Can AI rewrite our human genome?🧬

Hey!

Welcome to this week edition of Supermedic!

Moderna integrates ChatGPT into its organization, and its CEO wants his employees to use it 20 times a day. The first Benchmark for GenAI applies to medical tasks is out. Human gene therapy leaps forward with AI-powered editor but ethics issues remains.

Let’s get into it!

Victor

TODAY’S MENU

• Moderna Partners with OpenAI (VIDEO)

• Profluent Successfully Edits Human Genome with AI

• ChatGPT Shows Promise in Polypharmacy Management (STUDY)

• AI Can Now Evaluate Cardiovascular Risk From CT Scan

• Huggingface Releases Benchmark to Test GenAI on Health Tasks

Read time: under 5 minutes

AI + BIOLOGY

Moderna Partners with OpenAI to Supercharge mRNA Research

Moderna, Biotech leader, announced a partnership with OpenAI, the company behind ChatGPT, to further incorporate AI across its mRNA drug development and manufacturing operations.

Key Takeaway:

• Stephane Bancel, CEO of Moderna, envisions a complete overhaul of their business processes through AI, enabling their workforce to be 30x more productive.

• Moderna has already deployed 750 GPTs (custom-designed), with employees engaging in an average of 120 ChatGPT conversations weekly.

• The collaboration empowers Moderna's custom-designed GPTs to tackle a wide range of tasks, from optimizing drug dosages to streamlining branding and internal communications.

Looking Further:

This collaboration has the potential to help Moderna outpace its plan to roll out 15 new products within the next 5 years but also serve as a roadmap for other companies to integrate ChatGPT into their own operations.

THEY SUPPORT US ❤️

85% of all AI Projects Fail, but AE Studio Delivers

If you have a big idea and think AI should be part of it, meet AE.

We’re a development, data science and design studio working with founders and execs on custom software solutions. We turn AI/ML ideas into realities–from chatbots to NLP and more.

Tell us about your visionary concept or work challenge and we’ll make it real. The secret to our success is treating your project as if it were our own startup.

Talk to Us Now

FUTURE IS NOW

Profluent Successfully Edits Human Genome with AI

Profluent, a californian based startup, claims to have developed the world's first AI-made gene editor, called OpenCRISPR-1.

How it works:

• Profluent has trained large language models (LLMs) on a massive dataset of diverse CRISPR-based gene editing systems. This allowed the AI to learn and generate novel gene-editing proteins that do not occur naturally.

• OpenCRISPR-1 is capable of targeting and modifying DNA sequences within human cells. This enables more accurate editing of genes responsible for various hereditary conditions and diseases.

• OpenCRISPR-1 performances are on par or better than a naturally occurring editor.

Why it matters:

The introduction of OpenCRISPR-1 represents a significant leap forward in gene therapy. The hope is that the technology will eventually produce gene editors that are more nimble and more powerful than those that have been honed over billions of years of evolution. However, there are also some ethical concerns about the use of AI-powered gene editing, such as the potential for misuse and the unintended consequences of altering the human genome.

CHATBOT

ChatGPT Shows Promise in Polypharmacy Management for Elderlies

A new study suggests that AI could be a valuable tool for managing polypharmacy, the use of five or more medications in older adults. This is important because polypharmacy increases the risk of adverse drug interactions, but deprescribing unnecessary medications can be complex and time-consuming for primary care providers.

Researchers from Mass General Brigham tested ChatGPT on various clinical scenarios involving elderly patients on multiple medications. The AI was asked yes-or-no questions about reducing medications.

Here's what they found:

• ChatGPT consistently recommended deprescribing for patients without a history of heart disease.

• It was more cautious with patients who had heart disease, suggesting they keep their current medications.

• The AI seemed to prioritize deprescribing pain medication over other types of drugs.

• There were some inconsistencies in its recommendations, possibly reflecting the data it was trained on.

CARDIOLOGY

AI Can Evaluate Cardiovascular Risk During CT Scan

Researchers at Cedars-Sinai Medical Center have developed AI algorithms that improve cardiovascular risk assessment from routine chest CT scans, initially performed for lung cancer screenings.

How It Works:

• Utilizing two distinct AI algorithms, the team efficiently analyzes low-dose chest CT scans. One algorithm measures coronary artery calcium (CAC) scores while the other evaluates cardiac chamber volumes.

• These algorithms process nearly 30,000 patient scans, delivering CAC scores and chamber volumes in just seconds, with a remarkably low failure rate of only 0.1%.

Why It Matters:

This AI approach provides a non-invasive, accurate method to assess cardiovascular risks, surpassing traditional methods by radiologists and potentially reducing the need for invasive tests.

GENERATIVE AI

Open Medical-LLM: The First Medical-LLM Benchmark

Hugging Face, an AI startup, teams up with researchers from Open Life Science AI and the University of Edinburgh’s NLPG, to develop Open Medical-LLM, a benchmark test for evaluating the performance of Large Language Models on a range of medical-related tasks. It combines existing test sets designed to evaluate models' capabilities in core medical areas like anatomy, pharmacology, genetics, and clinical practice, along with general medical knowledge.

The ranking:

Med-PaLM 2 (Google & Deepmind) and GPT-4 (OpenAI) are far ahead from all other Medical-LLM, ranking higher in all types of medical tests.

Caution Advised: Open Medical-LLM is a useful tool for evaluating the performance of Medical-LLMs but it is not a substitute for careful real-world testing. Ultimately, the decision of whether or not to use a Medical-LLM in a clinical setting should be made on a case-by-case basis, taking into account the specific risks and benefits of the model.

Thanks for reading!

Let’s improve the newsletter together!🌟