Anirudh Joshi
@anirrjoshi
Followers
506
Following
1K
Media
2
Statuses
113
Co-Founder and CEO @valarlabs AI+Medicine 🇺🇸
Stanford, CA
Joined February 2020
Excited to share our new work "Dr Summarize: Global Summarization of Medical Dialogue by Exploiting Local Structures" which was accepted for publication at EMNLP Findings 2020 ! @CuraiHQ @stanfordnlp. Read on for a synopsis!. 1/n.
2
8
45
The 2008 financial crisis was followed by the fin tech boom (Square, Stripe, Plaid, Credit Karma, Robinhood) the 2020 healthcare crisis will be followed by the health tech boom.
0
2
27
I’m excited to announce Valar Labs (@valarlabs) and thrilled to be working with @VisweshKrishna, @dvrabac and @pranavrajpurkar to bring AI to decision making in oncology. We are joined by @a16z, @pearvc and a team of phenomenal physicians and engineers.
1
4
19
It’s a big moment in bladder cancer. We’ve been able to successfully validate computational histology based AI biomarkers to predict outcomes and response to first line treatment. Press Release:
1
2
12
Very excited to be working with @vintweeta and @JayRughani on building @valarlabs as a category defining company in oncology.
Digital pathology is evolving. v1.0 = automate pathologist workflow to drive status quo efficiency. v2.0 (via @valarlabs) = EXTEND pathologist w ML-based insights humans can't even see. Use digitized features + past pt data to predict #cancer therapy response. Software as dx!.
0
2
12
Sharing our rapid fire 🔥 questions with our @Pearvc Accelerator founders. Hear how they are prepping ☕ and feeling 🤩 about our demo day on October 13!.
2
0
10
Wonderful working with @HBhambhvani @BladderCancerMD @ShkolyarE @SunnyKaoMD on this and excited for more to come!.
Excited to present our work at #ESMO2022 to predict neo-adjuvant chemotherapy outcomes in bladder cancer with our AI based digital pathology assay. Great collaboration with @BladderCancerMD @ShkolyarE @SunnyKaoMD @StanfordUrology @anirrjoshi @dvrabac.Link:
0
1
10
Great working with @DrHendifar @PancPathologist Eric Collisson and Christos Fountzilas on the next generation of diagnostics in oncology.
We are excited to present our work on first line chemotherapy selection in #pancreas cancer with an AI + pathology based Dx @asco GI 2023! Great collaboration with @DrHendifar @PancPathologist Eric Collisson Christos Fountzilas.Abstract: #GI23.
0
1
6
It was a lot of fun working with the ML Research team @CuraiHQ. The convenience and ease of their healthcare delivery is clearly the future of primary care.
Read about the great work from our recent batch of ML Research interns at our latest tech blog post:
1
1
8
Excited to share our new work #GloFlow that was accepted to Neurips ML4H. We see this technology as a way to lower the barrier to slide digitization in pathology and biology. Partner with us to pilot this technology : Paper:
0
0
7
We are actively looking for collaborators who would be interested in piloting a novel slide digitization technique that can be done over a microscope without expensive scanning equipment. Please leave your information here: @StanfordAILab @StanfordMed.
Excited to share #GloFlow to create a whole slide image stitch from a video scan, enabling slide digitization directly from microscope. Partner w/ us: Paper: @visweshkrishna @anirrjoshi + al. @StanfordAILab @StanfordPath. 1/7
1
0
7
Honored to collaborate alongside experts in bladder cancer @ylotan @UroDocAsh on our international validation of #CHAI biomarkers to predict outcomes and treatment response for patients. @valarlabs @urotoday.
AI-based assessment tool for predicting BCG response in #BladderCancer pts. @ylotan @UTSWUrology joins @UroDocAsh @MDAndersonNews to discuss this innovative tool presented by @valarlabs aimed at refining the management of BCG-unresponsive bladder cancer >
0
0
8
A new look into clinical generalization of deep learning. We find that even if performance drops on distribution shifts, the resulting performance is still equivalent to radiologists on that distribution. Read more in our publication @CHILconference (.
Can medical imaging models maintain radiologist performance when tested on external clinical settings without fine-tuning?. Surprisingly, the answer is YES!. Paper: @anirrjoshi @anujpareek @AndrewYNg @mattlungrenMD @StanfordAILab . 1/6
0
1
5
@valarlabs 👀.
Today is the big day! 🍐 . A throwback to Day 1 in June when it all began… . from 🐛 to 🦋 . #pears21 #demoday
0
1
5
@SWilliams_MD has been a tremendous supporter from the beginning and it’s been an honor to work together.
Leveraging AI biomarkers to navigate the evolving landscape of #BladderCancer treatment @SWilliams_MD @UTMBUrology joins @zklaassen_md @GACancerCenter to discuss this innovative AI biomarker for #NMIBC > @valarlabs
1
0
3
Excited to be joining @CuraiHQ as an intern soon. Their approach of consumer first free healthcare will be a model for the future.
0
0
3
COVID-19 is causing mass digitization in industries that shied away from the internet. Restaurants -> cloud kitchens , medicine -> telemedicine. What else is next ?.
3
0
4
This is going to change education.
So @OpenAI have given me early access to a tool which allows developers to use what is essentially the most powerful text generator ever. I thought I’d test it by asking a medical question. The bold text is the text generated by the AI. Incredible. (1/2)
0
0
4
Thrilled to release our work on studying how chest x-ray interpretation algorithms generalize to challenges seen in real world deployment. Hopefully this moves us one step closer to increasing accessibility to AI screening technologies. @pranavrajpurkar @AndrewYNg @mattlungrenMD.
Do #AI models for chest x-ray interpretation generalize to challenges seen in real-world deployment?. Our new work investigates! @anirrjoshi @mattlungrenMD @AndrewYNg .@anujpareek @jeremy_irvin16 @mirkiani .@StanfordAIMI (thread) 1/
0
0
2
Audio first experiences are exploding from Clubhouse, Discord and now Twitter. AirPods could be a new app platform 👂.
0
1
4
@pranavrajpurkar OpenAI is probably saving on inference costs with a distilled version of the models.
0
0
2
If you'd like to be part of building the next generation of oncology decision support, join us!.
0
0
3
We are fortunate to have an incredible group of oncology and pathology collaborators at the world’s best cancer centers. If you share our vision and are interested in collaborating with us please reach out at hello@valarlabs.com.
1
0
3
@DrHendifar @CSCancerCare @valarlabs We are thrilled to be working with you and @CSCancerCare as well!.
0
0
3
Aligning NLP automated metrics with human evaluations has been notoriously difficult. This is an impressive step towards using RL to assist with that.
We've used reinforcement learning from human feedback to train language models for summarization. The resulting models produce better summaries than 10x larger models trained only with supervised learning:
0
1
2
For the last 8 years I’ve pursued a dream that AI could fundamentally change how medicine is practiced and drastically scale our access to medical insight. Valar Labs is the next phase of this journey and I couldn't be more proud of what our team is building.
1
0
2
0
0
2
0
0
2
🔥.
Lambda Networks: SOTA on ImageNet. Again Transformer-like architectures dominating over a new field 🚀. CNNs had a good run 😰. Paper: Awesome video from @ykilcher:.
0
0
1
@nealkhosla With low interest rates and reduced traffic on the roads it’s the perfect time for infrastructure investment.
0
0
2
@vintweeta @suchisaria @VanAllenLab @mattlungrenMD @nevillesanjana @AuslanderNoam Interpretability can also serve as a means of robustness and generalization of AI to different distributions and patients. As these models are typically more “simple” they tend to overfit less to specifities of a cohort.
1
0
1
@VivHo @valarlabs @VisweshKrishna @dvrabac @pranavrajpurkar It’s been so fun working with you and the @pearvc team!.
0
0
1
@CuraiHQ we want to scale quality healthcare to everyone. Automated summarization of medical conversations between providers and patients is an avenue in improving scalability however a challenge for NLP systems since medical conditions have to be captured accurately. 2/n.
1
0
1
@nikitaandersso3 @dvrabac @VisweshKrishna @valarlabs @pearvc @a16z @jschildt @HummingbirdVC Great to be working with you and the Hummingbird team! 🎉.
0
0
1
@prabhalah @pranavrajpurkar @visweshkrishna @StanfordAILab @StanfordPath Hey! This is being done under 40x. In practice the image seen under the microscope will be the same quality / level of blurring seen in the stitched image and there aren’t additional artifacts introduced by the method.
1
0
1
@vintweeta @suchisaria @VanAllenLab @mattlungrenMD @nevillesanjana @AuslanderNoam Likely models will start interpretable off smaller cohorts and as medical datasets hit GPT scale the field may get the best of both world i.e performance and robustness.
0
0
1
@VivHo @valarlabs @VisweshKrishna @dvrabac @DCVC @a16z Thanks to you and @pearvc for the everlasting support! Agreed boba ROI is stratospheric 🧋.
0
0
1
@mattlungrenMD has been one of the best mentors I could ask for and an amazing PI. AWS is very lucky, cant wait to see what he achieves there.
As co-director of #StanfordAIMI @mattlungrenMD helped build the AIMI Center from an abstract idea to an engine for outstanding scholarship. Congratulations Matt on your incredible achievements @Stanford! We wish you great success discovering new ways to improve human health! 🎉
0
0
1
@Aiims1742 @valarlabs @Stanford @UCSFCancer Really appreciate the support Dr. Maitra! Excited to work together on aiding treatment decisions in pancreatic cancer!.
0
0
1