Hi /你好，
Hi /你好，
I'm Degan Hao, Ph.D.
I'm Degan Hao, Ph.D.
a machine learning scientist from Morgan Stanley.
a machine learning scientist from Morgan Stanley.
Embedded Files
I am dedicated to applying trusthworthy artificial intelligence to solve real-world problems. My research efforts are centered on advancing both algorithmic innovation and business applications to shape a better world.
I am dedicated to applying trusthworthy artificial intelligence to solve real-world problems. My research efforts are centered on advancing both algorithmic innovation and business applications to shape a better world.
Email: oliverhaopit A.T. gmail.com
Email: oliverhaopit A.T. gmail.com
Current questions
Current questions
1. Deep learning are vulnerable to adversarial attacks. How to make safer AI for healthcare?
See our recent work on "Adversarially Robust Feature Learning for Breast Cancer Diagnosis"[link]
1. Deep learning are vulnerable to adversarial attacks. How to make safer AI for healthcare?
See our recent work on "Adversarially Robust Feature Learning for Breast Cancer Diagnosis"[link]
2. With the surge of agentic AI, how to ensure AI agents are fulfilling your task as desired?
2. With the surge of agentic AI, how to ensure AI agents are fulfilling your task as desired?
Past questions
Past questions
1.When working in the field of medicine, where labeled data can be expensive, what are the effective approaches for label-efficient machine learning?
1.When working in the field of medicine, where labeled data can be expensive, what are the effective approaches for label-efficient machine learning?
Hao, Degan, et al. "A self-training teacher-student model with an automatic label grader for abdominal skeletal muscle segmentation." Artificial Intelligence in Medicine 132 (2022): 102366. [link]
2.Inaccurate annotations can be common in some medical applications. What are the impact of inaccurate labels and how to identify them for weakly supervised learning?
2.Inaccurate annotations can be common in some medical applications. What are the impact of inaccurate labels and how to identify them for weakly supervised learning?
Hao, Degan, et al. "Inaccurate labels in weakly-supervised deep learning: Automatic identification and correction and their impact on classification performance." IEEE journal of biomedical and health informatics 24.9 (2020): 2701-2710. [link]
3.Censored data often occur during patient follow-ups. How to handle censored data for explainable survival prediction?
3.Censored data often occur during patient follow-ups. How to handle censored data for explainable survival prediction?
Hao, Degan, et al. "Identifying Prognostic Markers From Clinical, Radiomics, and Deep Learning Imaging Features for Gastric Cancer Survival Prediction." Frontiers in oncology 11 (2021): 725889-725889. [link]
And how to integrate multi-modal data for end-to-end survival prediction?
And how to integrate multi-modal data for end-to-end survival prediction?
Hao, Degan, et al. "SurvivalCNN: A deep learning-based method for gastric cancer survival prediction using radiological imaging data and clinicopathological variables." Artificial Intelligence in Medicine (2022): 102424. [link]
Can we enhance the accuracy and explainability of survival prediction with medical knowledge?
Can we enhance the accuracy and explainability of survival prediction with medical knowledge?
Hao, Degan, et al. (2024). Medical Knowledge-Enabled Multi-Task Learning for Gastric Cancer Survival Prediction. International Symposium on Biomedical Imaging, Athens, Greece, 2024.
4. Humans have lots of specialties that AI struggles to learn. For example, can we teach radiologists' reading behavior to AI?
4. Humans have lots of specialties that AI struggles to learn. For example, can we teach radiologists' reading behavior to AI?
Hao, Degan, et al. (2022). Incorporate radiograph-reading behavior and knowledge into deep reinforcement learning for lesion localization. Medical Imaging 2022: Computer-Aided Diagnosis, 12033, 244-249. [link]
In another example, can we simulate doctor's attention mechanism using AI?
In another example, can we simulate doctor's attention mechanism using AI?
Hao, Degan, et al. (2023). Predicting Outcomes in Long COVID Patients with Spatiotemporal Attention. Eleventh IEEE International Conference on Healthcare Informatics (ICHI). 2023 IEEE 11th International Conference on Healthcare Informatics (ICHI), Houston, TX, USA, 2023 pp. 162-167. [link]
5. AI safety. How to protect medical image diagnosis from adversarial attacks?
5. AI safety. How to protect medical image diagnosis from adversarial attacks?
Arxiv. [link]
Get in touch at d.hao@pitt.edu
Get in touch at d.hao@pitt.edu
Page updated
Google Sites
Report abuse