#6 | First UK Patients Receive mRNA Cancer Therapy

Quote of the Week 💬 

First UK Patients Receive mRNA Therapy For Cancer

⌛ The Seven Second Summary: The first patients in the UK have now received an mRNA therapy as part of a phase 1/2 clinical trial. The trial aims to evaluate the gene therapy’s safety and its potential to treat melanoma, lung cancer, and other 'solid tumor' cancers.

🔬 How It Was Done:

• The mRNA therapy introduces common tumor markers to a patient’s immune system. Once administered, the therapy trains the patient’s immune system to recognize and fight off cancer cells expressing these specific markers, so the patient’s body is more adept at eliminating suppressive tumor cells with these same markers throughout the body.

• Trial patients will receive either the mRNA therapy by itself, or the mRNA therapy as well as an existing cancer drug.

• Trial patients will be monitored for up to 34 months to evaluate the safety and effectiveness of the mRNA treatment option.

💡 Why This May Matter: Cancer vaccines are entering clinical trials globally as an emerging treatment strategy to offer more effective ways for a patient’s immune system to combat cancer with fewer side effects. By the end of the decade, these studies may provide evidence of safe and promising treatment options for cancer patients to live longer and more fulfilled lives after diagnosis.

🔎 Elements To Consider: This is neither a randomized trial nor an open-label trial. Therefore, all patients are receiving the same treatment, and both clinicians and patients are aware of the treatment being administered. If the gene therapy advances to Phase III trials, subsequent study designs should employ randomized, blinded trials to provide stronger assessments of causality for this therapy when compared to alternative treatment options available to patients today.

📚 Learn More: Imperial College London.

Stat of the Week 📊 

A 2000-Year-Old Greek Scroll Is Deciphered By AI

2,000 years old

⌛ The Seven Second Summary: The Vesuvius Challenge is an international competition where advanced technologies are used to decipher ancient texts, and its 2023 winners have been announced.

🔬 How It Was Done:

• 2,000 years ago, a volcanic eruption buried an ancient library of papyrus scrolls now known as the Herculaneum Papyri.

• About 800 Papyri scrolls have since been recovered, but they are so damaged from the eruption they look like lumps of ash and crumble if anyone attempts to roll them open.

• As an alternative, 3D CT images of the scrolls were created and in March 2023 a competition was launched where a community of researchers utilized computer vision and machine learning to detect and the black ink that was used to write on the scrolls from the deep charcoal color the scrolls are now left in.

• Eventually, the machine learning models predict where the black ink is throughout a scroll, and other methods are utilized to verify the accuracy of the machine’s predictions.

🧮 Key Results:

• The criteria of eligibility for the 2023 prize was to decipher at least 85% of four passages containing 140 character each. The winning entry far exceeded these expectations by deciphering more than 2000 characters written on a scroll. This amounts to about 5% of the first scroll in its entirety.

• Now that there is a foundation for how the research community can solve this problem computationally, the 2024 Vesuvius Challenge is to decipher 90% of four ancient scrolls.

• Afterwards, it should be possible to decipher all 800 Papyri scrolls that have been recovered soon after.

💡 Why This May Matter: This technological breakthrough may be a watershed moment for archeology and other classical studies to recover and understand ancient text throughout Egypt, Mesopotamia and other parts of the historical world. What might we learn as we read the messages from past civilizations?

🔎 Elements To Consider: Over $1M of prizes was awarded to winning participants throughout this challenge’s various progress prizes, with $700,000 going towards the winning team. This research may not have received as much community interest, or achieved its results as quickly, if the prize money was awarded as a research grant to academic institutions instead. Thus, the structure of this competition may be an example to emulate for other research funding in the future.

📚 Learn More: Scroll Prize. Github.

AI x Science 🤖

New AI Framework Significantly Improves Machine Reasoning

Google Research and USC researchers released SELF-DISCOVER, which is a framework to enable Large Language Models (LLMs) to determine the most appropriate reasoning methods to solve complex tasks more effectively. Why? Well, presumably, every conceivable task requires a specific reasoning structure to be successfully completed. For instance, mathematical thinking is used to solve financial or physics-based problems. Emotional intelligence is used to consider how decisions may affect other people and their feelings. Analytical thinking is used to evaluate and prioritize different variables or options that may exist in various scenarios. The list goes on.

With this in mind, SELF-DISCOVER’s framework involves a self-discovery process for LLMs to identify the unique reasoning structure needed to solve a specific task. This is done by assessing a task’s core components and determining what reasoning methods are required to complete each component, such as creative thinking, critical thinking or step-by-step thinking. Once a reasoning approach has been determined for all the components within a task, the system produces a structured set of steps for the model to follow in order to comprehensively solve the task at hand.

When SELF-DISCOVER’s framework was applied on LLMs like GPT-4 and PaLM2, the models experienced up to 32% performance improvements at solving reasoning tasks. Furthermore, SELF-DISCOVER outperformed other state-of-the-art reasoning methods for LLMs by more than 20% while operating 10-40x more efficiently.

These results are significant because they introduce a novel approach for models to improve performance while significantly reducing computational burdens. It may not be long until we reach a new inflection point and witness another remarkable display of capabilities by these machines. arXiv. 

Our Full AI Index

• Research: Researchers from Moffitt Cancer Center developed a model called IRMMa to predict how well a patient with multiple myeloma may respond to various cancer therapies. The model assesses clinical, demographic, genomic, and therapeutic data to determine its predictions, and outperformed all other prognostic models it was compared against when analyzing patient data from a Phase III clinical study. Moffitt. ASCO. Github.

• Open Source: Hugging Face launched Hugging Chat Assistant, which is a free and open-source alternative to the GPT Store OpenAI released last month to allow its users to build personalized AI chatbots. Hugging Face. Venture Beat.

• Business: Apple and UC Santa Barbara researchers released MGIE, an open-source AI system to edit images via natural language commands. This research was released shortly after Apple’s Q1 earnings call, where CEO Tim Cook mentioned the company will share AI announcements ‘later this year’. Perhaps this is an indication of what the company’s forthcoming updates will be about? arXiv. Github.

• AI Safety: The U.S. Department of Commerce announced the launch of an AI Safety Institute Consortium support of the development and deployment of safe and trustworthy AI. The consortium includes over 200 organizations, and unites civil society institutions, as well as academics and government and industry researchers from OpenAI, Kaiser Permanente, Stanford, MIT and elsewhere to support the safe development of AI. U.S. Department of Commerce. AISIC Members.

• Policy: In an effort to reduce the rise of scams from bad actors, the U.S. Federal Communications Commission announced a new ruling which makes AI-generated voices in robocalls illegal. FCC. 

Other Observations 📰

Using Black Solar Panels To Make Rain In The Desert

Since 2010, the United Arab Emirates (UAE) has been exploring climate engineering methods as one way to support the region’s water scarcity needs. One of the UAE’s most commonly known climate engineering techniques is cloud seeding, which is when they use an aircraft to spray particles into passing clouds to trigger rainfall. However, according to a new study, black, city-sized solar farms may be another way to provide more rainfall for the region.

The research explains how the heat emitted by large artificial black surfaces, such as massive solar farms which are already in the process of being built, affects localized temperature conditions and lead to more rainfall. The study highlights a correlation between the size of these black surfaces and their impact on precipitation levels. They also suggest solar farms exceeding 20 km in size represent the lower threshold to influence rainfall outcomes.

The researchers estimate a single 20 km black solar farm could enhance the downpour of 10 rainfall events per year, which would provide enough extra yearly rainfall to benefit over 30,000 people. Similarly, a 50 km solar farm may yield enough extra rainfall to support over 125,000 people, suggesting a promising solution to provide much-needed rainfall in the world’s most arid regions. EGU.

Our Full Science Index

• Accessibility: A team of researchers from various universities developed a temperature-sensitive prosthetic limb, enabling amputees to more accurately perceive the different temperatures and materials of the objects they touch. Sant’ Anna. Cell Press.

• Energy: The International Energy Agency launched Global Observatory, a new tool designed to track countries' progress in their clean energy transitions. The goal of this tool is to identify best practices worldwide, to support governments in implementing more effective clean energy policies in their respective jurisdictions. IEA.

• Phase III Cancer Trials: GSK shared findings from an interim analysis of their study results, showing its antibody-drug, Blenrep, when used in a three-drug regimen, led to nearly triple the progression-free survival rates in patients with relapsed or refractory multiple myeloma, from 13.4 months to 36.6 months. Furthermore, their results show Blenrep reduced the risk of disease progression or death in patients by 59%. An amazing development if these trends continue over time. GSK.

• Space: NASA’s latest satellite mission to study ocean health, air quality, and the effects of a changing climate launched successfully into orbit last week. The PACE mission will study microscopic life in water as well as microscopic particles in the air from hundreds of miles above the Earth. Its insights will inform scientists about how the how the ocean and atmosphere work together, and how climate change may affect these conditions over time. NASA.

• Public Health: Guyana’s Minister of Health announced progress in the country’s efforts to eradicate filaria, leprosy and tuberculosis from the nation. The Minister mentioned tuberculosis cases have dropped from over 500 a few years prior, to just 5 cases in 2023. Loop News.

• Policy: Last week, the EU Commission announced details for its plan to reduce net greenhouse gas emissions 90% by 2040 when compared to 1990 levels. European Commission. 

Media of the Week 📸 

Autonomous Robots Completing Tasks Around The Office

1X released a demo of their autonomous android robots navigating an office environment, opening doors, tidying shelves, and interacting with people without any manual intervention. Is this what work environments will look like in our lifetime?

Disney’s Newest Robotic Character For Its Resorts

We couldn’t resist another robotics video. Last December Disney engineers unveiled their latest robotics character to visitors at their Shanghai Resort. This clip is a behind-the-scenes look at how Duke Weaselton was tested and brought to life to create the magical storytelling experiences the company’s resorts are known for.

This Week In The Cosmos 🪐

No major astronomical events this week. How sad…

Until Next Time 💭