Trishan Panch
@basslinetherapy
MD MPH. President @HSPH Alumni. Co-founder@Wellframe
ID: 61020756
28-07-2009 21:14:04
1,1K Tweet
815 Followers
819 Following
        
        
        New today, in this Opinion piece Panch Trishan Panch et al, propose a distributed approach to the regulation of #clinicalAI that includes an amalgam of centralized and decentralized regulation #DigitalHealth dx.plos.org/10.1371/journa…
                        
                    
                    
                    
                
        A pleasure to work on this with friends MIT Critical Data Heather Mattie Felix Greaves Gopal Kotecha The current approach to the regulation of clinical AI will not scale in my opinion and a new distributed approach is needed.
        
        
        Join our next VRT: AI at the Clinical Frontlines, Mon 6/13 12pm EDT bit.ly/20220613AIFron… #keepconnectedstaysmart Harvard T.H. Chan School of Public Health Trishan Panch Harvard Health Harvard Medical School Harvard Business School
                        
                    
                    
                    
                
        Amazing – tour de force journalism piece by Linda Villarosa in The New York Times on multigenerational impacts of forced sterilization. A disquieting, crucial read. Structural violence against Black women, the disabled, the imprisoned, and the push for reparations. hsph.me/eugenics
        Impressive work with huge potential clinical impact from David Sontag and his lab re LLMs applied to a clinical corpus. New state of art performance in 4 areas WITHOUT retraining (zero vs few shot). Compelling both from pt and clinician workflow pov. arxiv.org/abs/2205.12689
        
        Dr. Trishan Panch (Trishan Panch)—a Program Director for our #AI for Health Care: Concepts and Applications course—was a participant in a Microsoft Research panel on Responsible, Equitable, and Ethical AI moderated by junaid bajwa. Watch the panel here: youtube.com/watch?v=pq7xMa…
        Thanks so much for your leadership and vision in putting this together junaid bajwa
        How can we translate health AI from research to benefit citizens and patients? Hear how working together as a community is key to this with junaid bajwa Microsoft Research, June Raine MHRAgovuk, Trishan Panch Harvard T.H. Chan School of Public Health, and Amy Abernethy Verily -> aka.ms/HealthFuturesF…
                        
                    
                    
                    
                
        When training ChatGPT from human preferences, why use policy gradient? I built a crude RLHF system on top of Andrej Karpathy's nanoGPT and instead tried backpropagating through the reward model using the Gumbel-Softmax trick. It works for a toy example. github.com/sanjeevanahila…
        @sanjeevanahilan Andrej Karpathy Needs a name! Is this... GDHF / "gradient descent from human feedback"?
        
        “I believe healthcare is based on relationships, which is based on trust, which is earned”~Trishan Panch Trishan Panch MD and Co-Founder of Wellframe TEDxBoston #TEDxBoston
                        
                    
                    
                    
                
        We need more than AI, we need AE — Artificial Empathy ~Trishan Panch AI + AE = less cognitive load “We are missing the boat” if we are just thinking about the tech. TEDxBoston #TEDxBoston
                        
                    
                    
                    
                
        My TEDxBoston #TEDxBoston talk Massachusetts Institute of Technology (MIT) on #artificalempathy #ae and #AI youtu.be/mo7QGo_qwl8
        new paper hot off the presses with Trishan Panch and Felix Greaves #PLOSDigitalHealth: Hallucination or Confabulation? Neuroanatomy as metaphor in Large Language Models dx.plos.org/10.1371/journa…