This article has been written by Peter Moore, Regional Managing Director, Asia Pacific and Japan Public Sector, Amazon Web Services
2020 was a year like no other. By March, COVID-19 had spread around the world impacting families, businesses, and communities. And, one year later, many are still fighting the spread of the virus, which has since introduced several new variants that are threatening our communities. The speed at which the virus spread left diagnostics for the disease lagging and the healthcare community looking for new ways to use technology to help.
As countries grappled with the challenge of scaling COVID-19 testing, we launched the Amazon Web Services (AWS) Diagnostic Development Initiative to help organizations around the world apply the power of the cloud to accelerate diagnostics research and development. Through this initiative, AWS committed $20 million in computing credits and customized expertise from the AWS Professional Services team to support customers using AWS to drive diagnostic innovations.
In the first phase of the initiative, AWS has helped 87 organizations in 17 countries ranging from startups, nonprofits, research institutions, and businesses. We have awarded $8 million supporting a range of diagnostic projects including molecular tests for antibodies, antigens, and nucleic acids; diagnostic imaging; wearables; and data analytics tools that use artificial intelligence and machine learning to detect the virus.
As we launch the next phase, we are excited to broaden the AWS Diagnostic Development Initiative’s scope and distribute the remaining $12 million this year. From 12 April 2021, we are expanding the scope to three new areas: 1) early disease detection to identify outbreaks at the individual and at the community level, 2) prognosis to better understand disease trajectory, and 3) public health genomics to bolster viral genome sequencing worldwide. While AWS will prioritize COVID-19 projects, we will also evaluate projects focused on other infectious diseases. We will accept applications through the end of the year, with priority consideration given to applications received before July 31. Interested organizations can apply here.
AWS has seen transformative innovations in how startups and organizations diagnose disease over the past year, from machine learning-powered X-ray imagery analysis to new developments in rapid, high quality, and direct-to-consumer tests. These changes will continue to evolve and improve a country’s ability to respond to future outbreaks.
Speeding customer innovation
The AWS Diagnostic Development Initiative has accelerated projects that are changing what is possible with medical diagnostics and having an immediate impact on COVID-19 detection. These projects are not only enabling the medical community to rapidly respond to COVID-19, but also have implications for many other infectious diseases. Here are some examples of the projects funded by the AWS Diagnostic Development Initiative:
Medo uses Artificial Intelligence (AI) to help healthcare professionals quickly identify severe COVID-19 cases
Medo is an AI healthcare startup founded in Singapore and headquartered in Edmonton, Canada, that leveraged support from the AWS Diagnostic Development Initiative to expedite the development of its Medo Lung solution. Medo Lung would allow a quick ultrasound scan of the lung to be reviewed by an AI algorithm to detect whether a patient has normal lung function or is suffering from complications such as interstitial pneumonitis, which leads to many of the severe complications associated with COVID-19. These scans could be performed in thousands of patients together with diagnostic swabs in COVID-19 screening clinics, to assist with rapid, accurate patient triage by helping identify those who urgently need to go to hospital. This would also eliminate the need for patients, particularly the elderly, to leave their homes and visit a doctor unless absolutely necessary, preserving hospital resources, and avoiding potential exposure in the process.
By leveraging AWS and ultrasound, Medo has helped radiologists and clinicians to perform and facilitate the diagnosis of several other medical conditions, in addition to the lung. As a result, caregivers are able to more quickly and accurately diagnose common and critical conditions, and therefore are able to understand the right course of treatment for patients quicker.
Medo currently pilots its Medo Lung solution in Canada. In addition to COVID-19, Medo Lung will also scan for diseases like pneumonia, pleural effusions, pneumothorax, and pulmonary edema.
“COVID-19 really crystallized our vision of democratizing medical imaging and bringing our solution to as many people as we can by highlighting the need for AI-driven diagnostics that can be administered by any caregiver at the point of care where it’s needed the most, keeping patients comfortable and safe. With the help of advanced AWS AI and ML services like Amazon SageMaker and Amazon Textract, we were able to develop Medo Lung that will make a big difference for patients in both the short term and long after the pandemic subsides.” — David Quail, Co-founder, Medo
Artificial intelligence and machine learning are helping us to democratize ultrasound self-diagnosis across the planet with the help of AWS. The cloud has allowed us to greatly improve patient outcomes, including in remote areas that would not otherwise have the infrastructure needed to perform or manage such patients. Scanning an organ takes just a few seconds and our algorithm provides reliable results in less than a minute, making access to medical imaging and early reliable diagnosis a reality for all.” — Dr Jeevesh Kapur, Radiologist and Co-Founder, Medo
Oncophenomics detects mutated COVID-19 virus quickly through next-generation sequencing
Oncophenomics is a healthtech startup based in Hyderabad, India that develops diagnostics solutions for cancer and infectious diseases. The company is studying the genetic epidemiology of the COVID-19 virus in India and is addressing the need for rapid, accurate, and affordable testing and applying it at scale so that the country can implement a test and track approach to identify infected individuals and impose quarantine measures on those infected with the more dangerous variants of the COVID-19 virus.
Not all diagnostic tests in the market are able to detect these new variants of concern. Confronted with the lack of variant-specific diagnostic tests, Oncophenomics has developed a two-step comprehensive saliva-based COVID-19 diagnostics solution that is able to test patients for COVID-19 on-site in minutes (self-administered test; no need for swabs, or requiring the support of a phlebotomist or lab technician) and the positive samples are shipped to a centralized laboratory to accurately diagnose COVID-19 virus variants using third-generation real-time long-read sequencing (Oxford Nanopore Technologies) approach to sequence the complete viral RNA within hours with the same saliva sample.
The sequencing data is being uploaded and analyzed on AWS Cloud using Oncophenomics’ proprietary bioinformatics pipelines and allows the startup to compare every patient sample against a global repository of more than one million COVID-19 virus genomes via GISAID (Global Initiative on Sharing Avian Influenza Data – an initiative that promotes the rapid sharing of data from all influenza viruses and the virus causing COVID-19) within minutes and generate a variant report quickly with clinical correlation.
Governments, public health initiatives, laboratory and healthcare professionals can use this platform to effectively identify the individuals infected with COVID-19 virus variants and plan appropriate interventions to prevent the further spread of the virus in the community. Such rapid interventions are needed to combat the second wave of COVID-19 cases rising in India and other countries.
Oncophenomics’s saliva-based COVID-19 diagnostics solution will undergo ICMR (Indian Council of Medical Research – the apex body in India for the formulation, coordination and promotion of biomedical research, is one of the oldest and largest medical research bodies in the world) performance evaluation and clinical validation by June 2021 for CDSCO (Central Drugs Standard Control Organization – India’s national regulatory body for pharmaceuticals and medical devices) regulatory approval and will be launched in India first, followed by Singapore and other APAC markets.
“AWS makes it easy for us to work on the cloud. Our goal is to address the issue of COVID-19 test performance in light of new variants affecting the conventional RT PCR (Real-Time Polymerase Chain Reaction) tests – its accuracy, specificity, and sensitivity. We have devoted all our resources to developing a robust end-to-end COVID-19 diagnostics solution within a short time frame and at a critical time where the world is facing an imminent threat of a new wave of COVID-19 infections. Some of the data analysis is computationally intensive and can only be run on the cloud.
Without the publicly available datasets (GISAID) and AWS’s scalable cloud computing capabilities, it would have been impossible to develop this COVID-19 variants testing solution. Currently, our saliva-based testing solution, the part-1 point of care test is targeting India specific mutations (while the part-2 nanopore sequencing can discover all mutations), but we hope to customize it for any region across the world to help combat the new COVID-19 wave.” — Dr Shibichakravarthy Kannan, MBBS, PhD, Founder & CEO, Oncophenomics Inc.
Stanford University School of Medicine develops smartwatch-based “alarm system” for diagnostics
Researchers at the Stanford University School of Medicine’s Healthcare Innovation Lab have developed a smartwatch app designed to correctly flag signs of fighting a potential COVID-19 infection. The app is powered by an algorithm that detects changes in an individual’s resting heart rate and step count. Early results are promising, and a pilot trial successfully alerted newly infected individuals as early as 10 days before they became aware of any symptoms.
The app has entered the next phase of study, and the Stanford team is recruiting participants with the goal of reaching 10 million participants to increase its ability to detect signs of COVID-19 in real-time. This smartwatch-based early detection system was built on AWS with the support of the AWS Professional Services team, who has collaborated with the researchers to help the study scale its data processing pipeline.
“We’re hopeful that ongoing screening using wearable devices can provide scalable diagnostics solutions to overcome current testing barriers, and that expanding data access to a broader range of researchers will contribute to new discoveries that improve human health. We look forward to continue pushing the boundaries of what is possible with the cloud.” — Michael Snyder, PhD, professor and chair of genetics for Stanford University’s School of Medicine
We continue to be inspired by the ingenuity of our customers across the world in their use of cloud technology to accelerate diagnostics to help citizens fight against COVID-19 during this trying time, and support governments to prioritize healthcare resources to save more lives. We look forward to supporting broader uses of cloud technologies to enable organizations and communities to identify and respond even faster to future outbreaks.
Rehabilitation services have gained increasing significance, as highlighted by Deputy Prime Minister Heng Swee Keat during RehabWeek 2023. The demand for rehab services is growing worldwide due to an ageing population and a rising incidence of chronic diseases. To meet this demand and improve outcomes, the field of rehabilitation is embracing innovation, particularly through advancements in technology, robotics, and digitalisation.
Rehabilitation plays a crucial role in enabling individuals, regardless of age, to regain independence and participate meaningfully in daily life. With the World Health Organisation estimating that 1 in 3 people globally may benefit from rehab services, the importance of this field cannot be overstated.
Beyond individual well-being, rehabilitation contributes to productive longevity and reduces downstream medical costs when integrated into holistic care plans. Thus, it aligns with the United Nations Sustainable Development Goal of “healthy lives and well-being for all at all ages.”
Deputy Prime Minister Heng shared his personal experience as a stroke survivor, emphasising the pivotal role that therapists and early rehabilitation played in his recovery journey. Early rehab interventions were instrumental in mitigating the debilitating effects of extended bed rest in the ICU. Dedicated therapists, combined with intensive rehab, enabled him to regain full functionality, underscoring the transformative potential of rehabilitation services.
Innovations in rehabilitation leverage broader trends like robotics and digitalisation. These innovations offer precision rehabilitation, tailoring treatment plans to individual needs. They also mitigate manpower constraints by augmenting human efforts with technology.
For instance, robotics-assisted physiotherapy and games-based cognitive exercises are becoming increasingly prevalent. Moreover, virtual rehabilitation has gained prominence during the COVID-19 pandemic, enhancing convenience and empowering patients to take charge of their rehab journeys from home.
Many societies are facing the dual challenge of an ageing population and a declining workforce to provide rehabilitation services. Technology is critical in augmenting these efforts to meet growing demand. Innovations in rehabilitation enhance its effectiveness and accessibility, ensuring that patients follow through with and benefit from rehab programs.
Singapore is at the forefront of innovative rehabilitation practices. Its acute hospitals offer excellent rehab care services and conduct research to improve care. Notably, Tan Tock Seng Hospital is a pioneer in rehabilitation medicine. Changi General Hospital houses the Centre for Healthcare Assistive and Robotics Technology (CHART), facilitating the synergy between clinical needs and technological innovation.
The One-Rehab Framework is a recent innovation in Singapore, ensuring timely access to rehabilitation care. This framework enables seamless care coordination across different settings and care team members through a common IT portal and harmonised clinical outcomes. It streamlines the sharing of relevant patient information and encourages right-siting of care within the community, reducing the burden on acute hospitals.
According to Deputy Prime Minister Heng, RehabWeek serves as a platform for delegates with diverse expertise and a shared commitment to advancing rehabilitation care. It encourages the sharing of best practices and useful technologies to strengthen collective impact, especially when addressing global challenges.
Singapore stands ready to collaborate with international partners, offering its strong ecosystem in research, innovation, and enterprise to advance the field of rehabilitation for the benefit of people worldwide.
He added that rehabilitation is evolving and embracing technological innovations to meet the increasing demand for its services, especially in ageing societies. “Collaboration, innovation, and a focus on the last-mile delivery of care are crucial for ensuring that individuals can live well and maximise their potential through effective rehabilitation,” Deputy Prime Minister Heng said. “Singapore’s commitment to these principles makes it a valuable partner in advancing the frontiers of rehabilitation on a global scale.”
The agricultural sector continues to experience technological advancements. Artificial Intelligence (AI) has become a part of the modern agricultural industry. AI technology is used in various aspects, from production and management to marketing. Agriculture heavily relies on weather, soil, and the environment. Therefore, AI technology related to drones and sensors is essential to support precision agriculture
Drones’ ability to rapidly scan areas with high-quality sensors is beneficial in various applications, including crop mapping, soil analysis, environmental surveys, livestock monitoring, and infrastructure surveillance.
In light of this, the Food Crops Research Centre (PRTP) of the Agriculture and Food Research Organisation (ORPP) under the National Research and Innovation Agency (BRIN) held an occasion regarding AI technology in the development of drones and sensors and its applications in agriculture.
Puji Lestari, the Head of ORPP BRIN, expressed that this occasion would benefit BRIN and other stakeholders. She emphasised that combining drone and sensor technology would create innovative solutions to address food availability challenges.
Furthermore, Puji also highlighted that precision agriculture is closely tied to the availability of tools. Implementing AI in rapid data analysis as a basis for decision-making, ranging from planting and feeding to irrigation and harvesting, is expected to benefit farmers.
The AI-based capabilities, including high-quality sensors and scanning, enable rapid work and real-time data processing, plant identification, and decision-making to support productivity targets. Therefore, the Food Crops Research Centre should provide more opportunities to utilise AI-based technology that supports increased crop productivity,” he emphasised.
At the same time, the Head of PRTP BRIN, Yudhistira Nugraha, also acknowledged that technological advancements have become inevitable. Through the science community, AI researchers are expected to actively contribute to utilising AI technology, turning it into a valuable science that can be applied to agricultural development in Indonesia.
“We can gain many benefits using AI technology for monitoring agricultural land, including fertiliser usage, fertility identification, plant growth, and with the help of AI technology, farmers can make decisions and take actions that can be applied in the farming system to increase productivity,” he explained.
Tri Surya Harapan, Research Manager at a company that provides sales of drones and surveillance services for agriculture, the environment, defence, forestry, and marine purposes, explained about multispectral cameras that provide information on plant health and management.
“AI is widely known for replicating human intelligence and can be simulated using computer systems. Automation sensors embedded in drones, such as camera sensors, LIDAR sensors, or other advanced sensors, provide valuable information as decision-makers in the field without direct human intervention,” he said.
“The use of AI with drone and sensor technology requires relatively high service costs, so in its implementation, collaboration with stakeholders on a large scale is needed,” Tri clarified.
Meanwhile, Senior Researcher at PRTP BRIN, Muhammad Aqil, discussed the Utilisation of Drone Technology in Food Crop Research. This is in line with the direction of the President of Indonesia in the 2021 National IPTEK Coordination Meeting, which emphasises the use of modern technology and contribution to the era of Industry 4.0, including the application of artificial intelligence technology to support all fields/activities, including agriculture.
“We have gone through several stages before reaching Industry 4.0, and now it’s time to use drone technology to monitor the nutrient status of plants, quickly detect pest attacks (OPT – Plant Pest Organisms), check strain contamination, inspect seed production data cells, and determine the harvest time,” said Aqil.
Aqil concluded that the vegetation index-based model developed for the selection of corn genotypes, which are tolerant to both NDVI and NDRE, has proven capable of predicting harvest yields and the best genotype types in corn variety selection in the field.
“By integrating drones and image analysis, it could support research activities, especially in the field,” Aqil added.
Scientists from Washington University in St. Louis have created a sonobiopsy method to diagnose brain disease. The Sonobiopsy method employs ultrasound and microbubbles to momentarily breach the barrier, enabling brain RNA, DNA, and proteins to enter the bloodstream for analysis. While this technique was initially tested on animals, a recent study demonstrates its safety and viability for human use. This innovation may pave the way for non-invasive brain disease and tumour diagnostics.
Eric Leuthardt, MD, co-senior author and co-inventor of the technology, stated that Magnetic Resonance Imaging (MRI) drastically transformed brain disease diagnosis in the 1980s and ’90s, offering structural and functional brain imaging capabilities.
Leuthardt, the Shi Hui Huang Professor of Neurosurgery and a professor of neuroscience at the School of Medicine in biomedical engineering and mechanical engineering at the McKelvey School of Engineering referred to sonobiopsy as the third revolution, emphasising its molecular aspect. This innovative technique allows blood sample collection reflecting gene expression and molecular characteristics at the brain lesion site, essentially performing a brain biopsy without the associated risks of surgery.
Eric Leuthardt and Hong Chen, PhD, Associate Professors of Biomedical Engineering at McKelvey Engineering and Neurosurgery at the School of Medicine, developed the groundbreaking technique, focusing on multidisciplinary research to create engineered solutions for neurological diseases.
The technique employs focused ultrasound to target a brain lesion at a millimetre scale. Subsequently, microbubbles are injected into the bloodstream, travelling to the designated area and bursting, creating minuscule, temporary openings in the blood-brain barrier. These openings naturally close within a few hours, causing no lasting harm. Within this time frame, brain lesion biomolecules can exit the bloodstream, facilitating their collection through a standard blood draw.
Hong Chen, another Senior Co-author and co-inventor of the technology described this innovation as initiating a new field for brain-related conditions. It offers the capability to noninvasively and nondestructively access all brain regions, enabling the retrieval of genetic information about tumours before surgical procedures.
This information aids neurosurgeons in determining the best approach to surgery, helping confirm the nature of suspicious findings on imaging. Furthermore, it paves the way for studying diseases that typically don’t undergo surgical biopsies, including neurodevelopmental, neurodegenerative, and psychiatric disorders.
Initially, the researchers utilised a commercially available ultrasound device combined with an MRI scanner, a setup limited by cost and MRI availability. To streamline the procedure, Hong Chen’s team designed a portable, handheld ultrasound probe that could be attached to a stereotactic pointer commonly used by neurosurgeons for pinpointing brain lesions. This device was seamlessly integrated into the clinical workflow, requiring no additional training for neurosurgeons.
Eric Leuthardt emphasised the user-friendliness of this device, stating that it was efficiently utilised during the study in the operating room but could also be employed in a clinic or at a patient’s bedside in a hospital. He noted that this approach was a significant step toward making advanced diagnostics more accessible, enabling the examination of patients’ brains without needing a high-tech, multimillion-dollar scanner.
In their research, the team conducted sonobiopsies on five individuals with brain tumours using this device. Subsequently, the tumours were removed surgically following the standard care protocol.
The analysis of blood samples collected before and after sonication revealed that the technique increased circulating tumour DNA, ranging from 1.6-fold to 5.6-fold, depending on the specific type of DNA examined.
Circulating tumour DNA holds crucial information about genetic alterations in a patient’s tumour, which guides treatment decisions regarding the tumour’s aggressiveness. Notably, the procedure showed no signs of causing damage to brain tissue, affirming its safety.
The advent of big data has opened up new possibilities for driving sustainable development and informed decision-making. In the context of New Zealand, harnessing the potential of big data presents numerous opportunities to address social, economic, and environmental challenges.
Police agencies in New Zealand are increasingly turning to advanced artificial intelligence (AI) technology to bolster their emergency response and risk assessment capabilities. Recent tragic incidents, such as the shooting of an unarmed constable in West Auckland in 2020, have prompted the development of innovative safety programmes aimed at improving law enforcement effectiveness. One intelligence system has emerged as a central component in this technological transformation.
By collaborating closely with major multinational technology companies specialising in data-driven policing systems, police agencies are harnessing the power of AI to redefine how they assess risks during emergencies. The intelligence system represents a leap forward in enhancing police intelligence systems, enabling law enforcement officers to make more informed decisions swiftly.
One of the critical achievements of the intelligence system is its ability to overcome the limitations of previous intelligence systems. The traditional system struggled to access essential information about criminal organisations, particularly gangs and firearms. This fragmentation hindered the ability of law enforcement to connect the dots and respond effectively to emerging threats swiftly. However, the intelligence system has revolutionised this process by providing instant access to vital connections and associations. This newfound capability significantly enhances police efficiency and decision-making in the digital age.
The intelligence system’s impressive functionality extends beyond mere data access. It leverages advanced AI technologies to deliver more valuable intelligence, particularly concerning firearm-related threats. By integrating data from various sources and employing machine learning algorithms, the intelligence system rapidly analyses and disseminates pertinent information. Front-line officers now can receive real-time updates directly on their smartphones, enabling them to respond effectively to evolving situations.
While the incorporation of advanced AI technology in law enforcement holds promise, it inevitably raises concerns surrounding privacy, transparency, and potential bias. This is not an isolated issue, as similar data-driven policing systems worldwide have grappled with these challenges. To address these concerns effectively, it is essential to conduct comprehensive privacy impact assessments and ensure the utmost transparency in the deployment of such technology.
Furthermore, the emergence of the intelligence system underscores the critical role of collaboration among organisations and the need for strategic partnerships to drive innovation. This initiative exemplifies how technology partnerships can push the boundaries of what’s possible and enhance capabilities beyond individual and organisational limits. In an era marked by rapid technological advancements, collaboration stands as the linchpin of resilience, enabling organisations to collectively address multifaceted challenges and fortify their defences against cyber threats.
The integration of advanced AI technology, exemplified by the intelligence system, into law enforcement operations, has the potential to bring public safety and police effectiveness. However, it simultaneously underscores the paramount importance of ethical considerations, transparency, and the responsible use of such technology to mitigate potential risks and biases.
In the pursuit of a safer and more secure digital future, collaboration remains indispensable, not just for technological advancement but also for achieving the overarching goal of creating a society where innovation thrives and security reigns supreme.
The emerging field of artificial intelligence (AI) has profoundly impacted the healthcare industry. Mahidol University has recognised the development and recently organised an event where experts and academicians gathered to discuss the importance of implementing artificial intelligence in the healthcare sector.
The discussions revolved around how AI technologies, such as machine learning and data analytics, can assist healthcare professionals in making more accurate diagnoses, optimise treatment plans, and personalise patient care. Moreover, AI can streamline administrative tasks, improve resource allocation, and enhance patient engagement.
On this occasion, Dr Somkiat Tangkitvanich, a Member of the Mahidol University Council and President of Thailand Development Research Institute (TDRI), acknowledged augmented reality’s increasing role in making the healthcare industry more efficient.
As AI advances, it transforms various aspects of healthcare, from diagnostics and treatment planning to patient care and administrative tasks. Augmented reality, a technology that overlays digital information onto the physical world, is becoming an invaluable tool in this transformation.
“By leveraging AI, it will not only expedite the diagnosis process but also enhance the precision of diagnoses,” he elaborated.
In healthcare, augmented reality finds applications in medical training, enabling students and professionals to observe intricate anatomical structures and surgical procedures in real-time. This technology enriches the learning process, leading to a deeper understanding of medical concepts.
Additionally, Professor Apichat Asavamongkolkul, Dean of the Faculty of Medicine at Siriraj Hospital, Mahidol University, added that AI has the potential to generate a seamless work-life balance for doctors and healthcare professionals. AI technologies continue to advance and integrate into healthcare, alleviating some of the daily burdens and challenges that healthcare professionals face.
AI can automate routine tasks, such as administrative work and data entry, allowing doctors to focus more on patient care. It can enhance diagnostics, provide doctors with more accurate and efficient diagnosis tools, and reduce uncertainty. AI can also optimise scheduling and resource allocation, giving doctors more predictable schedules and time efficiency.
“AI-powered diagnostic tools can help doctors reach more precise conclusions, potentially improving patient outcomes and reducing the need for extensive and invasive procedures.” Furthermore, AI-driven remote monitoring solutions enable doctors to track patients’ health remotely, reducing the need for frequent in-person visits. AI supports decision-making by providing data-driven insights, streamlining administrative processes like insurance claims and billing, reducing the administrative burden on healthcare providers,” he expressed.
Furthermore, they firmly believe that harnessing the potential of artificial intelligence (AI) in the healthcare sector will ensure its long-term sustainability and effectiveness.
On this occasion, Professor Apichat reminded us that Thailand’s demographic landscape has undergone a significant transformation, characterised by a notable increase in the elderly population, which now accounts for over 17% of the total populace. This demographic shift has officially classified Thailand as an ageing society. Addressing the unique healthcare needs of this growing elderly population presents a considerable challenge and opportunity for leveraging AI technology.
“Wise and responsible use of AI in healthcare involves addressing ethical concerns, data privacy, and ensuring that AI algorithms are transparent, unbiased, and validated for their intended purposes. Moreover, it’s essential to maintain a human-centric approach to healthcare, emphasising the importance of empathy, patient-provider relationships, and holistic care,” he elaborated.
Seeing this, it marks that Mahidol University underscored the commitment of healthcare professionals and academics to harnessing the power of AI to improve healthcare outcomes, which extends far beyond the confines of its campus. It is a declaration that they are not content with merely meeting the current standards of patient care and medical research; they aspire to set new standards and lead the way in shaping the future of healthcare and medical knowledge.
Generative artificial intelligence (AI) is at the forefront of transforming the boundaries of digital reality, promising to take simplicity and turn it into complexity through the creation of patterns in images, sounds, and text. Researchers at the Massachusetts Institute of Technology’s Computer Science and Artificial Intelligence Laboratory (MIT CSAIL) have delved deep into this realm, introducing an innovative AI model that bridges the gap between two unrelated physical principles: diffusion and Poisson Flow. Their work has led to the development of the “Poisson Flow Generative Model ++” (PFGM++), which is poised to redefine digital content creation across various applications.
The PFGM++ model represents a leap in generative AI, offering the capabilities to generate a wide range of content, from images to audio. Its potential applications span from the creation of antibodies and RNA sequences to graph generation. At its core, PFGM++ extends the foundation of the Poisson equation, a concept from physics, to enhance its data exploration and generation capabilities. This breakthrough underscores the power of interdisciplinary collaboration between physicists and computer scientists in advancing the field of AI, as highlighted by Jesse Thaler, a physicist at MIT.
Thaler emphasises the remarkable progress achieved by AI-based generative models in recent years. These models have generated photorealistic images and coherent textual content, challenging the boundaries of artificial intelligence. Notably, some of these powerful generative models draw inspiration from well-established physics concepts such as symmetries and thermodynamics. PFGM++ builds upon a century-old notion from fundamental physics—the existence of extra dimensions in space-time – and transforms it into a versatile tool for crafting synthetic yet authentic datasets. The infusion of ‘physics intelligence’ is revolutionising the landscape of AI.
In the PFGM model, data points take on the role of minuscule electric charges within a multidimensional space, shaping an electric field that extends into an extra dimension, ultimately forming a uniform distribution.
This process is akin to rewinding a video, starting with charges and retracing their path along electric lines to recreate the original data distribution. This process enables the neural model to grasp the electric field concept and generate new data that mirrors the original.
The PFGM++ model takes this concept further by expanding it into a higher-dimensional framework. As these dimensions continue to grow, the model’s behaviour unexpectedly begins to resemble another crucial category of models known as diffusion models. This work aims to strike a balance, as PFGM and diffusion models occupy opposite ends of a spectrum: one is robust yet complex to handle, while the other is simpler but less sturdy. The PFGM++ model introduces a balanced middle ground, combining robustness with user-friendliness, revolutionising image and pattern generation and marking a significant technological advancement.
In addition to its adaptable dimensions, the research team has proposed a novel training approach that enhances the model’s understanding of the electric field, further boosting its efficiency.
To bring this concept further, the research team tackled a pair of differential equations detailing these charges’ motion within the electric field. They evaluated the model’s performance using the widely accepted Frechet Inception Distance (FID) score, which assesses the quality of generated images compared to real ones. PFGM++ excelled in demonstrating enhanced error tolerance and resilience regarding the step size within the differential equations, solidifying its position as a game-changer in the realm of AI-generated content.
In the future, the researchers are committed to refining specific aspects of the model through systematic approaches. They aim to identify the optimal value of D, customised for distinct data sets, architectures, and tasks, by closely analysing the behaviour of neural network estimation errors. Moreover, they plan to leverage PFGM++ in contemporary large-scale endeavours, particularly in text-to-image and text-to-video generation.
MIT’s PFGM++ stands at the forefront of a digital content revolution, bridging the gap between AI and reality. By integrating physics principles and advanced AI techniques, this innovative model promises to reshape the way we create digital content, opening up new horizons for creativity and application across various industries.
The Department of Telecommunications (DOT), in partnership with the National Disaster Management Authority (NDMA), will conduct extensive testing of the Cell Broadcast Alert System. This effort is aimed at strengthening emergency communication during disasters and enhancing safety measures to protect the public.
The Cell Broadcast Alert System is an advanced technology that enables authorities to disseminate vital and time-critical disaster management messages to all mobile devices in specific geographic regions. This includes both residents and visitors, ensuring that crucial emergency information reaches as many individuals as possible promptly.
Government agencies and emergency services employ Cell Broadcasts to inform the public about possible threats and deliver vital updates during critical situations. This technology is commonly used for issuing emergency alerts like severe weather warnings (tsunamis, flash floods, earthquakes), public safety notifications, evacuation instructions, and other critical information.
The Cell Broadcast Alert System will undergo rigorous testing with multiple telecom service providers. These tests will be conducted periodically in various regions across the country to evaluate the emergency alert broadcasting capabilities of different mobile operators and cell broadcast systems for efficiency and effectiveness. As part of this endeavour, tests are being conducted in different states across India, with Punjab being the next state on the testing schedule for 29 September.
In a press release, DOT said that it is responsible for formulating developmental policies to accelerate the growth of the telecommunications sector in India. “Our mission is to ensure access to affordable and effective telecommunications services for all citizens while promoting innovation and safeguarding national security interests.”
The proliferation of digitalisation in both service and manufacturing domains has ushered in a global transformation. In recent years, the demand for digital connectivity has grown, and this vital role was highlighted during the pandemic, when there was a surge in demand across user segments, regardless of their geographical locations.
The Telecom Regulatory Authority of India (TRAI) has been overseeing the quality of telecom services nationwide through comprehensive studies and by issuing directives to stakeholders to improve facilities. Although there have been notable enhancements in the coverage of telecom services outdoors, there are still gaps in meeting the expected quality of service within buildings, whether they are residential or commercial areas.
Ensuring the quality of telecommunication services within buildings is a vital aspect of safeguarding consumer interests. TRAI has already implemented several policy initiatives, including the Recommendation issued on 20 February 2023, regarding the “Rating of Buildings or Areas for Digital Connectivity.” These recommendations establish an introduction for building ratings, aiming to deliver a satisfactory digital connectivity experience to consumers through a collaborative and self-sustainable approach.
To establish a regulatory framework, TRAI has indicated in its observations that it intends to develop the necessary regulations for the Rating of Buildings. It recently issued a consultation paper titled “Regulation on Rating Framework for Digital Connectivity in Buildings or Areas.” It deliberates on the regulatory measures needed to implement a rating framework.
The paper underscores the necessity of a rating system that not only caters to the current consumer expectations but is also adaptable for future expansion and upgrades. It should allow for evolving technologies and shifts in user demands. The paper also explores the benefits of a rating framework for end-users, service providers, and the broader ecosystem.
The consultation paper provides an overview of the ‘Rating Framework for Digital Connectivity’ based on international practices and existing rating frameworks such as GRIHA or Credit Rating in India. The consultation paper along with draft regulations have been uploaded to TRAI’s website, seeking inputs from the stakeholders and telecom consumers. Written comments will be accepted by 10 November and counter-comments by 24 November.