July 15th release

What is the link with AI? "La Vie" by Marc Chagall is exhibited near my home at the Fondation Maeght and it is one of my favorite paintings. The various elements of the painting are interconnected through lines and colors, creating a harmonious visual composition. In a similar vein, artificial intelligence strives to discover connections and correlations among diverse data sets to extract meaningful insights. Chagall portrays a range of life's elements, including love, family, music, death, and spirituality. Similarly, artificial intelligence is integrated into various aspects of human existence, such as healthcare, finance, art, and more. Just as animals and humans are integrated in Chagall's painting, evoking intense emotions, AI also explores the integration of humans with machines, prompting us to contemplate the meaning of consciousness.

Is there a better fit to introduce a new upcoming conference, "AI for Life" which will occur on December 7th in Geneva?

AI

Global Healthcare AI Market Size to reach 280 billon USD by 2032

The Global Artificial Intelligence in Healthcare Market Size was valued at USD 19.45 Billion in 2022 and the Worldwide Artificial Intelligence in Healthcare Market Size is expected to reach USD 280.77 Billion by 2032, with a compound annual growth rate (CAGR) of 30.6%, according to a research report published by Spherical Insights & Consulting. The market is driven by the increasing adoption of AI in healthcare for diagnostics, treatment, and patient care, with companies like #nvidia , #intel , #ibm , #google , #microsoft and others playing a significant role in the industry. The Asia-Pacific region is expected to experience the highest growth in AI adoption in healthcare, driven by a large and aging population, digital infrastructure, and government support for digital health initiatives.

Generative Ai is still on the hype but LLMs are not quite ready to be deployed to clinics

In the view of CB Insights Anjalika Komatireddy , there are three key areas of the healthcare sector where generative AI is booming the most in terms of both venture capital funding and the development of innovative technology: care delivery and navigation, digital therapeutics and wellness tools, and drug research and development. She discussed these three domains during a recent?webinar.As mentioned by Giles Bruce in Becker's Health IT, Google's generative artificial intelligence platform for healthcare is more than 90 percent accurate in making clinical decisions, on par with human providers, as per a?study?published in?#nature?on July 12th. However, the?large language models?aren't quite ready to be deployed in clinical settings, the study's authors acknowledged. The Journal of the American Medical Association reminds us here that using generative AI may face liability risks since these systems have not undergone FDA review. The accuracy and reliability of AI-generated information are not guaranteed, and physicians may struggle to evaluate the output independently. Liability for patient injuries will depend on the applicable standard of care and how the use of generative AI is perceived within the medical community. If Google’s medical AI chatbot is already being tested in hospitals, still in #nature, Eric Topol, MD and Bertalan Meskó, MD, PhD publish on the imperative for regulatory oversight of large language models (or generative AI) in healthcare. In their conclusion, they take our attention on the need to take a proactive approach to regulation, it is possible to harness the potential of AI-driven technologies like LLMs while minimizing potential harm and preserving the trust of patients and healthcare providers alike. UNC (university of North Carolina) Health chief analytics officer also claims that Responsible AI is required now!

Some concrete examples of AI transforming Healthcare

As explained by Tim Sweezy, a Fully AI-Generated Drug Is Being Tested On Humans For The First Time Ever. This is the result of the Hong Kong-based biotech startup, Insilico Medicine, led by Alex Zhavoronkov , has achieved a significant milestone by advancing its fully AI-developed drug, INS0_055, into Phase II clinical trials for the treatment of idiopathic pulmonary fibrosis. The company expects to have the results of Phase II by next year, demonstrating the potential of AI in drug discovery and development.

Another great example in research is the super speedy AI system for biology research developed by MIT scientists who have developed BioAutoMATED. This automated machine-learning system for biology research can select and build appropriate models for given datasets. The system simplifies data preprocessing and reduces the time required from months to just a few hours, making machine learning more accessible to researchers in the field of biology. BioAutoMATED’s open-source code is publicly available here and allows researchers to run initial experiments to assess if it’s worthwhile to hire a machine-learning expert to build a different model for further experimentation.

A promising example of revolutionizing healthcare is the Full-Body AI Scans which could be the Future of Preventive Medicine. As explained, body scan companies say they can diagnose cancer faster and that AI can make it cheaper. But there's still a gap between that dream and the medical community's reality...

Another promise: In Nature reviews neurology, AI-based tools have the potential to revolutionize the diagnosis and treatment of rare neurological disorders by analyzing large amounts of diverse data, identifying patterns, and assisting clinicians in making faster and more accurate diagnoses. In neurogenetic disorders, some AI tools (such as DeepVariant and Bluebee) target rare-genetic-variant calling and others (such as Congenica and Franklin) support interpretation by prioritizing genetic variants on the basis of their associations with specific phenotypes. Other machine learning algorithms, such as Face2Gene, can analyse facial features to generate a list of potential diagnoses. However, here as well, challenges such as data availability, interoperability, genetic diversity, and ethical considerations need to be addressed for the successful integration of AI in clinical practice.

Data

data democratization and data access

Democratizing data access

There is a growing need for democratizing access to health data for patients as described by HealthData Management in The growing need for patient-centric access and transparency. There is a need for individuals to have easy and secure access to their health information, just like they have access to other personal data in areas such as finance or entertainment. The article discusses the challenges patients face in obtaining their health data, the potential consequences of limited access, and the steps that need to be taken to improve individual access to health information. Earlier this year, the US Department of Health and Human Services approved the applications of an initial set of Qualified Health Information Networks (QHINs) for onboarding under the Trusted Exchange Framework and Common Agreement (TEFCA). The QHINs will help establish a universal floor for interoperability, as part of a nationwide framework for health information exchange. By connecting healthcare organizations via these networks, patients will be able to access their health data more easily and securely from a variety of sources.

Medical Image Data sharing

Gradient Health, a prominent medical AI data-sharing company providing access to over 300M+ diverse and standardized patient studies, has successfully raised $2.75 million. This investment will support the company in its ongoing efforts to build the largest annotated medical imaging library worldwide and create a secure platform. By providing researchers with access to diverse medical data, Gradient Health aims to facilitate advancements in diagnostics and therapies, addressing one of the major obstacles in accelerating AI development—access to large datasets of pathologies.

EHR data has surpassed expectations, but bias needs to be addressed

As described by HealthData Management in Exploring the potential of data convergence in the EHR, Electronic health records (EHRs) have evolved to become integral to the healthcare system, offering comprehensive views of patients' health information and serving as a central hub for vast amounts of data. EHRs have expanded their capabilities to include specialty applications, integrate #genomics data, and incorporate patient-generated health data from wearables and remote monitoring tools. However, challenges remain in striking the right balance of data sources, addressing administrative burdens, ensuring interoperability, and effectively harnessing the available data to provide contextual insights to healthcare providers and users.

If EHRs have surpassed the expectations of even the most visionary leaders, there is a growing need to address Bias in EHR data. As highlighted here, an article published in JAMIA(Journal of American Informatics Association) the NIH Pragmatic Trials Collaboratory’s Demonstration Project teams reflect?on strategies to address biases and lack of generalizability in electronic health record (EHR) data for pragmatic research and emphasizes the importance of mitigating biases and increasing inclusivity to ensure that research results are applicable to diverse and underserved populations, aiming to improve health equity.This work was a collaboration between the?Health Equity Core, the?EHR Core, and the?Patient-Centered Outcomes Core of the NIH Pragmatic Trials Collaboratory.

Secondary use of data: Challenges to leverage Medical real-world data stored in clinical, need to enhance the Ethical data review function and trust between Health Service& for-profit companies

As published in Nature article "What prevents us from reusing medical real-world data (RWD) in research". , the reuse of RWD is hindered by technical and cultural challenges. This analysis identifies the complexities of medical RWD, such as its sensitivity and heterogeneity, as well as legal, ethical, and project planning uncertainties. To enable practical and FAIR (findable, accessible, interoperable, and reusable) research on medical RWD, measures like standardized data curation frameworks, improved data management, simplified project approval processes, enhanced patient engagement, and technical solutions for data integration and anonymization are proposed. Overcoming these challenges would facilitate comprehensive and meaningful medical data science analyses, leading to advancements in healthcare. The importance of ethical review by data access committees is emphasized In another Nature article, particularly in the context of big data and artificial intelligence research. The article points out that many committees currently have limitations in reviewing such research, potentially due to the novelty of the field or a lack of relevant expertise in assessing the collective risks and benefits. Additionally, some committees may exempt certain cases involving de-identified data from review. The article suggests that data access committees should conduct ethical review of medical research databases, but with the condition that they enhance the review function by incorporating professional and lay ethical expertise. Another point is to create trust between public and private sector. This paper discusses the issue of public trust in sharing National Health Service (NHS) data for research and innovation with for-profit companies in the UK. Despite the NHS being a highly trusted institution, there are concerns regarding its data sharing practices, particularly with the private sector.

Data Breach Analysis, health breach notification rile (HBNR) and need for Zero trust architecture

A recent analysis of health data breaches in 2023 shows that hacking incidents, including ransomware attacks, continue to be the main cause of breaches and have affected tens of millions of individuals. The analysis highlights the persistent and extensive impact of hacking on protected health information, emphasizing the need for improved cybersecurity measures in the healthcare sector. Therefore, it does not come as a surprise that 95% of patients expressed concern about the possibility of data breaches affecting their medical records (reference: Health Gorilla’s Patient Privacy Report). For any health breach in US, there are clear health breach notification rules. Who Does It Apply To? The Health Breach Notification Rule (HBNR) issued by the Federal Trade Commission (FTC) applies to vendors of personal health records (PHRs) and other non-HIPAA-covered entities. The rule ensures that these entities are responsible for notifying consumers of a data breach involving protected health information (PHI). The HBNR covers three entity types: vendors of PHRs, PHR-related entities, and third-party service providers for vendors of PHRs or PHR-related entities. Take care as well to the expansion of the definition of healthcare providers: the FTC asserts that developers of health and wellness apps are “health care providers”—akin to HIPAA covered entities—for purposes of the Proposed Rule.The rule requires entities to provide breach notices when there has been an unauthorized acquisition of unsecured PHR identifiable health information. The notifications must be sent to impacted individuals, the FTC (for breaches impacting more than 500 people), and the media (for breaches impacting 500 or more residents of a particular state or US territory). Non-compliance with the rule may result in civil penalties. To avoid data breach, #VentureBeat explains Why healthcare in the cloud must move to zero trust cybersecurity It is indeed it crucial for healthcare providers to adopt zero-trust security measures and focus on identity access management and endpoint security. By implementing a zero-trust architecture, enforcing least privileged access, and prioritizing patient data protection, healthcare providers can strengthen their cybersecurity defenses and safeguard sensitive information.

EU's Data strategy and Europe-US agreement on data sharing

The European Union (EU) has developed a Data Strategy to establish itself as a leader in the data-driven society. The strategy aims to create a single market for data within the EU, promoting free data flow for the benefit of businesses, researchers, and public administrations. The EU's approach to data governance is closely tied to the concept of strategic autonomy and digital sovereignty, focusing on security, economy, and rights and values. See more info on: Geopolitical aspects of the EU’s Data Strategy - Elcano Royal Institute

The EU and U.S. this week agreed a new Data Privacy Framework, which will allow businesses to transfer data from the EU to the U.S. in a secure and compliant way. (See here). The agreement of the new rules will provide some relief to Meta and other U.S. tech giants, which share gargantuan amounts of user data around the world. However, the new agreement already faces the threat of legal challenges from privacy activists, who are unhappy with the level of protection offered to European citizens.

Platform

Medical imaging platform: Flywheel spins up $54m funding

Flywheel, a medtech company, has secured $54 million in a Series D funding round led by Novalis LifeSciences and NVentures. The funding will be used to further develop Flywheel's platform, which applies artificial intelligence to the interpretation of medical imaging data, and expand its presence in the pharmaceutical and public sector healthcare markets, as well as new areas such as providers, payers, system integrators, and software companies. Flywheel's cloud data management platform, Flywheel Enterprise, uses machine learning and AI algorithms to capture and curate medical images, enabling faster analysis and improving outcomes in drug development and healthcare.

Blockchain: Pharmaledger

in 2018, I advocated for "Banding Together for a Healthcare Blockchain Consortium" and we created with Marco Cuomo , Daniel Fritz , Disa L. , Adama Ibrahim, EMBA and few others the Pharmaledger IMI Project. This is great to read that 5 years later, PharmaLedger Reaches 30+ Members Demonstrating the Ecosystem Trend in Digital Health. Indeed, the PharmaLedger Association, which is now a nonprofit organization dedicated to establishing a secure Digital Trust Ecosystem in healthcare, has announced the addition of several new members to its organization. The new members, which include companies, government entities, educational institutions, and industry associations, further solidify the Association's role in driving transformation in the supply chain, clinical trials, and health data domains. The Association emphasizes the importance of collaboration and the need for a shared infrastructure to enhance trust, integrity, and efficiency in healthcare and clinical trials

generative AI: MosaicML acquired by Databricks

Databricks picks up MosaicML, an OpenAI competitor, for $1.3B. Databricks, a tech company specializing in AI, has acquired MosaicML, an open-source startup focused on neural networks, for $1.3 billion. MosaicML's expertise in training large language models and deploying generative AI tools will be integrated into Databricks' Lakehouse Platform, expanding its offerings in the AI space.The key differentiators with Mosaic’s approach are two-fold. First, it’s open source. Second, it’s focused on organizations building their own LLMs based on their own data. The company said that its latest release, MPT-30B, “has showcased how organizations can quickly build and train their own state-of-the-art models using their data in a cost-effective way.”

About BurstIQ, a trusted leader in data-driven healthcare solutions

BurstIQ announces the acquisition of Olive AI‘s business intelligence solution. This strategic move expands BurstIQ’s portfolio of innovative products, reinforcing its commitment to helping healthcare organizations navigate the complexities of data-driven solutions while maintaining strict privacy and compliance standards. The acquired solution, now known as LifeGraph Intelligence, further enhances BurstIQ’s offerings, revolutionizing how organizations see and use their data.

About Bing and ChatGPT

ChatGPT and Bing OpenAI’s ChatGPT app can now search the web — but only via Bing: OpenAI has announced a new feature called Browsing for subscribers to ChatGPT Plus, allowing them to search Bing for answers to questions within the ChatGPT app. This feature expands ChatGPT's knowledge beyond its original training data, particularly for current events and information. However, limiting the search capabilities to Bing has raised concerns as it restricts users to a single search engine and potentially excludes alternative search results. The partnership between OpenAI and Microsoft is evident in this decision, but it may be seen as a user-hostile move, considering Bing's history of potential bias and disinformation.

More Partnership and investments

1. Nvidia's generative AI cloud service BioNeMo: Recursion Pharma has secured a $50 million investment from Nvidia, which will give the chip designer a 4% stake in the tech-bio company. The funding will be used to combine Recursion's extensive biological and chemical dataset with Nvidia's generative AI cloud service BioNeMo, aiming to accelerate drug discovery and development in the biopharma industry.
2. Causaly platform: London-based healthtech company Causaly has raised $60 million in a Series B funding round to expand its AI-powered research platform. The platform, which can analyze 35 million scientific documents and clinical trial databases, provides instant answers to specific queries, helping scientists identify, create, and test new drugs more efficiently. The funding will be used to expand Causaly's customer base, establish a presence in the US, and further develop the platform for use in various areas of life sciences.
3. Vivpro's biointelligence software: The Critical Path Institute (C-Path) and Vivpro have announced a Memorandum of Understanding (MOU) to collaborate on advancing drug development. The partnership will leverage Vivpro's biointelligence software platform and C-Path's expertise in data management, biomarkers, and regulatory science to revolutionize regulatory strategy and intelligence. The collaboration aims to monitor the impact of C-Path's solutions on drug development and explore funding opportunities for further collaboration, while also promoting scientific activities and knowledge sharing between the two organizations.
4. Macro Trials has secured $6 million in seed funding for product development and expansion of their Precision Research Clinical Platform aimed at transforming clinical trials. The company addresses the challenges faced in the current clinical trial landscape, such as high failure rates, high costs, under-representation of diverse populations, and limited post-market success. Through their distributed "hub-and-spoke" model, Macro Trials conducts trials in various research settings, including in-person and decentralized approaches.

Engineering

Generative AI

1. OpenAI makes GPT-4 generally available. GPT-4 can now accept image and “Millions of developers have requested access to the GPT-4 API since March, and the range of innovative products leveraging GPT-4 is growing every day,” OpenAI wrote in a blog post. “We envision a future where chat-based models can support any use case. ”
2. As mentioned by Nayur Khan in his June monthly summary, There are over 140 #opensource?large language models now, and increasing by the week
3. As per this medium article, the highly anticipated AI model released by OpenAI, is probably not the groundbreaking breakthrough that was expected. It is revealed that GPT-4 is a combination of eight smaller models rather than a single innovative architecture.

OpenStudybuilder

OpenStudyBuilder Release 0.5 providing Data Exchange Standards for SDTM has been published. OpenStudyBuilder is an open-source project lead by #novonordisk for clinical study specifications. This tool is a new approach for working with studies that once fully implemented will drive end-to-end consistency and more efficient processes - all the way from protocol development and CRF design - to creation of datasets, analysis, reporting, submission to health authorities and public disclosure of study information. See the?GitLab repository

The Future of data engineering and chatting with your database using LangChain

In medium, Barr Moses writes about what's next in data engineering. Barr shares her predictions for the future of data engineering in 2023 and beyond. She predict that data engineering teams will spend more time on FinOps and data cloud cost optimization, data team roles will further specialize, data organizations will transition towards a data mesh while retaining a central platform, machine learning models will have a higher success rate in production, data contracts will move from pioneers to early-stage adopters, data warehouse and data lake use cases will start to blur, and teams will achieve faster resolution for data anomalies. The author emphasizes the need for processes that make the enlarged data universe more organized, reliable, and accessible. This article is a nice follow-up of a previous one Zero-ETL, ChatGPT, And The Future of Data Engineering

The rise of Large Language Models (LLMs) has also brought about a significant shift in technology, empowering developers to create applications that were once beyond imagination. LangChain is an orchestration tool for prompts that leverages the capabilities of LLMs (Large Language Models) to transform the way you communicate with your database. With LangChain, you can easily converse with your database and obtain precise responses in real-time, just as if you were talking to a close friend!

Other Great Newsletter on healthcare and AI

Stay up to date with the latest insights and industry trends by subscribing to these informative newsletters from my esteemed colleagues and professionals in the field.

• Healthcare in the Metaverse by Philippe GERWILL
• DigitalhealthReadingList by Wolfgang Schleifer
• Future Tech Trends by Bernard Marr
• All about Blockchain Adoption by Kamlesh Nagware