2020-02-26 FEB

Pathology AI - from AutoML to TF.js

Over the past 1 year Praful has developed the application where users annotate TMA's, then used to deploy a prediction service that responds in 1-2 seconds, and this week ported it to tf.js where it makes a prediction at zero cost (client side commodity laptop) 40-50x faster. Le's take our time to see how he does all this.

Recording of this session: https://www.youtube.com/watch?v=1XGKLdhZrT0&t=10s


Pathology AI - journal club

https://ai.googleblog.com/2020/02/generating-diverse-synthetic-medical.html

the paper: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6939343/

PWA

What are they, how are they developed, give it a try creating a web worker, have a go at one of the toolboxes

STRIDES

The NIH STRIDES Initiative is sponsoring a few training opportunities in March that you can attend free of charge. Please see below for the list of courses:


If you are interested in attending one or more of these trainings please reach out to me directly and let me know which offering(s) you’re interested in.

If you’re unable to attend, but know someone who might benefit from taking one or a number of these courses, please feel free to share this email with them and have them reach out to me personally. These classes are open to all NIH Employees and Contractors.

Thank you,

Matt Gieseke



This seminar may be of interest to our group and begins a 11:00 am

NCI CBIIT Data Science Seminar Series

Speakers:

Guilherme Del Fiol, M.D., Ph.D.

Richard Bradshaw, Ph.D.

University of Utah School of Medicine

Date: Wednesday, February 26, 2020

Time: 11:00 a.m. ET

Topic: Identifying and Managing Patients Who Meet Evidence-Based Criteria for Genetic Evaluation of Familial Cancer Using the FHIR Data Standard

In this talk, Drs. Del Fiol and Bradshaw will discuss OpenCDS, a clinical decision support platform, and show how it can be integrated with electronic health records (EHR) to identify and manage patients who may benefit from genetic evaluation of their hereditary cancer risk. They will describe the software architecture, clinical workflow, and show results of a study by the Huntsman Cancer Institute that piloted this application with patients from the University of Utah. This project is funded through NCI’s Informatics Technology for Cancer Research (ITCR) program (1U24CA204800-01A1).

Our speakers will present remotely. Register now and join us via Webex or attend the presentation in person at the NCI Shady Grove Campus, Room 2W032/034. Register now at bit.ly/2vaW20D

Preliminary Concept IDs

Nicole and Daniel have taken their first stab at creating Concept IDs and JSON schema for Connect and invite feedback.