research

My research fall under three main categories:

  1. Machine Learning with limited supervision and knowledge integration
  2. Machine Learning with multi-source/multi-modal data
  3. Data mining and subgroup identification

Application domains of my research include:

Precision Medicine of Brain Cancer.

This research aims to develop personalized machine learning models to predict regional tumor cell density from MRI. My contributions included developing a novel weakly-supervised transfer learning (WS-TL) model that leverages domain knowledge about the tumor, which addresses the small sample problem while quantifying intratumoral heterogeneity; and building a GUI for non-technical users to run pre-operative and post-operative model to assist cancer treatment.

This research is in collaboration with Drs. Kristin Swanson and Leland Hu at Mayo Clinic.

Publications:

  • Mao L, Wang H, Li J. Knowledge-informed machine learning for cancer prognosis and predictions: a review. (major revision at IEEE T-ASE). (website) (paper)
  • Mao L, Wang L, Hu L, Eschbacher J, Leon GD, Singleton K, Curtin W, Urcuyo A, Sereduk J, Tran L, Hawkins A, Swanson K, Li J. Weakly supervised transfer learning with application in precision medicine. IEEE Transactions on Automation Science and Engineering. doi:10.1109/TASE.2023.3323773 (paper) (🏆 Best Student Paper, IISE DAIS 2022)
Biomarker Discovery for Persistent Post-traumatic Headache

This research develops predictive machine learning algorithms for the prognosis of post-traumatic headache based on neuroimaging (MRI, fMRI, DTI), clinical questionnaires, and mobile-collected speech data. My contributions include: 1) developing prognostic models of headache persistence, 2) identifying patient subgroups with distinct headache evolution trajectories, and 3) developing a clinical tool for prognosis of headache persistence adaptable for patients with different data modalities.

This research is in collaboration with Drs. Catherine D. Chong and Todd Schwedt at Mayo Clinic.

Publications:

  • Mao L, Li J, Schwedt TJ, Berisha V, Nikjou D, Wu T, Dumkrieger GM, Ross KB, Chong CD. Questionnaire and structural imaging data accurately predict headache improvement in patients with acute post-traumatic headache attributed to mild traumatic brain injury. Cephalalgia. 2023 May; doi: 10.1177/03331024231172736. PMID: 37157808. (paper)
  • Mao L, Dumkrieger G, Ku D, Ross K, Berisha V, Wu T, Schwedt TJ, Li J, Chong CD. Developing multivariable models for predicting headache improvement in patients with acute post-traumatic headache attributed to mild traumatic brain injury: A preliminary study. Headache: The Journal of Head and Face Pain 63(1). doi:10.1111/head.14450 (paper)
  • Mao L, Li J, Schwedt T, Wu T, Ross K, Dumkrieger G, Smith D, Chong C. Identifying and Predicting Headache Trajectories Amongst Those with Acute Post-Traumatic Headache. (major revision at Headache).
Early prediction of Alzheimer’s Disease.

This research aims to do early prediction of AD conversion from multi-modal neuroimaging and genetics data (MRI, PET, SNP). My contribution include: 1) developing a novel Supervised Multi-modal Fission Learning (MMFL) model for predictive modeling under incomplete-modality data, and 2) developing a new mutual student-teacher multi-modal learning model with theoretical analysis of the knowledge distillation effectiveness.

This research is in collaboration with the ASU-Banner Neurodegenerative Disease Research Center

Publications:

  • Mao L, Wang Q, Su Y, Lure F, Li J. Supervised multi-modal fission learning. arXiv preprint arXiv:2409.20559 (2024) (paper)
  • Kwak M, Mao L, Su Y, Chen K, Weidman D, Wu T, Lure F, Li J. A cross-modal Mutual Knowledge Distillation framework for Alzheimer’s Disease: addressing incomplete modalities (major revision at IEEE T-ASE).
  • Ku D, Zheng Z, Mao L, Chen RQ, Su Y, Chen K, Weidman D, Wu T, Lure F, Lo S and Li J. A high-dimensional incomplete-modality transfer learning method for early prediction of Alzheimer’s disease. Alzheimer’s & Dementia, 19, p.e078606.. 2023, July. doi: 10.1002/alz.078606 (conference paper)
Analyzing Influence of Public Health Organizations on Social Media

This project aims to assess health organization’s influence on social media to derive dissemination strategy recommendations. We designed a four-dimensional framework to analyze topic-specific influence on Twitter. We applied this framework to 1M+ tweets from health organizations and analyzed their influence about dietary sodium intake, one of the risk factors of cardiovascular diseases. I mentored a masters, an undergraduate, and three high school students in this project.

This project is in collaboration with Dr. Yanfang Su, from University of Washington, School of Public Health

Publications:

  • Mao L, Chu E, Gu J, Hu T, Weyner B, Su Y. A 4D theoretical framework for measuring topic-specific influence on Twitter: development and usability study on dietary sodium tweets Journal of Medical Internet Research. 2023. doi:10.2196/45897 http://dx.doi.org/10.2196/45897 (paper)
  • Montoya A, Mao L, Drewnowski A, Chen J, Shi E, Liang A, Weiner B, Su Y. Tracking influencers in policy field on social media: a global longitudinal study of dietary sodium reduction tweets, 2006-2022. (major revision at Journal of Medical Internet Research).
Prediction of Unplanned Hospitalization of Medicare Patients

This project developed machine learning models to predict unplanned hospitalization of eight types of common diseases and four type of adverse events by analyzing over 15M Medicare claims and various public datasets. My methodological contributions included co-developing a multi-layered feature selection and dynamic personalized scoring algorithm to predict monthly hospitalization risk for individual patients.

This project was led by Drs. Julie Swann and Dr. Sara Shashaani from North Carolina State University. The NCSU team was selected top 25 out of 300+ teams into Stage I of the 2020 national $1M CMS AI Health Outcomes Challenge.

Publications:

  • Mao L, Vahdat K, Shashaani S, Swann J. Personalized predictions for unplanned urinary rract infection hospitalizations with hierarchical clustering. AI and Analytics for Public Health: Proceedings of the 2020 INFORMS International Conference on Service Science (pp. 453-465) (pp. 453-465). https://doi.org/10.1007/978-3-030-75166-1 34 (conference paper) (🏆 Best Student Paper Finalist, INFORMS ICSS 2020) (🏆 CMS AI Challenge)