Machine Learning and Clinical Informatics

Description

The National Library of Medicine has an opening for a postdoctoral fellow in the lab of Dr. Jeremy Weiss (https://www.nlm.nih.gov/research/researchstaff/WeissJeremy.html) for research in machine learning and clinical informatics. Research directions include: increasing performance and trustworthiness of deep learning in healthcare, improving natural language processing methods for risk forecasting, and developing bridging methods across algorithmic fairness and survival analysis.

Compensation

See the NIH Stipend Schedule: https://www.training.nih.gov/documents/11/2023_IRTA_and_VF_Stipend_Tables.pdf. Fellows with a quantitative focus may receive an additional supplement of up to ten thousand dollars per annum.

We are seeking to investigate methods that learn from longitudinal, multimodal (text-and-tabular) data present in electronic health records data that (1) are predictive and are useful for forecasting, and (2) that preserve interrogable properties. The research will address the trustworthiness of EHR data for clinical decision making. Relevant experience includes expertise in representation learning, signal processing, longitudinal data, ethical AI, and multi-modal learning, with applications in health domains. Relevant programming experience includes pytorch/tensorboard/huggingface/sklearn and R/tidyverse.

Recent relevant publications:

• Zhang, Wenbin, and Jeremy Weiss. ""Longitudinal fairness with censorship."" Proceedings of the AAAI Conference on Artificial Intelligence. 2022. (https://arxiv.org/pdf/2203.16024v2.pdf)
• Cheng, C. and Weiss, JC. ""Typed Markers and Context for Clinical Temporal Relation Extraction."" Machine Learning for Healthcare. PMLR, 2023. (https://static1.squarespace.com/static/59d5ac1780bd5ef9c396eda6/t/64d198242588467b55e5b7e0/1691457572386/ID23_Research Paper_2023.pdf)
• Noroozizadeh, S., Weiss, JC; and George Chen. “Temporal Supervised Contrastive Learning for Modeling Patient Risk Progression.” Machine Learning for Health, PMLR, 2023. (https://proceedings.mlr.press/v225/noroozizadeh23a/noroozizadeh23a.pdf)

• PhD, MD, or equivalent in related subject areas, including machine learning, computer
• science, medicine, bioinformatics, biomedical informatics and statistics/biostatistics;
• Publishing experience with peer-reviewed journals and/or conferences in the above areas;
• Appointees may be U.S. citizens, permanent residents, or foreign nationals (visa requirements apply).

Timing: postdoctoral fellowships are for 2 years with the possibility for extension. Candidates

are subject to a background investigation. Applicants will be reviewed on a rolling basis.

The NIH is dedicated to building a diverse community in its training and employment programs and encourages the application and nomination of qualified women, minorities, and individuals with disabilities.