Job ID 1482051
Provide operational, executional, tactical, and strategic support for the Good Algorithmic Practice initiative.
Help to implement recommendations from the Good Algorithmic Practice Initiative focused on standardization, reproducibility, and data transparency.
Help develop standards for the operationalization and fit-for-purpose use of AI/ML methods applied to Real-World Data (RWD). Produce exemplary educational material, including tools geared toward under-served communities and new learners, covering topics relevant to the use of RWD including but not limited to the treatment of missing data, case-control selection, minimization of type I/II error and bias, prevention of model overfitting and poor generalization and approaches to Electronic Health Record (EHR) data limitations.
Manage, schedule, and oversee outreach initiatives to underserved communities and extramural sites with limited bioinformatics capacities.
Support investigations from the Office of the Director and N3C leadership requiring biostatistical and/or machine learning knowledge and domain expertise. The outputs of these efforts would serve as exemplars of Good Algorithmic Practice and may result in publications.
Produce criteria to evaluate the incremental benefit of PPRL-linked datasets (i.e., improvements in the quality and/or scope of analyses of N3C data, reduction in data missingness, etc.).
Participate in N3C Enclave Knowledge Store object and code template peer review, integration and unit testing, and validation.
Minimum MS in computer science, (bio)statistics, or a similar quantitative degree.
Proven track record of working with real-world clinical data.
Experience in modern machine learning techniques and biostatistics. Preferably experienced in causal inference methods, and supervised, and unsupervised learning.
Strong oral and written communication skills and ability to actively participate in technical discussions and follow up on action items.
Sound judgment, positive attitude, and creative outlook required.
NIH experience preferred but not necessary.