Alyssa Pradhan
banner
alyssapradhan.bsky.social
Alyssa Pradhan
@alyssapradhan.bsky.social
Australian Infectious Diseases trainee | DPhil in Clinical Medicine @ndm.ox.ac.uk‬
Interested in equitable healthcare AI and tahini-based dips
Interesting discussion with Hoffman, ft. some concerning sidestepping about regulation, conflating any guardrails with stifling innovation. Also interesting to see the anxiety around #DeepSeek immediately leading to justification / apology for tech billionaires..
#MLSky #digitalhealth #AI
January 29, 2025 at 2:34 AM
Getting my predictions in early 😭 #IDSky
November 19, 2024 at 2:30 PM
Sorry I think my reply got deleted! I should have credited them, I believe it was Cornell.
November 14, 2024 at 2:23 AM
Apologies, I should have credited them, I believe it was Weill Cornell
November 14, 2024 at 2:10 AM
Some great analysis of the barriers to #digitalhealth adoption in safety net patients (including digital red-lining) and how collaboration, support of patients and cautious implementation can mitigate these #AMIA2024
November 14, 2024 at 12:50 AM
Practical examples of use of LLMs to combat stigmatising language in clinical notes #AMIA2024
November 14, 2024 at 12:50 AM
Team from Vanderbilt outlining use of adaptive resource allocation in recruitment to trials to improve representativeness across studies
#AMIA2024
November 14, 2024 at 12:50 AM
A number of groups discussed #ML strategies to combat inequity, including Northwestern use of reinforcement learning for allocation of scarce resources. #AMIA2024
November 14, 2024 at 12:50 AM
This includes analysis of what #AIfairness actually means. Definitions of group, individual, counterfactual fairness, as well as at the interpolation of different demographic groups as outlined by Dr Michael Kearns and Prof Hernandez-Boussard. #AMIA2024
November 14, 2024 at 12:50 AM
A great poster from both a content and user engagement perspective! #AMIA24 #indigenoushealth
November 12, 2024 at 11:55 PM
Lots of discussion about #NLP for unstructured clinical notes over the past few days. Data from UTHealth in suicide risk detection demonstrates fine tuned #LLM > in context learning > BERT models with significant caveats of cost and training corpus size.
#AMIA24 #MLSky #MedSky
November 12, 2024 at 7:52 PM
Joining @smcgrath.phd on his quest to make #AMIA24 happen. Some sobering results on the very high hallucination rates of GPT-4 use for ED discharge summaries by UCSF.
November 12, 2024 at 7:27 PM
Download an RSS feed app, I use inoreader. Paste the RSS feed address, subscribe and re-name as you wish.
November 11, 2024 at 9:15 PM
Rename your feed, select how many abstracts you want to see and copy the feed address.
November 11, 2024 at 9:15 PM
Search and select the show abstracts. Then hit "Create RSS".
November 11, 2024 at 9:15 PM
Go to Pubmed > Advanced settings > select what journals or terms you want to follow.
November 11, 2024 at 9:15 PM
First day at #AMIA24, some great insights about evaluation of #LLM in healthcare.

Also an apt time to scream into the #Medsky void, if you’re part of the #IDtwitter #Medtwitter #healthcareAI exodus, hmu
November 9, 2024 at 11:37 PM