Healthcare AI Guy Weekly Newsletter | 7/18

Google's AI doctor, first AI crunch, Nvidia invests in Recursion, and more


Good morning folks,

Read below to get a quick download of all the important things that happened in healthcare AI this past week:

Our Picks

Highlights if you’ve only got 2 minutes…

Google’s AI doctor

Google has a medically-focused chatbot called Med-PaLM 2 which they have been testing at hospitals including the Mayo Clinic. Trained on a curated set of medical licensing exams and demos, it can answer questions, summarize documents, and organize data. It still has some accuracy issues while diagnosing but initial reports suggest it performs more or less as well as real doctors at reasoning, consensus-supported answers, and correct comprehension.

One of the key points: customers will have total control of data, and Google can’t access it.

As the chatbot continues to improve, Med-PaLM 2 can be a valuable resource for areas or countries with more limited access to doctors. With this tech, doctors could also focus more on surgical procedures while Med-PaLM 2 could help treat more common health issues. Who knows, one day we might have an AI doctor integrated into our smart watch actively monitoring our health... (link)

First AI crunch?

Bank of America analysts estimate visits to ChatGPT were down about 11% in June to just over 51 million visitors per week. Similarly, Sensor Tower data showed a 38% month-over-month decrease in ChatGPT app downloads on iPhones in the US in June.

The first AI crunch seems like it’s starting to hit after ~1 year of hype, given that Stable Diffusion came out in August and ChatGPT in November of 2022. The drop is probably in part due to increased competition, more companies banning the use, or sensitivity to potential risks.

In terms of AI companies, some initial startups are starting to feel the increased competition — with Jasper AI, copy-writing AI, and Mutiny, website-text AI, both needing to cut staff. We’re starting to see who is just a ChatGPT “wrapper” and who actually has a competitive moat, which is hard to get in AI cause most of the tech is open source. I think everyone with an AI startup should scratch their head to figure out what their moat really is and if it’s defensible. (link)

Nvidia invests $50M in biotech Recursion for AI drug discovery

Chip designer Nvidia is investing $50M in Recursion Pharmaceuticals to speed up the development of AI models for drug discovery. Recursion’s stock went up 80% following the announcement and is up more than 100% since.

Recursion, based in Salt Lake City, UT, uses AI-powered models to identify and design new therapies and offers those models to other drugmakers, such as Roche and Bayer. The company will use its biological and chemical datasets exceeding 23,000 terabytes to train AI models on Nvidia’s cloud platform.

Nvidia's investment is the latest example of the AI pharma frenzy. Drugmakers are increasingly recognizing AI’s potential to get vital treatments to people faster and Nvidia, who provides all the picks and shovels to the AI industry, seems to have picked its horse in this race. (link)

Miscellaneous 🔍

News, podcasts, blogs, tweets, resources, etc…

  • Anthropic AI just released a ChatGPT rival - Claude 2 (link)

  • 65% of top AI companies have immigrant founders (link)

  • Scientists used AI to map a fruit fly’s brain (link)

  • AI tools are designing entirely new proteins that could transform medicine (link)

  • ML model identifies mild cognitive impairment from retinal scans (link)

  • Sapphire Ventures to invest $1bn in AI startups (link)

  • Medical AI is on a tear (link)

  • Intermountain Health to lead national consortium on AI usage (link)

  • 27% of jobs at risk from AI revolution says OECD (link)

  • Why healthcare is ready for genAI and gen AI is ready for healthcare (link)

  • What healthcare CEOs think of AI (link)

  • China unveils the final version of its genAI guidelines (link)

  • Highlights from Elon Musk’s xAI Twitter Space (link)

Deal Desk 💸 

Spotlight on latest capital raises, M&A and investments…

📈 Septerna: a biotech focused on G protein-coupled receptors based in SF, raised $150M in Series B funding. RA Capital Management led the round and was joined by various other investors. (link)

📈 Causaly: an AI platform for pharma testing based in London, raised $60M in Series B funding. Iconiq Growth led the round and was joined by Index Ventures, Marathon Venture Capital, EBRD, Pentech Ventures and Visionaries Club. (link)

📈 Recursion Pharmaceuticals: Recursion, an AI-based drug discovery platform based in Salt Lake City, received a $50M investment from Nvidia. (link)

📈 Verifiable: a health care credentialing and compliance automation startup based in Austin raised $27M in their Series B. Craft Ventures led the round and was joined by Highland Capital Partners, 137 Ventures, Cooley, The Altman Fund and Struck Capital. (link)

📈 Sonio: an AI prenatal screening startup based in Paris, raised $14m in Series A funding co-led by Cross Border Impact Ventures and Elaia. (link)

Tool Box 🧰

Latest on business, consumer, and clinical healthcare AI tools…

🔧 AiVF: AiVF, an Israeli startup offering an AI support platform for IVF treatments, just released a software called “EMA.” EMA is an AI-powered embryo evaluation software that uses an advanced algorithm to help fertility doctors select the best embryos for in-vitro fertilization, improving success rates and reducing costs. (link)

🔧 Lehigh Valley Health Network + Aidoc: Lehigh Valley Health Network is integrating three AI-based software programs into its radiology workflows at all of its medical imaging sites. The health system will use software from Aidoc, an Israeli technology company that develops computer-aided simple triage and notification systems, to help read images and analyze them for pulmonary embolism, pneumothorax and cervical spine fracture. (link)

🔧 National Health Services: The NHS developed a new health tool, called OSAIRIS, that can reduce radiotherapy waiting times by making treatment planning for cancer doctors 2.5x faster. (link)

🔧 MIT: MIT scientists have developed an automated machine-learning system called BioAutoMATED that can select and build appropriate models for biological datasets, reducing the time and expertise required for machine learning in biology research. (link)

AI Images of the Week 📸

Funny memes and pics from around the web…

Dormitories of different countries - Japan

Italy

Mexico

USA

Germany

See you next week 👋

That’s it for this week friends! Back to reading — I’ll see you next week.

— Healthcare AI Guy (aka @HealthcareAIGuy)

PS. I write this newsletter for you. So if you have any suggestions or questions, feel free to reply to this email and let me know