NASA opened applications for people to live a paid year in its CHAPEA Mars simulation. The job requires an agreeable personality and STEM degree.
Category: health – Page 84
How can water-based batteries help improve lithium-ion energy storage and technology? This is what a series of studies published in Advanced Materials, Small Structures, Energy Storage Materials, and Energy & Environmental Science hopes to address as a team of international researchers led by Liaoning University in China have developed recyclable, aqueous-based batteries that won’t succumb to combustion or explosion. This study holds the potential to help researchers develop safer and more efficient water-based energy storage technologies for a cleaner future.
While lithium-ion batteries have proven reliable, they pose safety risks due to the organic electrolytes responsible for creating the electrical charge, which can lead to them catching fire or exploding, limiting their development for large-scale usage. To solve this problem, the researchers used water for driving the electric current between the battery’s terminals, nearly eliminating the chance for a safety hazard.
“Addressing end-of-life disposal challenges that consumers, industry and governments globally face with current energy storage technology, our batteries can be safely disassembled, and the materials can be reused or recycled,” said Dr. Tianyi Ma, who is a team member and a professor in the STEM | School of Science at RMIT University. “We use materials such as magnesium and zinc that are abundant in nature, inexpensive and less toxic than alternatives used in other kinds of batteries, which helps to lower manufacturing costs and reduces risks to human health and the environment.”
According to a study published in the BMJ, a person’s chance of surviving cardiac arrest while receiving cardiopulmonary resuscitation (CPR) in a hospital is 22%, but that declines rapidly after only one minute to less than 1% after 39 minutes. The likelihood of leaving with no major brain damage is similar, declining from 15% after one minute of CPR to less than 1% after 32 minutes without a heartbeat.
Only around 25% of patients survive to hospital discharge after being admitted to the emergency department for cardiac arrest. This common catastrophic medical emergency with a high mortality rate is an important public health issue, affecting around 300,000 adults every year in America alone. Unfortunately, studies have shown that long resuscitation times are linked to lower odds of survival, but there are no specific recommendations on when to stop resuscitation.
This study was designed to measure the effects of CPR duration, using the largest cardiac dataset in the world, utilizing data from 348,996 adults with an average age of 67 years old who experienced an in-hospital cardiac arrest. CPR was defined as the interval between the start of compression and the first return of spontaneous circulation (ROSC) or the termination of resuscitation. The main measures of interest were survival to discharge and favorable function at discharge, defined as a brain performance score of 1 representing good cerebral performance, and 2 representing moderate cerebral disability on a 5-point scale.
Contrary to popular perception, traumatic brain injury (TBI) is not the reserve of car accidents and punishing contact sports; it’s surprisingly common. Up to 50 million new cases of traumatic brain injury are registered each year worldwide. Notably, 80% of TBI occurs in low-to middle-income countries, and it is also the leading cause of death and disability in young adults. Overall, the global economic burden of TBI is estimated at 400 billion USD.
Minimising the devastating effects of TBI doesn’t rely solely on reducing the risk of an injury; it’s also essential to improve treatment after one has happened. For that, physiological real-time monitoring of vital signals is critical. One inventor has made it his mission to create devices that can do this accurately, easily, anywhere, and what’s more, they are also non-invasive.
Professor Arminas Ragauskas is a founder and director of the Health Telematics Science Institute at Kaunas University of Technology in Lithuania, which develops innovative industrial and physiological measurement and process monitoring technologies. He is particularly known for his work on non-invasive intracranial pressure and cerebral blood flow autoregulation measurement devices. He was also the national coordinator of the CENTER-TBI project, funded by the European Commission and the EU industry, with a budget of 40 million EUR, and focused European efforts to advance the care of patients with traumatic brain injury.
Year 2020 face_with_colon_three
There’s a new disease-detecting technology in the lab of Sanjiv “Sam” Gambhir, MD PhD, and its No. 1 source of data is number one. And number two.
It’s a smart toilet. But not the kind that lifts its own lid in preparation for use; this toilet is fitted with technology that can detect a range of disease markers in stool and urine, including those of some cancers, such as colorectal or urologic cancers. The device could be particularly appealing to individuals who are genetically predisposed to certain conditions, such as irritable bowel syndrome, prostate cancer or kidney failure, and want to keep on top of their health.
“Our concept dates back well over 15 years,” said Gambhir, professor and chair of radiology. “When I’d bring it up, people would sort of laugh because it seemed like an interesting idea, but also a bit odd.” With a pilot study of 21 participants now completed, Gambhir and his team have made their vision of a precision health-focused smart toilet a reality.
W/ Andrej Karpathy
The Tokenizer is a necessary and pervasive component of Large Language Models (LLMs), where it translates between strings and tokens (text chunks). Tokenizers are a completely separate stage of the LLM pipeline: they have their own training sets, training algorithms (Byte Pair Encoding), and after training implement two fundamental functions: encode() from strings to tokens, and decode() back from tokens to strings. In this lecture we build from scratch the Tokenizer used in the GPT series from OpenAI. In the process, we will see that a lot of weird behaviors and problems of LLMs actually trace back to tokenization. We’ll go through a number of these issues, discuss why tokenization is at fault, and why someone out there ideally finds a way to delete this stage entirely.
Chapters:
00:00:00 intro: Tokenization, GPT-2 paper, tokenization-related issues.
00:05:50 tokenization by example in a Web UI (tiktokenizer)
00:14:56 strings in Python, Unicode code points.
00:18:15 Unicode byte encodings, ASCII, UTF-8, UTF-16, UTF-32
00:22:47 daydreaming: deleting tokenization.
00:23:50 Byte Pair Encoding (BPE) algorithm walkthrough.
00:27:02 starting the implementation.
00:28:35 counting consecutive pairs, finding most common pair.
00:30:36 merging the most common pair.
00:34:58 training the tokenizer: adding the while loop, compression ratio.
00:39:20 tokenizer/LLM diagram: it is a completely separate stage.
00:42:47 decoding tokens to strings.
00:48:21 encoding strings to tokens.
00:57:36 regex patterns to force splits across categories.
01:11:38 tiktoken library intro, differences between GPT-2/GPT-4 regex.
01:14:59 GPT-2 encoder.py released by OpenAI walkthrough.
01:18:26 special tokens, tiktoken handling of, GPT-2/GPT-4 differences.
01:25:28 minbpe exercise time! write your own GPT-4 tokenizer.
01:28:42 sentencepiece library intro, used to train Llama 2 vocabulary.
01:43:27 how to set vocabulary set? revisiting gpt.py transformer.
01:48:11 training new tokens, example of prompt compression.
01:49:58 multimodal [image, video, audio] tokenization with vector quantization.
01:51:41 revisiting and explaining the quirks of LLM tokenization.
02:10:20 final recommendations.
02:12:50??? smile
Exercises:
- Advised flow: reference this document and try to implement the steps before I give away the partial solutions in the video. The full solutions if you’re getting stuck are in the minbpe code https://github.com/karpathy/minbpe/bl…
Links:
Gene therapy company inks strategic partnership with animal health firm to develop ‘one-dose’ injection treatment for canine osteoarthritis.
A UC Davis Health study found that a single dose of Bacillus Calmette-Guérin (BCG), the vaccine for tuberculosis (TB), reduced liver tumor burden and extended the survival of mice with liver cancer. The study, published in Advanced Science, is the first to show the promising effects of the vaccine in treating liver cancer.
Hepatocellular carcinoma (HCC) is the most common type of liver cancer. It is also the third leading cause of cancer-related deaths worldwide. Current therapies include surgery, radiotherapy, chemotherapy, immunotherapy and liver transplant. Yet, the therapy outcomes for liver cancer remain bleak.
BCG, the century-old TB vaccine, is derived from the live bacteria Mycobacterium bovis. It is considered safe and widely used around the world.
“South San Diego County is in a total state of emergency related to transboundary pollution, and this is a public health ticking time bomb,” Imperial Beach Mayor Paloma Aguirre told ABC News. “We are living in conditions that nobody in this great nation should be living in.”
The Tijuana River – which has been classified as an impaired water body, according to the U.S. Clean Water Act — flows north for 120 miles from Mexico to California before reaching the Pacific Ocean on the U.S. side of the border in the Imperial Beach, San Ysidro and Coronado coastal areas.
Over the last five years, 100 billion gallons of untreated sewage, industrial waste and urban runoff have been dumped into the Tijuana River, according to the International Boundary and Water Commission.
Although people of European descent account for less than one-quarter of the world’s population, their DNA disproportionately drives genetics research. Between 2005 and 2018, the majority of genome-wide association studies were conducted with data from people living in just three countries — the United Kingdom, the United States, and Iceland.
“The paradox of precision medicine is that you have to have a ton of different kinds of people to figure out one person really well,” said Josh Denny, CEO of the All of Us research program. “There’s still so much we don’t understand about the human genome, especially about rare variation. Huge projects like ours are really helping to accelerate that understanding.”
All of Us has recruited more than 750,000 volunteers to provide survey responses about their health, medical records, and if they’re willing, biological samples for molecular and genetic testing. Genetic data from some participants have been available for researchers since 2020, but the new release this week includes the whole genome sequences of nearly 250,000 participants — half of whom are of non-European ancestry.