The mission of the University of Kentucky Division of Pathology Informatics is to leverage, develop, and promote the latest advancements in science, engineering, and technology in support of clinical practice, advancement of research, and the promotion of informatic training.
Historically, the preponderance of actionable patient data is derived from pathology and laboratory medicine. In recent years, especially in the areas of omics and imaging, the volume of data being generated in clinical practice is rapidly increasing. Pathology informatics supports the generation, analysis, validation, and reporting of data generated as part of clinical pathology practice. Making use of new sources of data and the latest technology, our division is well positioned to advance artificial intelligence (AI) efforts in healthcare.
Much like the broader field of pathology, our division collaborates across medical disciplines, from diagnostic testing and consultation, to the use of cutting-edge technologies for the prevention of disease. Within the field of pathology, activities include, but are not limited to:
- Anatomic Pathology
- Development of digital pathology slide repositories for clinical, research, and training purposes.
- Development of natural language processing (NLP) AI models based on pathology reports.
- Development of predictive AI models based on pathology imaging.
- Application of biostatistics and computational biology applied clinical and research problems.
- Development of next generation sequenicng (NGS) based diagnostics supporting efforts in oncology, constitutional disorders, cytogenetics, and microbiology.
- Cellular and Molecular Pathology
- Develop constitutional pipelines and repositories to support the processing of molecular data
- Advancements in the speed and scale of molecular processing through cloud-based and hardware-accelerated approaches.
- Development of predictive models making use of patient reports and curated molecular databases.
- Clinical Pathology
- Computational approaches to data generation, validation, verification, quality control, and quality assurance.
- Integration between laboratory devices and systems.
- Development of predictive AI models based on laboratory data.
V.K. Cody Bumgardner, PhD
Director of Pathology Informatics
Shadi Qasem, MD (Anatomic Pathology)
Kannabiran Nandakumar, PhD (Bioinformatics)
Shulin Zhang, MD, PhD (Cellular and Molecular Pathology)
Min Yu, PhD (Clinical Pathology)
Caylin Hickey (Data Science, Computer Science PhD Student)
Thilakam Murali, PhD (Bioinformatics)
Our division makes use of a range of resources from laboratory testing devices, next generation genomics sequencers, image scanners, to computational clusters. The majority of clinical efforts leverage the vast resources provided by cloud computing in conjunction with local hospital resources. Each service is designed to scale as needed to reduce computational delays.
Outside of clinical efforts, our division makes use of institutional research computing infrastructure, including high performance computational clusters and networks. We work closely with the Center for Computational Sciences and are affiliated with the Institute for Biomedical Informatics.
In addition, our division has the following computational resources:
- Large Memory Cluster: 10 Node, 800 Cores, 30.9 TB of RAM, and 60 TB of storage
- Hadoop Cluster: 11 Nodes, 264 Cores, 1,056 G of RAM, and 100 TB of storage
- Data Transfer Node: Server with direct 100Gb access to Internet/Internet2 used for high-speed data transfers
- Public Big-Data Node: 80 Core, 3TB RAM Server with direct 40Gb access to Internet/internet2 used fo rpublic facing large-data high-speed analysis
- Moonshot 'Streaming Data' Cluster: 4 x ARM nodes, with 64GB of RAM, and 2 x Intel Nodes, with 48 GB of RAM, used to process data in movement, also known as streaming data.
- Edge Computer Cluster: 3 x Raspberry Pi, 5 x NVIDIA Nano, 2 x NVIDIA TXS Jetsons, used for variosu distributed edge and edge-cloud computing testing and simulations.
These resources are connected to an advanced, research-specific network. This network insfrastructure includes a direct 100Gb connection to Internet2, 100 Gb Software Defined Network (SDN) research DMZ, and 100Gb connectivity to additional computational resources. This research network improves data-intensive sciences, providing greatly improved access to both campus, collaborators, and national resources. In addition, this connectivity enables the high-speed use of public cloud services.
Our division is committed to resident, graduate, and undergraduate education. The scope of work allows us to host students from pathology-specific subspecialties, general biomedical informatics, to fundamental research in computer science and engineering.
For interest in an observational learning experience, such as an internship or practicum, please send an email to firstname.lastname@example.org.