Language Models For Healthcare & CRISPR for Cows
Read about how generative language models could transform healthcare and CRISPR can be used on cows to curb greenhouse gas emissions
Generative Language Models Transforming Patient Care
Tech giants and startups alike are eagerly exploring the potential of generative AI tools in medical settings, with Google recently announcing the release of its medical large language model (LLM), Med-PaLM 2, to a select group of users. LLMs, for those unfamiliar, are AI tools that specialize in understanding and generating language. Med-PaLM 2 has been developed specifically for the medical domain, aiming to provide more accurate and safe answers to medical questions.
This move by Google comes on the heels of Microsoft's own foray into generative AI tools for healthcare, with the release of a new Azure Health Bot template. As these tech giants continue to innovate, it's clear that the race is on to harness the power of generative AI in the medical field. Google has been working closely with clinicians and non-clinicians alike to assess Med-PaLM and Med-PaLM 2, focusing on aspects such as scientific consensus, medical reasoning, knowledge recall, and potential harm. In addition to Med-PaLM 2, Google has also introduced new AI-enabled tools to streamline health insurance prior authorization and claims processing. These tools transform unstructured data into structured data, allowing experts to make faster decisions and ultimately improving access to timely patient care.
Doximity, a digital platform for medical professionals, has joined the race as well, releasing a beta version of a ChatGPT tool for doctors. This tool assists with time-consuming administrative tasks, such as drafting and faxing preauthorization and appeal letters to insurers. Abridge, a startup providing AI-powered medical transcription services, has deployed one of the largest generative AI projects in healthcare to date. The University of Kansas Health System has implemented Abridge's tools to assist over 1,500 physicians across more than 140 locations. This technology listens to patient-doctor visits and smartly summarizes key parts of the conversation for both parties.
Despite the enthusiasm surrounding generative AI, caution is still advised in its application within healthcare. The technology can sometimes "hallucinate" or invent a response when it lacks sufficient information. However, a recent survey by Huma.AI reveals that 86% of medical affairs leaders believe generative AI can be applied in life science companies. Generative AI has the potential to transform medical affairs activities, enhance engagement with healthcare professionals, and ensure products are used safely and effectively. It's important to keep in mind, though, that generative AI is not meant to replace human expertise but to augment and support human decision-making.
The recent surge of interest and innovation in generative AI within the medical field highlights the potential for significant advances in research and patient care. As tech giants and startups continue to develop and refine these tools, it will be crucial to strike a balance between the capabilities of generative AI and the need for human expertise in healthcare. The ultimate goal is to create a more efficient and effective healthcare system for all.
Source:
CRISPR on cows
Cow burps are responsible for over 27% of the methane that is released into the atmosphere as a result of human activity. When cows digest food, the microbes in their guts convert grass or hay into energy and release methane as a byproduct. The sheer number of cows that humans raise for beef, milk and other dairy products makes cows a major source of greenhouse gasses especially considering the fact that methane traps 25 times more heat than carbon dioxide.
With the advent of CRISPR, a gene editing technology that allows scientists to alter the genome of an organism by adding or removing DNA sequences, an exciting opportunity has opened up to explore the use of CRISPR in trying to genetically alter cows to make them produce less methane.
CRISPR has been around for some time and used in labs and on cells that are grown in dishes, there are some small animal use cases and clinical trials but CRISPR has not been used in non-controlled environments yet. Additionally, it is difficult to get the CRISPR machinery into bacteria, in a lab you can give bacteria an electric shock and get the CRISPR machinery inside of it but you can not do that when the bacteria is in a cow.
This is one of the big challenges that scientists at the Innovative Genomics Institute are working towards addressing. They need to find a way to get the CRISPR machinery inside the bacteria in a cow and also find appropriate genetic modifications to reduce how much methane bacteria in the gut microbiome produces. People have been able to use viruses as delivery vehicles for CRISPR in plants and bacteria but this is usually done in a pure colony of one bacteria and not a mixed population like what you would have in the gut microbiome.
Ideally, it would be a sort of pill that can be given to the cow when it is a calf and have a permanent modification in the calf so that a repeat dose or repetitive dosing is not required in the future. This is an exciting area of research and if successful can help us curb one of the biggest sources of greenhouse gasses.
Featured Fake News
After nearly 5 months since Dr. Fauci officially retired from the position of Director of NIAID, social media users have started circulating the claim that we have zero new COVID-19 strains since his retirement.
First of all, that is not true. The current strain which accounts for >80% of COVID-19 cases and the most transmissible strain to date was already spreading rapidly while Dr. Fauci was starting his retirement. Additionally, on March 30, 2023 WHO announced a new omicron sub lineage of interest currently called XBB.1.16.
Secondly, it is impossible for new variants to stop showing up unless the pathogen goes extinct. That is just how biology works. There will inevitably be mutations in the viral genome as they are replicated which will give way to new variants. This is true for both real organisms and laboratory engineered ones.