Surfing the AI wave

During my three-week stay on France's southwest coast, observing surfers and waves, I found ample time to reflect on the analogy between catching the AI wave and surfing the ocean. Just as surfers time and position themselves for waves, companies must recognize the ideal moment to engage and align resources for the AI wave's maximum impact. Just as paddling to catch a wave demands strength, effort, and wave understanding, capturing the AI wave requires companies to nurture capabilities and allocate resources. Like surfers predicting forming waves, organizations must anticipate trends and position themselves for AI integration.

The surge of a surfer's "take off" demands skill and swift execution. Similarly, a company's AI journey begins with a concerted push into data-driven innovation, evolving into a seamless process over time. Additionally, a wave exit strategy is vital: like surfers gracefully leaving waves, businesses must discern when to step back from certain AI endeavors, ensuring integration of the learnings and designed AI products into their operations to maximize the value of their investment.?

As we delve into AI's transformative potential in healthcare and explore below the news from the two last weeks, let's recall the art of riding waves — combining timing, strength, agility, balance, and strategic decisions to propel us to new healthcare innovation horizons

AI in Healthcare

The digital transformation of the Pharmaceutical industry

The pharmaceutical industry is undergoing a digital transformation, leveraging technologies like artificial intelligence (AI) and digital tools to enhance research and development (R&D), streamline processes, and improve stakeholder engagement. The industry's investment in digital technologies is expected to reach billions of dollars, with AI spending alone projected to exceed $3 billion by 2025." Several pharmaceutical companies have directed significant investments into their venture capital arms to tap into innovative digital health solutions. As an example, Sanofi increased its capital investment in Sanofi Ventures to over $750 million to support transformative science and digital innovation. As AI is increasingly being utilized for drug discovery and development. Biotech and pharma companies are collaborating with AI specialists to expedite the identification of new drug targets and enhance personalized treatment. Moderna, Eli Lilly, Novo Nordisk, and others are partnering with AI companies to harness data analytics and AI models for innovative drug development. Sanofi, AstraZeneca, Pfizer, and Novo Nordisk have formed collaborations with AI-focused companies to accelerate drug target discovery. Biotech and pharma companies are also investing in AI-based platforms to optimize internal operations, streamline clinical trials, and enhance patient engagement.

Ai in the hospital

As few ask if patients have the option to opt out of having AI used to facilitate their care, it becomes clear to everyone (hopefully) that Artificial Intelligence (AI) in healthcare is not meant to replace healthcare professionals but to augment their abilities and improve patient care.We have examples of AI being used to fight doctor burnout and AI technologies, such as intelligent triage systems, clinical documentation automation, and predictive analytics platforms, can streamline workflows, save time, and make informed decisions, enhancing patient outcomes.

Introducing AI into clinical settings has to be a gradual process. An illustration of this is seen in the Digital Pathology's AI Market, which encounters obstacles. Despite the projected rapid growth of this sector, challenges like the gradual uptake of digital pathology scanners, varying geographic acceptance, and unsteady venture capital funding pose difficulties for individual AI providers. Moreover, going too fast is also risky, as Google is under pressure from a US lawmaker to explain how it trains and deploys its medical chatbot Med-PaLM 2 in hospitals.

AI transforming Healthcare

In the last two weeks, we got additional evidences of practical AI solutions transforming healthcare:

1. Insilico Medicine has developed a cutting-edge AI tool called inClinico that accurately predicts the success of Phase II to Phase III clinical trials with 79% accuracy. This tool leverages generative AI and multimodal data, providing critical insights to biotech and pharma companies, helping them prioritize drug development programs and giving investors valuable information on which drugs are likely to succeed. The research, published in Clinical Pharmacology and Therapeutics, demonstrates the potential of AI in revolutionizing the drug discovery process and saving millions of dollars and decades of effort.
2. In a great interview, Carlos Ciller CEO of RetinAI Medical, explains how RetinAI is driving the healthcare revolution through their cutting-edge AI-powered platform. By facilitating drug discovery, clinical studies, and patient care with real-time image analysis, RetinAI's platform is redefining healthcare decision-making and advancing precision medicine.
3. As described in the Lancet paper "Mammography Screening with Artificial Intelligence trial (MASAI)," the use of artificial intelligence (AI) for mammography screening is safe and effective. The AI-supported mammography screening protocol showed a similar cancer detection rate compared to standard double reading by radiologists. Additionally, the AI approach significantly reduced the screen-reading workload by 44.3%. These results indicate that incorporating AI in mammography screening can maintain the quality of cancer detection while improving efficiency by reducing radiologist workload. The study suggests that AI-supported mammography screening is a viable and promising approach for improving breast cancer detection and screening processes.
4. Researchers from MIT and Dana-Farber Cancer Institute have developed a machine learning model, called OncoNPC, that can predict the origin of cancerous tumors using the sequence of about 400 genes. The model was able to accurately classify 40% of tumors of unknown origin with high confidence, potentially guiding doctors toward personalized treatments for patients with cancers of unknown primary origin, improving the number of eligible patients for targeted treatments.

Data

Data sharing approaches

This great Nature article delves into the impact and effectiveness of clinical trial data sharing models used by the U.S. National Institutes of Health (NIH), specifically comparing the outcomes associated with the National Heart, Lung, and Blood Institute's (NHLBI) centralized model and the National Cancer Institute's (NCI) decentralized model. The research finds that the centralized data sharing model employed by the National Heart, Lung, and Blood Institute (NHLBI) yielded more trials sharing data and a higher number of shared data publications compared to the decentralized model used by the National Cancer Institute (NCI).

Simultaneously, China, vigorously aiming to assert its leadership in biosciences, is actively pursuing genetic data dominance by conducting a "national genetic survey" aimed at collecting and centralizing genetic information. New regulations, effective from July, have restricted foreign access to Chinese genetic data, a departure from the open data-sharing policies of numerous Western countries.

The imperative to share, analyze, and interpret data on a global scale is also crucial in the fight against cancer." In the realm of oncology, "global cancer registry software" serves as an indispensable tool, facilitating seamless data sharing, collaborative analysis, and knowledge exchange among diverse stakeholders worldwide. This software offers a comprehensive repository of cancer patient information, empowering researchers and clinicians to glean valuable insights into cancer trends, survival rates, and treatment efficacy.

Moreover, the significance of interoperability in healthcare technology is underscored in an article featured in HealthDataManagement. Interoperability not only encompasses data sharing but also the unrestricted flow of innovative ideas and concepts. Challenges arise from the existence of siloed workflows, outdated processes, and suboptimal technology within the industry, hindering comprehensive progress and achieving full-scale interoperability.

EMEA's Clarification on Clinical Data Publication and GCP Practice Update

The European Medicines Agency (EMA) has released an updated Q&A document on the publication of clinical trial data, providing applicants and marketing authorization holders (MAHs) with information to comply with the clinical data publication (CDP) policy. The update clarifies that interim study reports and clinical results from ongoing blinded studies are subject to publication, with the possibility of additional redactions to protect the study's integrity.

The FDA has released draft guidance with updated recommendations for the implementation of the International Council for Harmonisation's (ICH's) guidelines on good clinical practice (GCP). The focus is on modernizing clinical trial design and conduct, incorporating emerging technologies, such as decentralized trials (DCTs) and digital health technologies, while ensuring participant safety and data integrity. Furthermore, the revised GCPs encourage the incorporation of digital health technologies (DHTs), like wearable sensors, to facilitate agile data collection and patient recruitment. The FDA's commitment to modernizing its GCP guidelines is evident in its recent efforts. In addition to these draft recommendations, the agency released further documents that complement this initiative, including draft guidance on implementing decentralized clinical trials (DCTs) and a DHT framework document to guide the utilization of data from digital health technologies in regulatory decision-making. Stakeholders have an opportunity to provide feedback on the draft guidance until September 5, 2023, reflecting the FDA's commitment to an inclusive and collaborative approach toward refining clinical trial standards.

Data breach continues increasing the role of cybersecurity group

The role of Global Hospital Information Systems (HIS) in cybersecurity has never been more critical. These systems, which manage and store patient data, are increasingly targeted by cybercriminals, making the protection of this sensitive information a top priority. Here are few examples of this phenomenon:

• Hackers exploited a vulnerability in the corporate file transfer tool MOVEit Transfer to access the protected health information of 1.7 million Oregon residents. Performance Health Technology (PH Tech), a data management services provider for U.S. healthcare insurers, confirmed the breach, with hackers accessing personal and health information including names, birth dates, Social Security numbers, and sensitive health data.
• A class action lawsuit has been filed against a Seattle-area hospital, accusing it of allowing Facebook's online tracking tools to integrate with its website, resulting in the sharing of personal health data belonging to hundreds of thousands of individuals with Meta and other third parties. The lawsuit claims that code embedded in various web pages on the hospital's website allowed Facebook's parent company, Meta, to capture patient information from their interactions with doctors and online health service requests.

As mentioned in the last edition, the Federal Trade Commission’s (FTC) proposed revisions to its health breach notification rules and as the public comment period concludes, various consumer protection and privacy organizations emphasize the need to update health privacy regulations for the digital age."

Platforms

For those who would like to understand more platform business models, this article provides an in-depth explanation and describes how they work and their significance in today's digital world. It covers the key characteristics of platforms, advantages over traditional business models, potential risks and challenges, and examples of successful platforms across various industries.

Growing Number of Platform Providers Harnessing LMM

• GenHealth AI, a spinoff of 1upHealth, has raised $13 million in early funding for its large medical model (LMM), trained on medical event data rather than text, aiming to offer improved performance with reduced bias and hallucinations. The LMM is expected to automate healthcare decisions based on individual patient histories, targeting use cases in risk adjustment, care management, and clinical trial simulations for payers, providers, and pharmaceutical organizations.
• Clinova has introduced Healthwords, an AI health platform focused on delivering healthcare advice through conversational AI tools. The platform allows users to input questions or symptoms to receive medically verified answers or advice, aiming to transform the self-care industry and reduce the burden on GP appointments.
• Emids Unveils EPulseAI – A Generative AI Platform for the Healthcare Industry Emids, a global leader in healthcare and life sciences digital engineering solutions, has unveiled EPulseAI, Generative AI platform aimed at transforming product development in the healthcare industry. The platform promises to accelerate product engineering, enhance productivity by up to 50%, and revolutionize the way healthcare data is processed for diagnosis, treatment, and prevention, ultimately leading to better patient outcomes and improved customer experiences.

Clinical data management

According to the latest research by InsightAce Analytic, the Clinical Data Management Systems (CDMS) Market is valued at US$ 1.98 Billion in 2021 and is anticipated to Reach USD 7.1 Billion by 2031, with a CAGR of 12.44%. The key drivers of the CDMS market's growth include the increasing need for clinical data management, driven by the growing number of clinical trials and the demand for more accurate and timely data to comply with regulatory requirements. Furthermore, the advancement of technology has led to the development of more sophisticated and user-friendly CDMSs, enhancing their efficiency and usability. Additionally, the industry's growing focus on patient safety is prompting the development of CDMSs that can reduce errors and improve patient care. The market is segmented by delivery mode, including licensed enterprise (on-premise) solutions, cloud-based (SAAS) solutions, and web-hosted (On-demand) solutions. Furthermore, it is categorized based on end-users, such as contract research organizations (CROs), medical device companies, pharma/biotech companies, and others.

One of the key player, eClinical Solutions has been Named a Leader in Everest Group’s Life Sciences Clinical Data and Analytics (D&A) Platforms PEAK Matrix??Assessment 2023. New entrants are constantly trying to penetrate this lucrative market, for example Durham-based health tech startup, Clinetic, has successfully raised a Series A round of financing led by Sopris Capital, a prominent venture capital firm specializing in health tech investments. Clinetic, which originated from Duke University, aims to accelerate its innovative work in clinical trials technology and patient recruitment, offering a data-driven software platform that streamlines manual processes, expediting studies and advancing new therapies for enhanced patient care.

Engineering:

As described in the last newsletter, Amazon has introduced , an API that allows software companies to develop clinical note generation apps utilizing AI. Katie Adams explains explains that rather than competing directly with existing AI vendors, Amazon is collaborating with them to streamline the documentation process, but opinions on its success are divided. (there are still some doubts about whether HealthScribe will be successful in tackling the physician burnout problem.)

From TechCrunch, ChatGPT custom instructions expand: OpenAI this week announced that it’s expanding custom instructions — a way to give users more control over how ChatGPT responds — to all users, including those on the free tier of the service. The feature, which was first unveiled in July as a beta for ChatGPT Plus subscribers, allows users to add various preferences and requirements that they want the AI chatbot to consider when responding.

Considering the rise of AI Large Language Models, in a provocative post, Assaf Elovic explains that open-source models have gained momentum, raising questions about their potential to challenge tech giants like Google, Facebook, and Microsoft. However, despite their progress, several barriers hinder open-source models from outpacing proprietary ones, including limited resources, algorithmic shortcomings, and incumbents' advantages, indicating that while open source holds promise, it faces an uphill battle to disrupt the established AI landscape.

Technical debt: Throughout my career, I've witnessed firsthand the challenges and complexities that arise when software projects accumulate technical debt. Balancing innovation and time-to-market pressures often leads to shortcuts, quick fixes, and code that becomes harder to maintain over time. This "debt" can hinder progress, increase maintenance costs, and even jeopardize the stability of a project. I really enjoyed reading this Medium article which present SOLID principles and five foundational design principles that can help maintaining an acceptable level of technical debt.

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
• AI Frontier?by?Steve Nouri