In December, last year, when OpenAIs ChatGPT was struggling to find real use cases, Google decided to explore the use of large language models (LLMs) for healthcare, resulting in the creation of Med-PaLM an open-sourced large language model designed for medical purposes.

Since then, the team has released scaled-up versions of healthcare LLMs, including Med-PaLM-2 and Med-PaLM-M, both of which have had a direct impact on human lives. Currently, Med-PaLM-2 is also undergoing testing at renowned healthcare institutions such as the Mayo Clinic. One of the prominent contributors to these projects is Vivek Natarajan, an AI researcher at Google Health.

Currently, based in the San Francisco Bay Area, the Tamilian with deep Bengali roots, began his journey as an engineering intern at Qualcomm, progressing to a role with Meta AI, and ultimately finding a fulfilling place at Google Health.

However, there is a story behind why he chose to transition into the field of medical AI.

It is 2023, and Indias healthcare system still faces significant hurdles with insufficient medical infrastructure and a severe shortage of medical professionals, especially in rural regions. The ratio of doctors to patients falls well below global standards, with a mere 0.7 doctors per 1,000 people. Adding to that, we have only 0.9 beds per 1,000 population, and out of those, only 30% are in rural areas.

Most had to walk tens of kilometres, often in extreme conditions, leading to delayed diagnoses, poorly managed chronic conditions, and even untimely deaths. This healthcare disparity affected both the underprivileged and affluent individuals, underscoring the stark healthcare inequalities in these areas.

Having grown up in different parts of India, Natarajan witnessed these immense challenges faced by people in small towns and villages when it came to accessing medical care. It always bothered me that people should not have to suffer so much to receive basic healthcare, and I always wanted to do something about it, Natarajan told AIM in an exclusive interaction.

From starting out by building Ask the Doctor, Anytime Anywhere, an app aimed at democratizing healthcare access in 2013 to being the research lead behind Googles state-of-the-art LLM for medicine, Med-PaLM 2, Natarajan has come a long way. I guess the name gives away what we were trying to do. Ask the Doctor was bootstrapped using older AI techniques and a lot of rules, and it clearly did not work well, leading to its discontinuation, he said.

The app was made by leveraging pre-deep learning ML techniques a combination of expert systems and rules. However, even back in 2013, he had this intuition that AI would be the most important piece of solving this healthcare problem.

After completing a bachelors degree at NIT Trichy in Electronics Engineering and graduating with a masters degree in Computer Science from UT Austin in 2015, Natarajan joined Meta AI. Despite being in the pre-transformer era, Natarajans time at Meta AI, which was his first job, taught him the potential of deep learning. At Meta, he worked in various areas, from speech recognition to conversational and multimodal AI, and on various business-critical platforms such as Newsfeed and Messenger.

However things took a different turn. Unfortunately, it was during this period that his father began showing signs of an aggressive form of Parkinsons disease, which couldnt have been identified sooner due to the limited care options and resources. That persuaded me to go back to the problem that I always deeply cared about using AI to democratise access to healthcare and put world-class medical expertise in the pocket of billions, said Natarajan.

Coincidentally, this was also the time when researchers from Google Brain and DeepMind (now referred to as Google DeepMind), after some seminal medical AI papers, were coming together to form Google Health AI, aligning with his aim. So when Greg Corrado, co-founder of Google Brain and head of Google Health AI, offered me the chance to join, I took it up without hesitation, he added.

Since then, he has collaborated with esteemed AI researchers like Greg and Dr Alan Karthikesalingam to work toward the vision of making an AI doctor accessible to billions.

If not an AI researcher, Natarajan would have probably been a cricket commenter like Harsha Bhogle. Well, lets take a moment to appreciate that he didnt embark on that career, otherwise, we might have missed out on his stellar work in building Med-PaLM, Med-PaLM 2, Med-PaLM M, and related projects.

The core concept driving the development of Med-PaLM is the utilisation of general-purpose language models like PaLM and GPT-4, which excel in predicting text but lack specialised medical knowledge. However, the challenge lies in transforming these models into medical experts.So, we need to do the same with AI and send them to medical school if we want to use them for medical applications. Make them learn from high-quality medical domain information spanning human biology to practice of medicine as well as from clinical expert demonstrations and feedback similar to residency after medical school, he added.

However, the primary obstacle was the scarcity of large-scale medical datasets due to privacy concerns and healthcare in the global south not being digital. Additionally, theres a pressing concern about bias in LLMs used in healthcare. These cultural, social, racial, and gender biases can result in unequal access to care, misdiagnoses, and treatment disparities. The root of this problem lies in the reliance of healthcare LLMs on extensive datasets that mirror historical healthcare inequities, potentially leading to inaccurate diagnoses and treatment recommendations for marginalised communities.

The Med-PaLM models, derived from the PaLM general-purpose language models, are tailored for medical applications through fine-tuning with high-quality medical datasets and clinical expert demonstrations, covering areas like professional medical exams, PubMed research, and user-generated medical questions. These datasets, including the openly available HealthSearchQA dataset from Google, are instrumental in the development of Med-PaLM and its likes.

In the Med-PaLM paper, researchers introduced an evaluation rubric for assessing LLMs in medical applications, with bias being one of the key dimensions.Additionally, in Med-PaLM 2, we introduced adversarial questions evaluation, specifically targeting sensitive topics like vaccine misinformation, COVID-19, obesity, mental health, and suicide. These topics have a high potential to exacerbate bias and healthcare disparities through the spread of medical misinformation, said Natarajan.

Our approach to mitigating bias involves rigorous evaluation and expert clinician demonstrations to train the model. While its a complex challenge, we are steadily making progress in this area, he added.

Consequently, he added that the fine-tuning approach used depends on the available data. In the case of the first Med-PaLM, prompt tuning was employed, wherein the majority of the LLM parameters remained fixed, and only a small set of additional parameters were learned. However, for subsequent versions such as Med-PaLM 2 and Med-PaLM M, the team had access to more data, enabling them to fine-tune the models end-to-end in order to enhance performance and align them more closely with medical expertise.

As we continue to ride the generative AI wave, Natarajan believes that understanding LLMs is crucial, as they differ from human intelligence and require specialised methods, such as mechanistic interpretability or artificial neuroscience, posing a plethora of new challenges that need to be solved. According to him, there lies immense potential for exciting research beyond large language models. He is particularly excited about LLMs potential in biology and neurology, such as analysing the human genome and decoding brain signals.

Although he has no plans to directly revisit building a similar app like Ask the Doctor, he believes that his work on Med-PaLM and medical AI as a whole at Google will eventually lead to something very similar. While there is still a long way to go, given the incredible progress made in LLMs just last year, it appears that my dream of making an AI doctor accessible to billions is no longer science fiction. Fingers crossed! Natarajan concluded.

Read more: Pushmeet Kohli On Solving Intelligence at DeepMind for Humanity & Science

The rest is here:
Meet the Genius behind Med-PaLM 2 - Analytics India Magazine

Description

In a groundbreaking research endeavor, a team of experts at DeepMind has redefined Reinforcement Learning (RL) by providing a comprehensive, precise mathematical definition of Continual Reinforcement Learning (CRL). The paper titled A Definition of Continual Reinforcement Learning challenges conventional RL approaches and establishes a solid conceptual foundation for agents that perpetually adapt and learn from Read more

In a groundbreaking research endeavor, a team of experts at DeepMind has redefined Reinforcement Learning (RL) by providing a comprehensive, precise mathematical definition of Continual Reinforcement Learning (CRL). The paper titled A Definition of Continual Reinforcement Learning challenges conventional RL approaches and establishes a solid conceptual foundation for agents that perpetually adapt and learn from their experiences. By introducing a pair of operators on agents and formalizing core definitions, the team lays the groundwork for future research in the field of CRL.

The DeepMind research teams paper offers novel insights into the realm of Continual Reinforcement Learning (CRL), reshaping the conventional understanding of RL agents. Rather than merely addressing specific issues, these agents are designed to learn continuously and adapt indefinitely. The key aspect of CRL lies in agents that never stop their implicit search over a set of behaviors. This creates an environment where the best agents perpetually update and refine their behaviors based on experience, thereby pushing the boundaries of AI and reinforcement learning.

The core of the research revolves around the formalization of Continual Reinforcement Learning and the establishment of a clean, general, and precise mathematical foundation. The team begins by defining environments, agents, and their interplay. They view the interface between an agent and their environment as two pairs of countable sets of actions and observations, each represented by a history of action-observation pairs. Both the environment and the agent are formulated as functions that respect this agent-environment interface.

To capture the essence of Continual Reinforcement Learning, the researchers propose a two-fold approach

DeepMinds groundbreaking research not only provides a robust definition of Continual Reinforcement Learning but also offers invaluable guidance on designing principled continual learning agents. The implications of this work extend to the creation of AI agents that adapt, evolve, and optimize their behaviors continually, resulting in agents that do not merely solve problems but consistently improve and refine their decision-making processes based on experience.

The teams efforts open doors to a new perspective in designing AI agents, shifting the focus from creating agents that aim to solve specific problems to developing agents that never stop learning and refining their behaviors. This paradigm shift is expected to drive substantial advancements in the realm of Artificial Intelligence and Reinforcement Learning, paving the way for a new generation of smarter, more adaptive AI agents.

As the field of Continual Reinforcement Learning gains traction, the DeepMind research team acknowledges the need for further exploration. They intend to delve into the connections between the formalism of continual learning and the empirical studies in recent times. By bridging theory and practice, the researchers aspire to unlock new possibilities and refine their understanding of Continual Reinforcement Learning, enriching the AI landscape with agents that can learn, adapt, and make intelligent decisions in an ever-changing world.

DeepMinds pioneering work in establishing a precise mathematical foundation for Continual Reinforcement Learning is a significant breakthrough in the realm of AI and RL. By rethinking RL problems as endless adaptation, they have laid the groundwork for a new generation of AI agents that perpetually update their behaviors based on experience. This opens up exciting avenues for future research, pushing the boundaries of AI and Reinforcement Learning to new heights. As the world embraces the potential of CRL, the future of AI looks brighter and more promising than ever before.

See original here:
Is Reinforcement Learning Set to Transform with DeepMind's ... - Cryptopolitan

Murray Engelkeimer, pictured with his grandson Graham, has been dealing with Parkinsons, which causes a variety of challenging symptoms. Doctors are studying whether deep brain stimulation or DBS can help Parkinsons patients with sleep problems. Photo courtesy of Murray Engelkeimer.

About 20 years ago, Murray Engelkemier made a routine stop to pick up some dry cleaning. The ordinary errand took a turn, however, when he began making out a check for his order.

It was hard to write with my right hand, Engelkemier recalled. He soon discovered that the trembling was the result of Parkinsons disease (PD), which causes degeneration in areas of the brain that control movement. The motor symptoms of PD include the tremor that affected Engelkemier, as well as stiffness, slowed movements, balance problems and more.

Engelkemier, now 65, treated his disease with medications designed to increase the levels of dopamine in his brain. Dopamine is a chemical that plays a key role in nerve signaling that enables movement. His symptoms continued to progress, but the treatments worked well enough that he was able to continue for well over another decade in his role as a project and program manager with U.S. Bank.

I was pretty good until the last three years of my career, Engelkemier said. Over that final period, he said he burned out with monthly work-related travel to Spain, leading him to take disability and retire.

His biggest movement-related problem has been leg stiffness that causes him to shuffle and drag his feet.

My mind is always getting ahead of my body, Engelkemier said. But another problem not related to movement also plagued him: regularly disrupted sleep.

I would wake up at 4 a.m. and could not go back to sleep, Engelkemier said. My mind was racing. The fatigue detracted from my quality of life.

His problem is not unusual, said Dr. Alexander Baumgartner, an assistant professor of Neurology with the University of Colorado School of Medicine who practices with UCHealth.

We think of Parkinsons disease as having a lot of motor symptoms, but there are many other, non-motor symptoms, and sleep is one of the most common, Baumgartner said. He added that people with PD also frequently suffer from mood disorders, cognitive decline and dementia, all of which may be entangled with chronically poor sleep.

Broadly speaking, improving sleep is a way to promote general brain health, and that could decrease the risk of developing dementia in the future, Baumgartner said. However, we dont have great treatments designed for these non-motor symptoms [of PD]. There is an unmet need.

Baumgartner now leads a trial that aims to explore whether deep-brain stimulation (DBS), an established surgical treatment for movement-related problems in people with PD, could also help to improve their sleep and their quality of life.

Baumgartner and his colleagues are recruiting people who are considering DBS, a procedure that involves implanting leads, or electrodes, in the brain and connecting them to another implanted device that generates electrical impulses in particular areas of the brain that control movement. These impulses can help to disrupt and tame the haywire electrical signaling that causes the debilitating movement symptoms of Parkinsons disease.

Baumgartner said there are indications that DBS might also help patients experience more restful nights.

We have seen for many years that people who go through DBS often report that they are sleeping better after the procedure, he said.

The problem is a lack of rigorous research to validate and quantify the sleep-related benefits of DBS, he added. For example, some studies previously conducted after DBS surgery simply asked patients to rate their sleep quality subjectively. A small number of others assigned patients to a one-night stay in a sleep lab, hooked up to a monitor to measure their brain activity. In both cases, researchers were able to glean only limited information.

Baumgartners study, which gathers data from patients before and after their DBS surgery, aims to change that with improved technology. Instead of scheduling time and checking into a sleep lab, recruits complete the trial at home, with a headband device that records their brain activity over a three-night period. The in-home arrangement could improve not only the amount of data collected but also its quality, Baumgartner said.

[Patients] spend the night in their own bed, so they are acting more naturally and following their normal sleep patterns and behaviors, rather than spending the night away from home in an uncomfortable bed in a lab, he said. Were also able to study them over consecutive nights, so were getting more data with people in familiar environments that more closely approximate what their life is truly like.

Recruits also get a Fitbit-like wristband that they wear throughout the day to track their normal activity, another likely indicator of how well or poorly they have slept. More data comes from looking at sleep characteristics when patients are taking their PD medications, and when they are not, Baumgartner added.

The researchers ask the recruits to wear the devices for a series of several nights leading up to their DBS surgery. After the surgery, patients have a period for healing and adjustments of the implanted system to manage their motor symptoms as effectively as possible. After about three months, Baumgartner and his team approach them for their consent to participate in the post-surgical part of the trial. If they agree, they go through the same routine they followed before the DBS procedure.

The headband yields data that help the researchers evaluate what sleep pattern changes, if any, occurred after DBS surgery. The measurements include how much sleep patients got, how long it took them to fall asleep and how often they awoke after falling asleep.

The device also offers clues to the quality of sleep, such as how long patients were in Stage 3, which Baumgartner explained is especially important for restoring and recharging mind and body and maintaining healthy memory function and cognition. Recruits offer their own perspectives through questionnaires that ask about how well they slept, their energy levels throughout the day, their pain levels, how often they had to get up during the night and other ways they evaluate their sleep.

Murray Engelkemier said he was initially resistant to the idea of DBS surgery. I didnt want anybody drilling holes in my head, he said. But as his leg problems, in particular, persisted, his providers at UCHealth encouraged him to speak with other patients who had had the procedure, and he was impressed by the positive results they reported.

He decided to seek an evaluation in the spring of 2022, and neurosurgeon Dr. Steven Ojemann performed DBS procedures on both sides of his brain in November and December of that year. Baumgartner, who is his neurologist, also informed Engelkemier about the sleep study, and he agreed to participate.

Today, Engelkemier said he is still working to improve his shuffling gait and control involuntary muscle movement by fine-tuning his DBS programming. His sleep, however, has improved significantly. He said he now regularly gets six to seven solid hours of sleep and no longer stays awake after a trip to the bathroom. The changes have improved his quality of life, Engelkemier added.

I used to be sure to get a nap every day, he said. Ive very seldom needed one since DBS. He works to stay active by riding his bike, playing golf and walking. He participates weekly in tai chi with other PD patients and wants to try paddle boarding.

Of course, Engelkemiers experience doesnt itself establish that DBS improves sleep or other non-motor symptoms in PD patients. But Baumgartner said he hopes his study can increase understanding about possible links between the two and perhaps open the door to more research.

For example, he said, researchers and clinicians might learn how to directly intervene with targeted DBS that stimulates specific areas of the brain involved in sleep. That could lead to improvements not only in sleep but also mood and cognition, Baumgartner said.

Were learning more and more how critical sleep is for the brain, both in adults with and without Parkinsons disease and other neurological conditions, he said. Thats true for kids and teenagers too.

For more information on the study, contact Lisa Hirt,[emailprotected].

Read the rest here:
Parkinson's disease and sleep problems: Finding out if deep brain ... - UCHealth Today

The website of X.ai is now live with a landing page summing up the mission, vision, and leading team.

It will be the super app, the everything app, old Twitter turned into an AI platform that questions understanding the universe. X is an answer to the OpenAI for-profit Microsoft platform of GPT or the open-source AI platform of Facebook, Llama, which Facebook is opening for developers.

X is the official announcement of an unannounced war between the tech giants: Google, Elon Musk, Facebook, and Microsoft.

X.com now redirects to Twitter.

Lets not forget Tesla is one of the major AI companies in the world which is about to finally tackle the auto-pilot problem. Tesla has 200 engineers on the software side, 100 on the chip design side and 300 on the labeling side of AI development.

Elon Musk is also the only international person and Tesla is the only company in China who is really welcome to work in China. China has the only super app, WeChat, currently available, and Elon Musk praises WeChat often.

Tesla GigaShanghai was Tom Zhu and he is now the president of Tesla GigaTexas thanks to his extraordinary skills in building and managing Tesla China. Zhu is seen by many as the next CEO of Tesla if Elon Musk needs to focus on another even larger company, which can be X.

This means that Chinas AI power wind is also behind Xs development.

We shall see in time if this will be one of the great Xs of Elon such as SpaceX, X.com, and his son X. If it works, it may also be everyones favorite letter.

In engineering, there is a saying to mention a great number of things and that is n number of things. For Elon Musk, x is everything or everything is x.

It may also be worth mentioning that Elon Musk is the biggest brand and power in the tech sphere now. So, hundreds of thousands of skilled people would get in line to work with him to create the vision, and millions of people would sign up for any service he launches almost instantly.

One more thing, Threads is dead already?

View post:
Innovation Party Weekly. Another AI-era week with lots to talk | by ... - Medium

TheArtificial Intelligence In Lawreport is an in-depth examination of the global Artificial Intelligence In Laws general consumption structure, development trends, sales techniques, and top nations sales. The research looks at well-known providers in the global Artificial Intelligence In Law industry, as well as market segmentation, competition, and the macroeconomic climate. A complete Artificial Intelligence In Law analysis takes into account a number of aspects, including a countrys population and business cycles, as well as market-specific microeconomic consequences. The global market research also includes a specific competition landscape section to help you better understand the Artificial Intelligence In Law industry. This information can help stakeholders make educated decisions before investing.

Leading players of Artificial Intelligence In Law including:

AIBrain, Amazon, Anki, CloudMinds, Deepmind, Google, Facebook, IBM, Iris AI, Apple, Microsoft, Intel

Free Sample Report + All Related Graphs & Charts @https://www.mraccuracyreports.com/report-sample/346221

The report is classified into multiple sections which consider the competitive environment, latest market events, technological developments, countries and regional details related to the Artificial Intelligence In Law. The section that details the pandemic impact, the recovery strategies, and the post-pandemic market performance of each actor is also included in the report. The key opportunities that may potentially support the Artificial Intelligence In Law are identified in the report. The report specifically focuses on the near term opportunities and strategies to realize its full potential. The uncertainties that are crucial for the market players to understand are included in the Artificial Intelligence In Law report.

As a result of these issues, the Artificial Intelligence In Law industry has been hampered. Because of the industrys small number of important enterprises, the Artificial Intelligence In Law area is heavily targeted. Customers would benefit from this research since they would be informed about the current Artificial Intelligence In Law scenario. The most recent innovations, product news, product variants, and in-depth updates from industry specialists who have effectively leveraged Artificial Intelligence In Law position are all included in this research study. Many firms would benefit from Artificial Intelligence In Law research study in identifying and expanding their global demand. Micro and macro trends, important developments, and their usage and penetration across a wide variety of end-users are also included in the Artificial Intelligence In Law segment.

The market analysis done with statistical tools also helps to analyze many aspects that include the demand, supply, storage costs, maintenance, profit, sales, and production details of the market. Furthermore, the global Artificial Intelligence In Law research report provides the details about the Artificial Intelligence In Law share, import volume, export volume, and the gross margin of the companies.

Artificial Intelligence In Law Segmentation by Type:

Blockchain, The cloud, Others.

Artificial Intelligence In Law Segmentation by Application:

Speech Recognition Software, Document Capture and Automated Workflows, Redaction and Encryption, Others

Artificial Intelligence In Law report answers some key questions:

Highlights of the Report:

Please click here today to buy full report @https://www.mraccuracyreports.com/checkout/346221

Table of Content:

1 Scope of the Report1.1 Market Introduction1.2 Research Objectives1.3 Years Considered1.4 Market Research Methodology1.5 Economic Indicators1.6 Currency Considered2 Executive Summary3 Global Artificial Intelligence In Law by Players4 Artificial Intelligence In Law by Regions4.1 Artificial Intelligence In Law Size by Regions4.2 Americas Artificial Intelligence In Law Size Growth4.3 APAC Artificial Intelligence In Law Size Growth4.4 Europe Artificial Intelligence In Law Size Growth4.5 Middle East & Africa Artificial Intelligence In Law Size Growth5 Americas6 APAC7 Europe8 Middle East & Africa9 Market Drivers, Challenges and Trends9.1 Market Drivers and Impact9.1.1 Growing Demand from Key Regions9.1.2 Growing Demand from Key Applications and Potential Industries9.2 Market Challenges and Impact9.3 Market Trends10 Global Artificial Intelligence In Law Forecast11 Key Players Analysis12 Research Findings and Conclusion

MR Accuracy Reports is the number one publisher in the world and have published more than 2 million reports across globe. Fortune 500 companies are working with us. Also helping small players to know the market and focusing on consulting.

Follow this link:
Artificial Intelligence In Law Market Is Thriving Worldwide- AIBrain ... - Glasgow West End Today