Tiezheng Bu

Accelerating and augmenting biology with #AI: This month, we’re exploring how Owkin has created HE2RNA – our AI model for predicting RNA expression from whole slide digital images. 👇 magic isn't it? Through its interpretable design, HE2RNA can be used as a virtual spatialization tool, providing heatmaps of any gene’s expression. This AI model can be used to predict tumour immune infiltration, characterize the spatial expression of RNA, and predict treatment-relevant mutation in disease. #AugmentBiology #AI 

110

11 Comments

5 intriguing AI developments I learned about last week: 1: Komodo Health + LangChain It’s one thing to see AI Agents and Compound AI Systems being launched in Retail etc but seeing Komodo Health build an AI assistant that “leverages the power of NLP and GenAI to provide personalized and informative responses to healthcare queries” within a highly regulated industry is significant. Big shoutout to the rockstars over at LangChain including Harrison Chase + Jacob Lee. No surprise they're involved. Link: https://lnkd.in/gjPMcH2S 2: AbbVie Gen AI enabled Enterprise Product Incubator for Capabilities (EPIC) Same as #1. This is happening in a highly regulated industry… “What does it look like to scale Generative AI to the enterprise from a grass roots level?” “Seventy-plus of the power users and builders leveraging our AbbVie Intelligence platform capabilities met to dig into the new advances in our vendor agnostic, enterprise supported, GxP ready, secure LLM powered ILIAD platform. Congratulations to Jon Stevens and the rest of the Enterprise Product Incubator for Capabilities (EPIC) team for coordinating such a great event!” - Brian Martin Chief AI Product Owner, ACOS Senior Research Fellow at AbbVie Link: https://lnkd.in/gS2RC-Q5 3: Google's NotebookLM A quote like this most definitely gets my attention - “The best software ever created for leveraging AI in your creative work.” - Tiago Forte, author of “Building a Second Brain.” Link: https://notebooklm.google/ 4: Arcade Company “Arcade enables anyone to design, purchase, and sell custom, manufacturable products with a simple text or image prompt.” “Arcade brings together the power of generative AI with a global network of top artisans to turn user ideas into personalized, physical products in a quantity of one. Arcade aims to redefine commerce by offering unprecedented personal choice, expression, and meaning in product creation. Link: https://lnkd.in/gGud-jHD 5: “Every white collar job will have an AI co-pilot. Then an AI agent.” - Angela Strange, Andreessen Horowitz “Which roles might change most? James da Costa and I took a look at the Bureau of Labor statistics to inspire some ideas.” Link: https://lnkd.in/gMJ8VASz Preview of next week? Ethan Mollick comments below about LLM enabled ad models and the reactions are likely to make it... ;) “No one has figured out how you integrate advertising with LLM replies.” “This is sort of a big deal, given that where the sloshing pool of ad dollars flows determines almost everything about the digital spaces in which we operate.” Enjoy your Monday!

3

1 Comment

Langgraphs and multi agentic flow with contextual memory can enable the best customer service. We see a future where the Lang graphs deep technology over open source platforms can enable data structuring and even a better informed ai assistants throughout the spectrum. #langchain #crewai #llm #langgraph #andrewng DocsTribe AI

3

🚀 Exciting News! With Med, we just applied for the SPC Founder Fellowship in Fall 2024, and we highly recommend that other early-stage startups do the same. The SPC Fellowship offers an incredible opportunity to connect with a vibrant community of founders, gain valuable mentorship, and accelerate your startup’s growth. If you’re an early-stage founder looking to take your venture to the next level, don’t miss out on this chance. Check out their blog in the comments to learn more about the fellowship and how to apply. #startup #entrepreneurship #innovation #fellowship #SPC #earlystage

1

4 Comments

Here’s one way to use ChatGPT in Patient Services. Well technically, not ChatGPT, but Large Language Models (LLMs). LLMs are a type of AI and they are the underlying technology that power ChatGPT. What makes LLMs so powerful? They’re really good at analyzing and processing text. You know who deals with a lot of text? Field Reimbursement & Hub teams. We’ve been exploring the boring applications for LLMs that don’t make the headlines but could be very valuable to some teams. Here’s one example: ✅ Automatically processing EOBs for HCP reimbursement If you deal with a buy & bill drug, you’re probably processing lots of EOBs to reimburse providers for copay assistance. Today, many hub teams manually receive their EOBs as faxes. But these teams get lots of faxes. So usually this fax gets added to a backlog of all faxes. Someone then reviews every fax. Whenever it’s an EOB, they have to review it and figure out if it meets the criteria for processing payment. This process is obviously very manual. And if you don’t do this in a timely fashion, HCPs get frustrated, which negatively affects their experience on your program. They also will call you more frequently to get a status update on their reimbursement. This makes your phone lines busier & can hurt your KPIs like first-call-resolution. We can use AI to automate this process. We can use LLMs to automatically detect the EOBs within your faxes. And we can review the text in the EOBs to streamline this process. We can do this in a secure, HIPAA compliant way. This lets you process your claims faster, free up your staff, and reduce the number of calls that you have to deal with. It also improves the HCP experience on your program. 📣 If you face this problem, and you’re interesting in learning more, feel free to DM me. I’ll also be at Asembia if you want to connect there. ❓If you’re in patient services, field reimbursement, or hubs, here's my question for you: Which day-to-day processes in your life do you want to see streamlined?

7

ShiftRx is announcing our $3.5M seed round today (see Axios announcement here: https://lnkd.in/gVqMiGea ) led by Work-Bench, followed on by Opal Ventures, Tau Ventures, Inflect + more. My co-founder Autumn-Kyoko C. and I started this company a little over a year ago as an attempt to fix U.S. Healthcare staffing. You can read her letter here (totally sappy but, very much true): https://lnkd.in/gnFdJy6C When I started my career as a pharmacy technician at CVS Health 10+ years ago, I was working full-time and pursuing the goal of being a PharmD. I chose this path after watching my dad battle colon cancer and being hands-on with his medications and treatment plans. Many pharmacists I worked with tried to talk me out of this goal, and told me “it’s not the same as it used to be”. But for me, this mission was very much personal. I hadn’t initially asked myself, why can’t we be the ones to change it? Around the time I got accepted into a pharmacy school, my dad passed, and I ended up taking a corporate job instead. This was one of the hardest decisions for me to make. I was unsure if the years I had spent working towards this goal had been wasted and if what I had done, had really made an impact. It was then, I committed to making the change I wanted to see. I quit my job and co-founded ShiftRx. I can say with absolute certainty today that the lessons I have taken with me were not lost, and that healthcare was not what it once was. That is why we are on a mission to make it better than before. It starts with us -- as current healthcare providers, former providers, patients, and advocates. #ShiftTogether

131

20 Comments

Incredibly excited to share our updated Healthcare AI thesis. This is not a market map – in 4,000 words, we dive deeply into the research, products, business models, and market entry points that are reimagining healthcare with AI. More specifically, we cover: • Translating AI research into real-world products • Mapping modalities (software, copilots, agents, services, diagnostics, therapeutics) to problems in healthcare • Designing AI business models that unlock billion-dollar markets • Closing current gaps in healthcare AI infrastructure • Distinguishing AI platforms from point solutions Let us know what you think ⤵️

515

67 Comments

AI startups are commanding massive investments, with 1 in every 3 VC dollars now going to AI. Notably, Safe Superintelligence Inc. (SSI), an early-stage startup, secured a $1B round, becoming the 9th AI company to achieve a billion-dollar round this year. Despite investor confidence and FOMO, many fledgling AI startups may not meet high expectations and could face significant challenges ahead. Will AI investments continue at this pace? #AI #VC #TechInvesting #Startups #Innovation

6

We’re getting to that point in a technology’s journey where we open or keep closed source. Meta just released Llama 3.1, the largest-ever open-source AI model, which apparently outperforms GPT-4o and Claude 3.5 Sonnet on several benchmarks. MZ says that Meta AI will be the most widely used assistant by the end of this year! Interesting times ahead for what is set to be the biggest tech advance since the internet. Where do you sit in terms of open and closed source and does the same rule apply for AI? #ai #genai #chatgpt #metaAI

11

1 Comment

This study is another AI milestone: an RCT demonstrating clear and significant (used in both the colloquial and statistical sense) advantage to using an AI alert intervention system. Fore example, in high-risk ECG patients, there was a 40%(!) reduction in all-cause mortality. There are still several important caveats to consider, but as these systems begin outperform traditional methods for caring for patients, the ethical discussion shifts: Is it justifiable to *avoid* using AI in healthcare when it offers such clear benefits to patients? Shoutout to Ran Balicer for bringing this groundbreaking research to my attention! #HealthTech #ArtificialIntelligence #PatientCare https://lnkd.in/gJ_nkYQe

17

1 Comment

“... native solutions built entirely around vectors will provide the ‘speed, memory safety, and scale’ needed as vector data explodes.”  “... we do advanced vector search in the best way possible.” At Epsilla (YC S23), we share the same vision and approach. Thus we have been leading the charge with our unique innovations in a 10x faster, cheaper, and better vector search engine, as the built-in foundation for our no-code RAG as a Service Platform. You can experience the accuracy and relevancy from our advanced technologies via a number of high quality AI-powered smart search and chatbot applications below: 1. Smart Search on 180+ RAG Research Papers (https://lnkd.in/eMPdVae5): This smart search app leverages our RAG platform and vector database technology to help user navigate through the massive 180+ papers with ease and keep up with the latest and greatest advanced research on this important topic. . 2. AI Assistant Chatbot for Knowledge Graphs (https://lnkd.in/eeBTFe_Q): Inspired by insights from Mike Dillinger's posts, this AI assistant chatbot leverages our RAG platform and vector database technology to generate highly relevant, nuanced and contextually aware responses to help user stay ahead of the game on topics like LLM, knowledge graph, and GenAI. 3. AI Assistant Chatbot for Taylor Swift (https://lnkd.in/e_m6dB4P): Powered by our RAG platform and vector database technology, this interactive AI-powered chatbot can answer any questions about Taylor Swift, one of the most impactful artists and successful business women in the world. These high quality AI applications underscore the fact that  Epsilla's vector database technology is not just keeping up with the demands of generative AI, but also setting new standards for what is possible. In addition, our no-code RAG platform allows AI app builders to develop and deploy high quality apps like these within hours. Join us as we continue to innovate and lead in this exciting era of AI-driven applications! (https://epsilla.com) #VectorDatabase #RAG #Chatbots #GenerativeAI #NoCodePlatforms

31

8 Comments

With the release of the Meta Llama 3.1 series of LLMs, previously proprietary-only levels of language intelligence is now open weights! How has this development impacted your organization's AI strategy? We're moving from just prompt engineering to a full fine tuning workflow to create custom models for our classifier needs! Without this quality baseline training wasn't feasible for us due to costs. Now, custom models are approachable for small enterprise applications. Can't wait to see the next wave of innovation these open models will drive!

1

Trying to run LLMs locally? 🤔 Our course Enterprise RAG and MultiAgents Applications, covers all that and much more. Ever wondered how to use quantized LLMs like TheBloke's GGUFs on platforms like Hugging Face ? 🧐 Here's the deal: quantizing a model allows for compression, making it faster for inference with fewer bits. However, the more quantized the model, the lower the accuracy and sophistication it can provide. 🤓 Good news! Ollama lets you load a local GGUF model effortlessly. Simply obtain the GGUF files, store them, write a Modelfile, and voilà! You're all set to use it on Ollama. 🚀 From loading local models to becoming an (LLM) hero, our notebooks and lectures in Enterprise RAG and MultiAgents Applications are now available on Maven. Join us there to learn more! 📚💻 Signup for our course: https://lnkd.in/g7Z-7SAU #LLM #QuantizedModels #Ollama #EnterpriseRAG #MultiAgentsApplications #Maven

50

1 Comment

"OpenAI's o1: The AI That Thinks Before It Speaks 🧠💡" 🚀 OpenAI has just unveiled o1 (codename: Strawberry), a groundbreaking AI model that's changing the game in machine reasoning. Here's what you need to know: 🔍 What is o1? • A large language model (LLM) designed to excel in complex reasoning and problem-solving • Utilizes a novel reinforcement learning approach, moving beyond traditional pattern recognition 🧠 How was it trained? • Employs large-scale reinforcement learning, teaching the AI to "think productively" • Learns to improve its own chain of thought, leading to more robust reasoning capabilities 💪 Capabilities: • Outperforms humans in competitive programming (89th percentile on Codeforces) • Matches top 500 US students in math olympiad qualifiers • Exceeds PhD-level accuracy in physics, biology, and chemistry problems • Demonstrates multimodal prowess, combining text and visual information processing 🔬 Key Features: • Chain-of-thought reasoning: Breaks down complex problems into manageable steps • Self-correction: Recognizes and rectifies mistakes during problem-solving • Scalability: Performance improves with increased compute power and thinking time 🌟 Why it matters: • Represents a significant leap in AI's ability to tackle complex, multi-step problems • Opens new possibilities in fields requiring deep reasoning like science, mathematics, and coding • Paves the way for more sophisticated AI assistants in various industries 🤔 What's next? • Potential applications in healthcare, research, education, and more • Ongoing development to enhance safety and alignment with human values • Integration of o1's reasoning capabilities into more accessible AI tools The o1 model is not just another incremental step in AI—it's a giant leap towards machines that can truly think and reason. This is the first real step towards AGI. #AI #MachineLearning #OpenAI #Innovation #TechNews

1

Imagine a world where doctor-patient interactions are led by empathy, understanding, and their efficiency is enhanced 💪 . I'm thrilled to share that I have joined Roi Shabshin and Claire Lim to co-found Script AI, a company with a profound mission to redefine healthcare. At Script AI, we are dedicated to transforming doctors' operations, enhancing patient care, and unlocking new medical possibilities. The reality we're creating with Script AI is one where doctors no longer will have to do paperwork (!), and patients receive the best possible care. Our AI-powered platform, which strictly sticks to the highest data privacy and security standards, captures patient conversations in real-time, automating documentation and streamlining clinical processes, leading to a more efficient and effective healthcare system. Exciting times lie ahead as we revolutionize healthcare. Stay tuned for more updates on our journey to reshape the future of medicine! #HealthcareInnovation #AIinHealthcare #HealthTech #DigitalHealth #PatientCare #EmpathyInMedicine #MedicalTechnology #HealthcareRevolution #FutureOfMedicine #EfficientHealthcare #HealthcareTransformation #HealthcareTech #MedicalInnovation #AIForGood #EmpatheticMedicine #DataPrivacy #MedicalAutomation #ClinicalEfficiency #HealthcareLeadership

67

13 Comments

That's a good summary Manny Bernabe - Reflecting on an insightful article from O'Reilly about LLMs, I was struck by this point: "LLMs are also broadly accessible, allowing everyone, not just ML engineers and scientists, to build intelligence into their products. While the barrier to entry for building AI products has been lowered, creating those effective beyond a demo remains a deceptively difficult endeavor." This democratization reminds me of the printing press's impact on biblical texts, which expanded access but also introduced new interpretive challenges. As we navigate these advancements, it's crucial to balance accessibility with responsible use. At Ushur, we've been at the forefront of leveraging LLMs in our backend/knowledge work automation as well as carefully introducing them on the conversational side via our Chatbots and other modes. More will be coming out on that. Read more here: O'Reilly Article #AI #LLM #Technology #Innovation #History #Democratization #ResponsibleAI #Ushur

12

1 Comment

building AI in pathology is becoming effortless. like sending a text vs sending a telegraph image foundation models are getting super powerful, fast our AI research scientists have been using this the last year. results have been 🤯 * 13x faster AI development for biomarker prediction * 80 models in <24 hours (on a laptop) * processed WSIs in 2.5 hours vs. 33.4 hours * embeddings = super lightweight substrate for computation (203GB > 19 GB) (91% reduction vs image) we think it's time for the world to have access to this

60

1 Comment

"Med42-v2 Released: A Groundbreaking Suite of Clinical Large Language Models Built on Llama3 Architecture, Achieving Up to 94.5% Accuracy on Medical Benchmarks Researchers from M42 Abu Dhabi, UAE, have introduced the Med42-v2, a suite of clinical LLMs built on the advanced Llama3 architecture. Developed by the team at M42 in Abu Dhabi, these models are meticulously fine-tuned using specialized clinical datasets, making them particularly adept at handling medical queries. Unlike generic models, which are often preference-aligned to avoid answering clinical questions, Med42-v2 is specifically trained to engage with such queries, ensuring that it can provide relevant and accurate information to clinicians, patients, & other stakeholders in the healthcare sector. The development of Med42-v2 involved a two-stage training process designed to optimize the models for clinical use. The Llama3 models were fine-tuned in the first stage using a curated dataset that included medical and biomedical information, chain-of-thought reasoning, and conversational examples. This stage accounted for 26.5% of the final training dataset and enhanced the models’ ability to understand and generate responses relevant to clinical contexts. The second stage focused on preference alignment, ensuring the models’ outputs aligned with human expectations and ethical standards. This stage utilized datasets such as UltraFeedback and Snorkel-DPO, allowing the models to be iteratively refined to meet clinical requirements..... Read full take on Med42-v2: https://lnkd.in/gC-NzsUt"

4

September 27 2024 – 5 Notable Articles To Read Article: Google DeepMind AI drug discovery spinout Isomorphic Labs issues £182M in new shares Takeaway: The UK-based company, which has created an AI-driven drug discovery platform for the pharmaceutical and biotech industries and is ultimately owned by Alphabet Inc., Google's parent company, filed the document earlier this month. Source: Sifted Link: https://lnkd.in/desGyh_b Article: Germany unveils €12B initiative to bolster startup scene Takeaway: Politicians aim to raise €12B from both public and private investors by 2030. The WIN Initiative, announced earlier this year, plans to secure funding from public and private capital sources, including Deutsche Bank, Allianz, Commerzbank AG, Deutsche Telekom, AXA Deutschland, and insurance companies such as HUK-COBURG. Source: Sifted Link: https://lnkd.in/danWh2tr Article: Schroders Capital receives approval to launch UK venture and growth LTAF Takeaway: Schroders Capital, the $97.3B specialist private markets arm of Schroders Group, has obtained regulatory approval from the UK’s Financial Conduct Authority to launch its third Long-Term Asset Fund (LTAF), and the first focused exclusively on UK VC. Source: Funds Europe Link: https://lnkd.in/dzp8gk3T Article: As AI Chip Startups Sell to Mideast, More Hire in Region Takeaway: Chip startups SambaNova Systems, Cerebras Systems, and Groq, aiming to capture a share of the $71BAI chip market, are targeting sales to Middle Eastern countries with substantial financial resources and ambitious AI goals. Source: The Information Link: https://lnkd.in/dAbR3mUr Article: Alaska Airlines completes its acquisition of Hawaiian Airlines, which will remain a separate brand Takeaway: Alaska Airlines finalized its $1B acquisition of Hawaiian Airlines, just a day after the federal government cleared the final regulatory hurdle. As part of the deal, Alaska will also take on approximately $900 million of Hawaiian's debt. Source: The Associated Press Link: https://lnkd.in/d4vEwvsS #news #research #business #ai

7