Data Science Postdoctoral Fellow

Penn State University

College of Medicine at the Milton S. Hershey Medical Center
Date Announced:
Date Closing:
open until filled
Job Number:
Work Unit:
Penn State College of Medicine


The Center for Optimizing Radiology Value (CORVA) in the Department of Radiology at Penn State Hershey Medical Center and Penn State College of Medicine invites applications for up to 3 data science postdoctoral positions, initially one year and renewable for a second year by mutual agreement. CORVA is at the interface between digital data (electronic medical record data and imaging data) and delivery science I implementation science. Our objective is to transform institutional data assets into clinical insights with the power to improve healthcare delivery locally and more broadly to increase the field`s knowledge of data science. While we`re based in Radiology, our remit is broad. Our department touches more than 85% of patients in the system, performing nearly 400,000 imaging studies. Our team works across internal medicine, the emergency department, ICUs and orthopedics. Our `business partners` across the institution have a large and well funded portfolio of grant funding from PCORI, NIH, and other federal agencies. Successful applicants will be expected to take advantage of the institution`s many clinical departments and our excellent working relationships there to source interesting projects at the intersection of novelty and practicality. The position will also provide opportunities for diverse research experiences in a highly collegial environment at both the College of Medicine campus at Hershey and the University Park campus. Successful applicants will have preferential access to CORVA`s 3 database scientists, a team of 3 IRB staff, a pre award grant specialist and dedicated administrative support. You will be mentored by senior clinical staff and experienced health services researchers. We anticipate that as much as half of your time will be protected to pursue your own program of research. Representative current data science projects include: predictive analytics for after discharge outcomes among chronically ill patients, impact of celebrity news on breast cancer screening, mining of the universe of Facebook posts for side effects of statins, re-analysis of the SPRINT blood pressure trial, prospective implementation of clinical decision rules for in "hospital mortality among critically ill patients, etc. Future projects could include: supervised machine learning around image data in radiation oncology or bone mineral density inferences in fragility fractures, operations research models in scheduling and staffing of imaging, cost predictive analytics patient reported outcomes and social determinants of health, etc. After fellowship, we expect our postdoctoral fellows to place academically in Divisions of Healthcare Delivery Science or in Centers of Clinical Innovation or more traditional health services research departments. On this track, we'd be surprised if successful applicants did not obtain some early career funding and did not publish three or more papers a year, although neither is a requirement. For those not intending to become academics, we expect postdoctoral fellows to be very competitive in the consulting, deep learning, EMR vendor, or hospital IT spaces where strong quant skills mated to intensive situational fluency in healthcare operations will be invaluable. On this track, we`d be surprised if successful applicants did not create valuable, IP-protected algorithms, from which royalties would accrue to the developer, although this is not a requirement either. There are few - not nearly enough - formal data scientists in healthcare today. That`s why we`re not overly focused on your current doctoral experience. Healthcare experience is not required; situational fluency is something you`ll develop on the ground. Rather, promising computer science, physics, math, economics or statistics graduates, and even quant heavy marketing science or business school graduates are at least as likely to make good healthcare data scientists as traditional health services researchers. The only mandatory requirement is a superlative understanding of methods and extremely strong practical skills in at least one of the following, with strong preference for model building: data extraction side - Building scalable data pipelines, data engineering, and feature engineering. Database - Constructing, analyzing and administering relational databases, SQL, No SQL, business intelligence tools such as IBM Cognos, SAP Business Objects. Big data - Distributed programming languages such as Hadoop/MapReduce. Data mining - Bayesian networks, CHAID, CART, hierarchical mixed models, and generalized linear models. Machine learning - Information-, similarity-, probability-,and error-based approaches including random forests, gradient boosting, SVMs, clustering, graph theory and neural networks. Statistics - Bayesian and frequentist methodologies, unimpeachable mathematical understanding of statistical testing, ideally from a clinical trials or more general biostatistics background. Model building - Prototyping, building, validating and deploying predictive analytics models in R, Stata, SAS Enterprise Miner, Matlab, or using scripting languages, Python, Pearl, Caffe, Scikit-Learn. Image processing - deep learning interests and training in feature extraction, target classification and diagnostics in natively digital data environments such as radiomics, imaging, breast cancer computer aided detection etc. Applicants should be a U.S. citizen or noncitizen national, or must have been lawfully admitted for permanent residence and possess an Alien Registration Receipt Card (I 151 or 1-155) or some other verification of legal admission as a permanent citizen; individuals on temporary or student visas are not eligible. Candidates should upload a one page application letter addressing the points above, a curriculum vitae and contact information for three references.

These salary bands have been established to provide salary guidelines for staff positions.

Salary Band Minimum Midpoint Maximum
A $16,584 $24,456 $32,328
B $18,240 $26,904 $35,556
C $19,728 $29,592 $39,456
D $21,708 $32,568 $43,416
E $24,312 $36,468 $48,612
F $27,228 $40,848 $54,456
G $30,012 $45,744 $61,500
H $34,188 $52,140 $70,080
I $38,988 $59,424 $79,908
J $43,716 $67,740 $91,812
K $50,712 $78,600 $106,488
L $58,836 $91,176 $123,528
M $68,232 $105,756 $143,292
N $80,508 $124,788 $169,068
O $93,492 $147,252 $201,024
P $110,340 $173,760 $237,192
Q $126,396 $199,056 $271,728
R $151,668 $238,872 $326,088