About
Welcome to the Emerging Diagnostic and Investigative Technologies EDIT program in the Department of Pathology and Laboratory Medicine at Dartmouth Hitchcock Medical Center. Professor Joshua Levy Levy Lab, Professor Louis Vaickus, and Professor Matthew Hayden serve as leaders, along with Hitchcock's Steve Firing and Benjamin Koziol, program coordinators Ben Mattern and Moses Odei Addai, and Dartmouth's Cancer Center leadership. The EDIT ML program hosts a national summer research program for high school and undergraduate students that provides exposure to, and instruction reguarding biomedical research.
"Emerging technologies informed by and for clinicians have the potential for lasting translational impact."
Diversifying STEM Workforce:
This program plays a vital role in mitigating bias in AI-augmented clinical decision making by diversifying the STEM workforce. While AI technologies have the potential to revolutionize healthcare by augmenting clinical decision making, bias in AI algorithms can negatively impact historically underserved groups and lead to misdiagnosis. The lack of representation from underrepresented racial and ethnic groups, women, and under-resourced rural and urban school districts in STEM training and career exploration exacerbates this bias. By broadening participation in the development of AI algorithms and diversifying the STEM workforce, we can ensure that AI technologies benefit clinicians and patients from all backgrounds.
Equitable Access to Medical AI Education:
The realities of internet technologies and the limitations imposed by the pandemic have further highlighted the need for remote learning opportunities. In-person STEM learning often favors students who can travel and afford such experiences. The pandemic has limited access to in-person learning and exposed the inaccessibility of tuition-charging or unpaid programs. Remote learning, while expanding educational experiences, presents challenges for students without reliable internet and devices, especially in communities with limited infrastructure. By offering a remote learning opportunity, we aim to bridge these gaps and provide equitable access to AI education and STEM opportunities.
Program Structure and Resources
Through a series of lectures, guided projects, and IRB supported basic research, Program members develop algorithms to explore diagnostic spaces in pathology from cancer detection, to gigapixel image manipulation, to text prediction. Students are placed into teams to design and pitch projects and adhere to a team culture which promotes broad collaboration. Dr. Levy meets weekly with lab project teams to discuss updates and provide guidance on the technical aspects of their projects (including presentation/manuscript preparation), while providing tutorials (e.g., overview of operating in an HPC environment), a lab GitHub based wiki page, and over the summer a weekly seminar series Seminar series to help them better understand emerging themes in the field. Dr. Levy also holds weekly office hours for general inquiries. Students in the EDIT ML lab have access to vast compute resources:
-
QDP-Alpha HPC resource for research applications, a computational node system comprising:
- 400 CPU cores
- 416GB of GPU memory
- 1.5TB of RAM
- 150TB of storage
-
Pathology Local Cluster (PLC), a modular HPC system for mature clinical applications comprising a genome analysis node:
- 128 CPU cores
- 768GB RAM
- 12GB GPU memory
- 100TB storage
And a machine learning node:- 40 CPU cores
- 128GB GPU memory
- 512GB RAM
- 100TB storage
- Personal computing equipment, including access to surface tablets to study histology slides and a local computing workstation for rapid algorithm prototyping.