I had a great time discussing our research developing #NLP methods to actively support patients undergoing radiotherapy and other cancer treatment
AtlanticLIVE
#PeopleVCancer
@massgenbrigham
Brigham and Women’s Research

🗞️ Our latest work out today in Nature Communications

🌟 Led by star research fellow Anna Zapaishchykova

🧠 Automated temporalis muscle quantification and growth charts for children through adulthood rdcu.be/dqFGu

🧵 1/

🚨 #Cardioonc #lungcancer study highlight! tinyurl.com/3ezfpwtn

Led by Bonnie Ky Paco E. Bravo, MD Cliff Robinson Ray Mak, MD Marcelo Di Carli Joshua Mitchell Nitin Ohri @SalmaJabbour1Michael Soike Manuj Agarwal Steve Feigenberg & William Levin

Abramson Cancer Ctr. #LungCancerAwarenessMonth

Join us on Wednesday 11 EST- what's new in🫁lung ☢️ #radonc ⚡️webinar details below! #LCSM @iaslc Fiona McDonald Houda Bahig, MD PhD Pranshu Mohindra Shankar Siva Corinne Faivre-Finn 💙 Dr. David Palma Drew Moghanaki 🐕 Matthias Guckenberger Joe Y Chang

Really enjoyed the discussion! Thanks to Abhinav Suri, MPH, Hesham Elhalawani Charles Kahn, MD and RadiologyAI for invite, for organizing and insightful questions on AI-opportunistic (ambient) screening. #RadAIChat

T5. Before we deploy them we should prove they are relevant to our patients, develop efficient macros/separate reports as needed, and provide extensive referring clinician education. Ideally patient, too. Mammo letters are not a bad notification model. #RadAIchat

T4. Here’s a cool one. ILD screening on #radiotherapy planning CT scans. Identifies undiagnosed ILD, a patient subset in which pulmonary RT dose can lead to fatal outcomes. Deployment is #radonc facing, but refers to a radiologist for consult. #RadAIchat
buff.ly/3FIKv9U

T3. As with any screening, main concerns are false positives leading to unnecessary procedures, anxiety, etc, with the added layer of complexity of black-box algorithms with outputs that may not be explainable. #RadAIchat

Radiology: Artificial Intelligence T3: I think one of the biggest downsides is the need to have resources available to patients who receive this information. If a patient learns they have osteoporosis but can't get calcium supplements or see someone who can address this what good have we done? #RadAIChat

T3. As with any screening, main concerns are false positives leading to unnecessary procedures, anxiety, etc, with the added layer of complexity of black-box algorithms with outputs that may not be explainable.

Radiology: Artificial Intelligence T2. Giving a plug for using CT scans to screen for diabetes. Our lab showed that biomarkers derived from pancreas segmentations can predict diabetes development in the future! #RadAIchat Pritam Mukherjee NIH