Dilip Prasad

🌍 Heja Sverige! Microsoft invests €3 billion in 🇸🇪 Today, Microsoft revealed a massive €3 billion investment in Sweden over the next two years to advance artificial intelligence. This announcement has been warmly received by the Swedish government, which believes it will help the country "remain at the forefront of global competitiveness," according to Microsoft's CEO. Microsoft plans to invest SEK 33.7 billion (€3 billion) in AI development in Sweden over the next two years—the largest investment the American company has ever made in the Nordic country. The firm aims to train 250,000 people by 2027 to "increase knowledge and skills" in AI, and will also expand the capacities of its three data centers. In Sweden, Microsoft will provide 20,000 advanced graphic processors, crucial for training AI models, and will bolster its data storage centers in Sandvikens kommun, Gävle kommun, and Municipality of Staffanstorp. Prime Minister Ulf Kristersson commented, "AI is a catalyst for many things. It will also help accelerate development in other areas. This significant investment in Sweden could pave the way for further investments." Brad Smith, President of Microsoft, stated, "This investment should enable Sweden to stay at the forefront of global competitiveness." Global Investment Program Recently, Microsoft has announced substantial AI investments in several countries worldwide. For example, the company committed €4 billion in France - my home country - €3.4 billion in Germany, and significant plans in Asia: €1.6 billion in Indonesia, €2.05 billion in Malaysia, and €3 billion in Japan. In Malaysia, this funding will build AI and cloud infrastructure, create an AI center of excellence, and provide AI training for up to 200,000 Malaysians. AI Raises Concerns Among its achievements, Microsoft has developed VASA-1, an AI technology that transforms a facial photo and a voice sample into an ultra-realistic talking video. Although impressive, this technology raises concerns about potential misuse for fraud and misinformation. To address these, Microsoft has established an AI advisory board to guide the technology's development responsibly. Moreover, Microsoft does not plan to release this new tool or provide technical information until they are confident it will be used responsibly and in accordance with regulations. Amazon’s Massive Cloud Investment Microsoft is not alone in investing in AI and cloud technologies. Amazon announced on May 22nd a €15.7 billion investment over ten years to enhance its data centers' capacity in Spain. The company also plans to invest nearly €8 billion in Germany through its Amazon Web Services (AWS) subsidiary. Additionally, Amazon will invest over €1.2 billion in France, creating more than 3,000 jobs. How is your country attracting investments in AI? #AI #foreinginvestments #microsoft #amazon #GAFAM #fantasticfour #magnificientseven #sweden #france

3

What a phenomenal development!!! Swedish Scientists Make World’s First 'Living' Computer Using Human Brain Tissue. While this is very different than the Neural Network based AI-ML models that we have seen in operation, this can certainly be considered one of the next logical steps to improve the Computing Systems and their capacity to scale to unprecedented levels in future. What caught my attention is the comparison of power consumption figures of Human Brain when compared to one of the best computers in the world, such as the Hewlett Packard Enterprise Frontier. If similar figures are being projected for these new type of computers; then we all must take a serious note of it. I am eager to witness the next leap that this technology is promising to us!! https://lnkd.in/dwtT235B

8

👉🏼 Comparative Analysis of Artificial Intelligence Virtual Assistant and Large Language Models in Post-Operative Care 🤓 Sahar Borna 👇🏻 https://lnkd.in/e3BfiPQF 🔍 Focus on data insights: - AIVA demonstrated superior performance with significantly higher accuracy (mean: 0.9) compared to BARD and ChatGPT-4. - Lower knowledge gap observed in AIVA (mean: 0.1) indicating better understanding and response quality. - AIVA's responses received higher Likert scores for appropriateness, showcasing its effectiveness in delivering contextually relevant information. 💡 Main outcomes and implications: - Specialized AI tools like AIVA are more effective in providing precise and relevant information for postoperative care compared to general-purpose LLMs. - ChatGPT-4 shows promise but exhibits varying performance, particularly in verbal interactions. - Emphasizes the need for tailored AI solutions in healthcare to ensure accuracy and clarity for improved patient outcomes. 📚 Field significance: - Customized AI solutions are crucial in addressing specific medical contexts and enhancing patient care. - Further research is necessary to develop AI tools that cater to the unique requirements of healthcare settings. 🗄️: [#ArtificialIntelligence #VirtualAssistant #LargeLanguageModels #PostOperativeCare #Healthcare #DataInsights]

1

1 kommentar

🤖 AI is making strives in the Medical imaging world once again. 👩‍🔬 A study in Denmark has shown that AI could improve cancer detection while reducing false positives. Radiologists used AI to analyse mammograms from two groups of women. The AI identified suspicious areas and radiologists reviewed these images. Cancers were smaller in the group using AI. #ai #medicalimaging #breastcancer

18

https://lnkd.in/drSnd4pw I watched this interview of prof Li Fei-Fei, Sequoia Professor of Computer Science in Stanford, a god-mother in AI with her ImageNet project which tremendously boosted the growth in AI especially in deep learning. She mentioned now the academics world lack the means to compete in the cutting edge of AI due to limited resources. Stanford's NLP (Natural Language Processing) lab only has 64 GPUs, so much dwarfed by the tech giants of the world. #ai #ml #machinelearning #deeplearning #gpu #nlp #genai

16

2 kommentarer

Can A Minute Per Day with Wearable AI Reduce Stress? The first paper from the MOODS study was presented at ACM CHI’24 this week (https://lnkd.in/eEW9aP_A). This 100-day study with 122 participants showed that one minute per day with the MOODS app helped participants self-initiate fourteen types of behavior changes and lower their stress. The high utility and low burden resulted in a 30-day retention rate of 81% and an app approval rating of 85%. It took 17 years and contributions from many people for Wearable AI to reach this milestone. I want to start by first acknowledging the MOODS team. The MOODS software consisted of an app for Fossil smartwatches, apps for Android and iPhones, and cloud software. It was developed by Timothy Hnat, Anandatirtha Nandugudi, and Dr. Nasir Ali. The study was designed and conducted by Sameer N., Mithun Saha, and Shahin Samiei with contributions from Rabin Banjade, Soujanya Chatterjee, Ph.D., and Md Azim Ullah, Ph.D. The data analysis was guided by our esteemed collaborator david almeida. Our team’s journey began in 2007 when our multidisciplinary team was selected by the GEI program to detect stress in the natural environment. Emre Ertin designed the AutoSense wearable and Mustafa al'Absi conducted the lab and field studies that produced the data necessary to develop the first version of our AI algorithm to detect stress (in 2011). Kurt Plarre, Andrew Raij, PhD, Somnath Mitra, Amin Ali, and Syed Monowar Hossain in Memphis led the AI development. Our collaborators included Prof. Tom Kamarck, Prof. Dan Siewiorek, and Prof. Asim Smialegic. By 2016, Karen Hovsepian, Timothy Hnat, Moushumi Sharmin, Mahbubur Rahman, PhD, Rummana Bari, and Hillol Sarker in Memphis developed a robust version of our AI (called cStress) and implemented it on Android smartphones to detect stress in real-time, in collaboration with Bonnie Spring, Prof. Billie Nahun-Shani, and Susan Murphy. By 2019, we developed the next version of our AI that could run directly on commercial smartwatches. MOODS study showed that it is now ready for real-life applications. Back in 2007 (having just completed my Ph.D. involving sensors, but with zero experience in health or wellness), I had naively announced at the first GEI meeting that we would solve stress detection in four years. Some experienced people remarked at the time that it may take 20 years. They were much closer than I was. It took us 17 years. Now that we see evidence of the real-life benefits of this AI, we have embarked on the next phase of our journey, i.e., to bring its benefits to everyone for improving their health and wellness via CuesHub. I want to thank many amazing program officers from NIH and NSF (Will M. Aklin, Ph.D., Richard Conroy, Tiffani Bailey Lash, Ph.D., Wendy Nilsen, Vinay Pai, Misha Pavel, Ed Ramos, William Riley, Sylvia Spengler, and several others), whose encouragement and unwavering support have been instrumental in our journey.

63

16 kommentarer

📢 Exciting Announcement! 📢 I'm thrilled to invite all my esteemed contacts in academia, life sciences, biotech, and medtech in Copenhagen to an engaging workshop I'm leading, hosted by Erudita.tech. This event is designed to empower researchers in digital health, providing an excellent opportunity to learn how to conduct systematic reviews efficiently. Event Details: Title: Workshop with Erudita.tech: Empowering Researchers in Digital Health Date and Time: Thursday, June 13th, 17:00 onwards Location: Nørre Allé 49, 2200 København, Lighthouse KUB Nord (Entrance via Frederik Bajers Plads) Agenda: Introductions by Hosts Overview of Erudita – Learn about its application in digital health research. Workshop Session: Participants will be divided into groups. Each group will create their own systematic review protocol using Erudita. 20-30 minutes for protocol development. Interactive Discussion: Groups present their systematic review protocols. Feedback and discussions (recorded for platform improvement). Important Notes: Participants need to use their Gmail or LinkedIn email addresses to access Erudita. Networking breaks will be incorporated. Please bring your computer for this hands-on session. I hope many of you will join us and leverage this event to start your own systematic reviews. Looking forward to seeing you there! For more information and registration, please visit this link: https://lnkd.in/dm33tjPi #DigitalHealth #SystematicReview #HealthcareInnovation #Research #Biotech #Medtech #CopenhagenEvents #Entrepreneurship #HealthcareProfessionals #AcademicNetworking, #EruditaTech Let's make a significant impact in digital health research together! 🚀

12

Flow Computing, a spinout from Finland’s acclaimed VTT Technical Research Center, says its Parallel Processing Unit (PPU) can enable '100X Improved performance for any CPU architecture.' The startup has just secured €4M in pre-seed funding and its founders and backers are asking the tech world to brace for its self-titled era of "CPU 2.0." #emergingtechnologies #cpu #cpu20 #computing #computerscience #ppu #techstartups #itstartups #earlystagestartups #breakthrough #innovation #parallelprocessing #cpuperformance #futuretech #innovation #technologydevelopment #futurevision #technologydeployment #futureresearch #technologyinnovation #emergingtech #disruptivetechnologies #scienceandtechnology #highspeedcomputing #chips #supercomputers #chiptechnology #dataprocessing #chipmakers #processors #gpu #semiconductors #semiconductorindustry #microelectronics #ai #aitechnology #artificialintelligence #aiapplications #superchargedcpu

14

She says one question drives her work: “Which pillars of gravitational physics are just not true?” Jennifer Chu April 9, 2024 A black hole imprint In 2014, midway through her PhD work, Engelhardt honed one of her questions about quantum gravity and spacetime emergence to a specific problem: how to compute the quantum corrections to the entropy of gravitating systems. “There are surfaces (in spacetime) that are sensitive to gravitational (curving) called extremal surfaces,” Engelhardt explains. “There already was a formula that used such surfaces to compute the entropy of gravitational systems in the absence of quantum effects. But in realistic quantum gravity, there are quantum effects, and I wanted a formula that took that into account.” She and postdoc Aron Wall worked to construct a general equation that would describe how entropy of gravitating regions should be computed when quantum effects are taken into account. The result: quantum extremal surfaces, a quantum generalization of the old classical surfaces. But in 2019, as a postdoc at Princeton, Engelhardt and others realized that this equation might give a very different prediction for what a quantum extremal surface might do, and what the corresponding quantum gravitational entropy would be, in one specific situation: as a black hole evaporates. What’s more, what the equation predicts could be the key to resolving the longstanding black hole information paradox. “This was a very dramatic moment,” she recalls. “Everyone was working around the clock to try to figure this out, not really sleeping at night because we were so excited.” After three sleep-deprived weeks, the physicists were convinced that they had made a dramatic step toward resolving the paradox: As a black hole evaporates and releases radiation in a scrambled form of the information that originally fell into it, a new, completely nonclassical quantum extremal surface emerges, resulting in a gravitational entropy that shrinks as more information radiates away. They reasoned that this surface can serve as an imprint of the radiated information, which could in principle be used to reconstruct the original information, which Stephen Hawking had shown would be lost forever. “At the end of the day, if you really want to resolve the paradox, we have to explain what Hawking’s mistake was,” Engelhardt says. Her hunch is that he was in a way computing a different quantity altogether. She believes Hawking’s work, which raised the paradox to begin with, might have been computing a different type of gravitational entropy, that appears to result in information loss when run forward as a black hole evaporates. However, this other form of gravitational entropy does not correspond to information content, and so its increase would not be paradoxical.

1

A DETAILED ALGORITHM JOURNEY MAP THAT SPANS DEVELOPMENT, INTEGRATION, AND MAINTENANCE PHASES ⬇ This paper discusses the systematic integration of the SepsisWatch AI tool into Duke University's healthcare systems, providing a detailed algorithm journey map that spans development, integration, and maintenance phases. 1️⃣ The study emphasizes the necessity of bridging theoretical AI concepts with practical applications, underscoring the challenges of operationalizing AI in healthcare. 2️⃣ It provides a comprehensive "algorithm journey map," detailing steps from the identification of healthcare problems to the integration and lifecycle management of AI solutions. 3️⃣ The paper highlights key lessons learned in modeling assumptions, stakeholder engagement, and organizational structure during the implementation of SepsisWatch. ✍🏻 William Boag, Alifia Hasan, Jee Young Kim, PhD, Mike Revoir, Marshall Nichols, Will Ratliff, Michael Gao, Shira Zilberstein, Zainab Samad, Zahra Hoodbhoy, Mushyada Ali, Nida Saddaf Khan, Manesh Patel, MD, Suresh Balu, Mark Sendak. The algorithm journey map: a tangible approach to implementing AI solutions in healthcare. npj Digit. Med. 7, 87 (2024). DOI: 10.1038/s41746-024-01061-4. 💡 This paper outlines the nuanced interactions and dependencies within healthcare ecosystems when introducing AI technologies. It serves as a crucial reminder of the importance of detailed planning, comprehensive stakeholder inclusion, and adaptability to feedback throughout the lifecycle of an AI tool. The structured methodology presented can significantly aid healthcare providers and administrators in navigating the often turbulent waters of AI integration. 💥 Join our community of 30.000 curious minds! ✅ Subscribe to our newsletter for your front-row seat to the latest groundbreaking studies. Get started here: https://lnkd.in/eR7qichj.

85

8 kommentarer