Advancing Research with AWS

Amazon Web Services (AWS) has provided Pitt with $100,000 in credit for the no-cost use of applicable AWS cloud services for conducting research. The program aims to support projects that seek to build a cloud-hosted service, software, or tools, and migrate research processes or open data to the cloud. More than 30 requests have been received for up to $5,000 in AWS promotional credit.

“The availability of the AWS cloud service helps improve our research performance. It provides scalable and cost-effective computational resources, customized based on our need, as well as easy access to a wide range of bioinformatics tools and databases for efficiently processing and analyzing large genomic datasets. In addition, AWS is a stable running environment that we can access remotely and that remains unaffected by severe weather or electronic outages.”

- Lujia Chen, Assistant Professor, Department of Biomedical Informatics, Developing Deep Learning Models for Precision Oncology

“AWS seed grant funding has played a pivotal role in advancing my research in AI and medical imaging. By offering crucial resources and infrastructure through the AWS cloud platform, it has empowered me to efficiently showcase and validate the outcomes of my research endeavors. This invaluable support has not only amplified the scalability and accessibility of my work but has also fostered collaboration and innovation within the field, making it an indispensable asset to the success of my research projects."

- Jiantao Pu , Professor of Radiology and Bioengineering, Department of Radiology, CT Scanning for Pneumonia vs. Traditional Xray

"The AWS seed grant allows us to utilize the state-of-the-art cloud computational architecture and software to data-mine electronic medical records for outcomes research. The Pitt CRC team and AWS team walked us step-by-step to go through the systems and recommended powerful software for our research. We really appreciate your professional services and help."

- LiRong Wang, Assistant Professor of Pharmaceutical Sciences, School of Pharmacy, Deep Learning on EMR Data for Outcome Research

Answering the Archiving Challenge

Storing large amounts of data on an on-premises server can quickly become expensive for University departments. Pitt IT launched a new Archiving Service to help departments move less frequently accessed data to a cost-effective cloud storage environment hosted by Amazon Web Services. The Department of Pediatrics is the first University unit to adopt the new service. They were able to archive 29% of their data, reducing their storage costs by more than 96%; a monthly savings of $1,473.

Enabling Research Compensation

UPMC discontinued administration and support of the service that is used to pay research participants, creating a potential gap. Pitt IT stepped in to assume administration and end user support of the service, ensuring continuity of research for the University and UPMC.

“Pitt IT has done a great job coordinating the technical complexities of keeping Vincent up and running for our important stakeholders who rely on Vincent to manage research participant payments.”

- George Cerminara, Project Manager, Office of the Treasurer

Enhancing Secure Research

To ensure the University remains a leader in computationally enabled health sciences research, Pitt must provide high-performance computing (HPC) resources that also meet stringent data security standards. To this end, Pitt IT is developing a customized environment that enables research with HIPAA data to be performed on a secure, controlled HPC cluster.

Envisioning the Data Center’s Future

The computing infrastructure hosted at the University’s Data Center must keep pace with changing technology and the needs of the University research community. Pitt IT and Facilities Management have engaged with an engineering firm to conduct a study that will evaluate the Data Center’s power and cooling infrastructure. The study will focus on the state of existing infrastructure and make recommendations for enhancements to support the University’s evolving needs.

Facilitating Research Discovery

Pitt researchers benefit when they are more easily able to search for and learn about other researchers’ projects so that they can reach out, collaborate, and network with others in their field. Pitt IT helped make this possible by developing a Research Discovery Tool prototype for the Office of the Senior Vice Chancellor for Research. Microsoft funded the prototyping and assisted in creating the intelligence behind the tool.

Improving Post-Award Business Processes

The current post-award business process found duplicate entry points and manual entry that can lead to errors in sponsored project activations. Pitt IT is working with Sponsored Projects Accounting and the Office of Sponsored Programs to streamline award activation and modify processes with the goal of minimizing these issues to benefit all University research administrators. Along with the process improvements, a total of eleven dashboards were developed and released to research administrators — increasing overall visibility for sponsored project administration and reporting.

"The Business Process Improvement (BPI) process led by Pitt IT has added value by identifying pain points, and provided new ideas and solutions. After several BPI processes, we now have access to metrics, dashboards, training resources, and tracking capabilities that have improved our performance."

- Kellie Sellers, Systems Manager, Sponsored Projects Accounting

Reclaiming Enterprise Storage Space

Pitt IT’s Dell Isilon storage environment provides high-volume storage, backup, and archiving of data for departments and researchers. As Pitt continues to generate more and more data, storage space is consumed at an ever-increasing pace. Working with Dell, Pitt IT implemented data deduplication software that recovered 25% of the consumed capacity of Isilon storage. To date, the software has succeeded in saving 821 TB of space, delaying the need to purchase additional hardware for storage.

Sharing Data Across Universities

Pitt IT is working with the Department of Biomedical Informatics (DBMI) on a project to securely collaborate and share data with other Universities for large research studies. Deidentified datasets from numerous sites will be preprocessed and prepared for a deep learning algorithm and machine learning, leveraging Globus for transfer of large data sets.

Supporting Diabetes Research

A Pitt research team in the Department of Biomedical Informatics working on the eDERRI model ― which uses electronic health records data to better predict hospital readmission risk among diabetes patients ― was interested in technology support for their research. Pitt IT worked with DBMI to provide the team with cloud-based tools, including a virtual machine built specifically for performing data science, to assist them in their work.

Viewing High-Performance Computing Progress

The Center for Research Computing (CRC) provides high-performance computing (HPC) resources to University researchers through an automated queuing system. Previously, researchers had limited visibility into the estimated completion time for their job after it was submitted to the CRC computing cluster. To address this need, Pitt IT launched a dashboard that enables researchers to view the wait time for their computing jobs. This complements dashboards that Pitt IT created to provide visibility into overall usage and computing load.