF)

  • Killian, M. O.†, Payrovnaziri, S. N.†, Gupta, D., Desai, D., & He, Z. (2021). Machine learning-based prediction of health outcomes in pediatric organ transplantation recipients. JAMIA Open, 4(1). (free-access link)

  • Salman, S.†, Payrovnaziri, S. N.†, Liu, X., Rengifo-Moreno, P., & He, Z. (2020, July). DeepConsensus: Consensus-based Interpretable Deep Neural Networks with Application to Mortality Prediction. In 2020 International Joint Conference on Neural Networks (IJCNN) (pp. 1-8). IEEE. (PDF)

  • Barrett, L. A., Payrovnaziri, S. N., Bian, J., & He, Z. (2019). Building computational models to predict one-year mortality in ICU patients with acute myocardial infarction and post myocardial infarction syndrome. AMIA Summits on Translational Science Proceedings, 2019, 407. (PDF)

  • He, Z., Barrett, L. A., Rizvi, R., Payrovnaziri, S. N., & Zhang, R. (2019). Exploring the Discrepancies in Actual and Perceived Benefits of Dietary Supplements Among Obese Patients. Studies in health technology and informatics, 264, 1474–1475. (PDF)

  • He, Z., Barrett, L. A., Rizvi, R., Tang, X., Payrovnaziri, S. N., & Zhang, R. (2020). Assessing the Use and Perception of Dietary Supplements Among Obese Patients with National Health and Nutrition Examination Survey. AMIA Summits on Translational Science Proceedings, 2020, 231. (PDF)

† equal contribution

Previous Research:

  • My MSc thesis “Partially Occluded Pedestrian Detection”, supervised by Professor Reza Safabakhsh, Department of Computer Engineering and Information Technology at Amirkabir University of Technology, Tehran, Iran, (please refer to: http://www.aut.ac.ir/official/main.asp?uid=safa), focused on one of the biggest challenges of computer vision on pedestrian detection, occluded input, to achieve higher accuracy in terms of classifying pedestrians. With the combination of HOG and LBP feature extraction techniques, and applying variety of SVM algorithms including different kernels, we examined the performance and accuracy of the proposed technique on INRIA dataset, while comparing the result to normal SVM and boosting algorithms. We found a close performance and accuracy behavior of our combined features applying SVM Intersection Kernel with HAAR features applying normal SVM algorithm, while HAAR+SVM revealed lower run time, HOGLBP+SVM Intersection revealed lower false positive. The proposed technique and required comparisons was implemented in MATLAB.

  • Supervised by Professor Saeed Shiry (http://ceit.aut.ac.ir/~shiry/ ) as a machine learning graduate course requirement, Department of Computer Engineering and Information Technology at Amirkabir University of Technology, Tehran, Iran, I worked on Dr. Manaf Zargoush (http://www.degroote.mcmaster.ca/profiles/manaf-zargoush/ )’s new boosting algorithm called “A New Boosted Classification Algorithm Based on Dummy Weighting” which was a continuing research on the application to predicting patients’ response to citalopram based on genetic features. I implemented the algorithm in MATLAB which the results got validated and verified by Dr. Zargoush. Therefore, I was recognized for the best machine learning project that semester.