Healthcare AI Guy Weekly Newsletter | 8/1

Our Picks ✨

Highlights if you’ve only got 2 minutes…

Amazon announces new healthcare AI service — AWS Healthscribe

Amazon Web Services unveiled a range of new AI products, including AWS HealthScribe which helps healthcare providers summarize doctor visits.

HealthScribe enables healthcare software developers and providers to create clinical applications that use speech recognition, AI capabilities, and advanced ML algorithms to help generate clinical documentation and enrich provider workflows. By offering a single API and eliminating the need to manage complex ML infrastructure, this service empowers healthcare software providers to build clinical applications rapidly and efficiently.

This move puts Amazon/AWS in closer competition with its rivals, as Microsoft, which acquired health AI company Nuance last year for $20B, is already using ChatGPT technology to sell transcribing and summarizing services to doctors and nurses. Meanwhile, Google is working on Med-PaLM, an AI model that can help with real-time clinical support.

The healthcare industry is shaping up to be a battleground for generative AI and this was a significant move by Amazon to persuade providers to run their own custom-built AI apps on AWS. Either way, the winners here are healthcare practitioners as the burden of clinical documentation is one of the main drivers of burnout. (link)

Google Cloud & CareCloud to develop AI solutions for outpatient providers

Google Cloud and CareCloud, a health tech company with a suite of solutions based in New Jersey, announced a partnership to develop generative AI solutions for outpatient/ambulatory providers!

The collaboration will enable AI technology and capabilities to benefit small and medium-sized healthcare practices and outpatient providers. CareCloud’s experience with downstream providers and extensive clinical and financial data sets, combined with Google Cloud’s AI stack and capabilities, can be brought to life and be used in meaningful ways via search and recommendation generation.

Potential use cases for the technology:

• Improving the pre-visit patient check-in process with smarter intake questions.

• Providing tailored recommendations and evidence-based guidelines curated to the patient.

• Concluding with providing billing and financial information post-visit.

Overall, this is an incredible benefit for ambulatory healthcare practices. While the majority of tech giants have focused on providing AI to large organizations, this provides access to an often overlooked yet critically important sector of healthcare. (link)

Google unveils Med-PaLM Multimodal (Med-PaLM M), a proof of concept for a "Generalist Biomedical AI"

Google Health, Google Deepmind and Google AI have unveiled Med-PaLM M, a large multimodal generative model that flexibly encodes and interprets biomedical data. It can handle various types of medical data, including clinical language, medical images, and genomics, and performs well on a wide range of tasks, all using the same set of model weights.

It was built by fine-tuning and aligning PaLM-E, a language model from Google AI, to the medical field using a specially curated open-source benchmark called MultiMedBench. MultiMedBench consists of 7 biomedical data types and 14 diverse tasks, such as medical question-answering, generating radiology reports, and identifying genomic variations. With over 1 million samples, this benchmark encourages the development of generalist biomedical AI systems.

Med-PaLM M excels in all tasks on MultiMedBench, often outperforming specialist models by a significant margin and even surpassing PaLM-E, proving the importance of adapting the model to biomedical data.

The key idea behind building a large-scale biomedical AI is to use language as a common framework for different tasks. This allows the AI to combine knowledge from various sources and transfer skills across tasks more effectively.

Excitingly, preliminary evidence suggests that Med-PaLM M can generalize to new medical tasks and concepts and perform multimodal reasoning without specific training. It can accurately identify and describe medical conditions in images using only language-based instructions and prompts, even if it has never seen such cases before.

The possibilities with such generalist biomedical AI that can encode the biomedical universe are limitless with applications spanning the continuum of scientific biomedical discovery to care delivery. (link) / (tweet)

Miscellaneous 🔍

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

• Overview of AI scribe landscape (link)

• Google, OpenAI, Microsoft, and Anthropic form Frontier Model Forum to develop safety standards for advanced AI systems (link)

• Hunna Technology, a UK healthtech startup, unveiled the first-ever AI to act as CEO of a company in Europe (link)

• Cigna says claims review process does not rely on AI/Algorithm (link)

• Several biotech stocks benefit from AI-fueled rally (link)

• AWS HealthImaging promises to ease medical imaging (link)

• ChatGPT comes to Android (link)

• 61% of Americans think AI threatens humanity’s future (link)

• AI can recreate voices otherwise lost to disease (link)

• Google announced RT-2, an AI model that translates both vision and language in robot motion (link)

Deal Desk 💸 

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

📈 RapidAI: a vascular and neurovascular conditions diagnostics company that helps diagnose strokes based in San Mateo raised $75M in Series C funding led by Vista Credit Partners. (link)

📈 Proprio: an AI-driven surgical navigation platform that generates a real-time 3D visualization of the surgery just raised $43M in Series B funding from new and existing investors, including Bird B. (link)

📈 A-Alpha Bio: a synthetic biology and ML company based in Seattle raised $22M in additional Series A funding. Perceptive Xontogeny Ventures led the round with Madrona and Breakout Ventures participating. (link)

📈 Hippocratic AI: a healthcare-focused generative AI startup, raised an additional $15M in seed funding, totaling to $65M. They also announced a handful of partners, including Cincinnati Children's Hospital, Universal Health Services, HonorHealth, Vital Software, ELNA, Sondermind and Capsule. (link)

📈 K4Connect: a senior-focused health tech company based in North Carolina raised $9M in funding. Bryce Catalyst and AXA Venture Partners led with Intel Capital, Forté Ventures, Topmark Partners, and the Ziegler-Linkage Fund also participating. (link)

📈 ReflexAI: a provider of AI-powered training and quality assurance tools for high-stakes call centers based in New York, raised $3.3M in seed funding. Footwork led the round with Emerson Collective, Altman Capital, Gaingels, and angel investors also participating. (link)

Tool Box 🧰

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

🔧 Cardiff University: Cardiff University researchers created an AI system to help doctors detect cancer by learning from radiologists' eye movements when analyzing medical images. The AI aims to support and enhance radiologists' decision-making without replacing them, potentially leading to earlier cancer detection. (link)

🔧 Monash University + Cortical Labs: A team of researchers from Monash University, based in Melbourne, Australia, and Cortical Labs, a biological computing startup, received funding from Australia's Office of National Intelligence for a research project exploring merging human brain cells with AI. Successful experiments show brain cells playing "Pong" in a Petri dish. The integration could revolutionize machine learning and grant significant strategic advantages in various fields. (link)

AI Images of the Week 📸

Funny memes and pics from around the web…