Generative AI on the hype

The world of Generative AI is abuzz with possibilities, as organizations strive to unlock its value. If Mahatma Gandhi's famous list of seven social sins, written in 1925, were to be published today, it is most likely that he would include "AI WITHOUT ETHICS" as a significant addition.

AI

Generative AI has emerged as a prominent topic that is capturing the attention of various companies, including hospitals, pharmaceutical firms, and startups. Many of these organizations have already embarked on Proof of Concept (PoC) initiatives or have started to deliver solutions, harnessing the power of ChatGPT. DataScienceCentral helps us understand the diverse use cases that Language Models and Machine Learning (LMMs) can address. In the Adam Roger speculates that ChatGPT might replace your doctor and this is quite scary to read that the chatbot's answers were rated as three to four times more reliable and seven times more empathetic than those from human doctors. While the ? is exploring how generative AI-based enterprise search functionality can help clinicians and other staff gather information from across the health system, they came up with a partnership with Google Cloud focused on generative AI. Tempus also rolled out an AI-enabled clinical assistant that leverages generative AI to give clinicians easier access to patient data and the Hybrid primary care company #CarbonHealth debuted hands-free charting—an AI-enabled notes assistant—in its proprietary electronic health record software?across each of its clinics and providers.

On the Drug Discovery side, MITnews Anne Trafton tells us that Large Language Models offer a way to speed up drug discovery and that by applying a language model to protein-drug interactions, researchers can quickly screen large libraries of potential drug compounds. This new approach "represents a significant breakthrough in drug-target interaction prediction and opens up additional opportunities for future research to further enhance its capabilities,” says Eytan Ruppin .

At HIMSS23 Europe, Harvey Castro, MD, MBA. , a physician and healthcare consultant, reminds that regulation plays a vital role in ensuring the ethical and responsible use of AI in healthcare. The proposed EU AI Act, expected to be adopted in 2023, aims to establish the first legal framework for AI technology in Europe. Just a few days ago, The FDA released two discussion papers to spur the conversation about AI, associated with AI/ML in drug development, addressing the ethical and security considerations as well as the need for explainability.

As explained by Dennis McWilliams , Venture capital firms see plenty of promises and are currently supporting the growth of generative AI, but companies will need to generate revenue, convince hospitals to invest in their algorithms, and need to address racial and ethnic disparities in healthcare by ensuring their data sets are representative of diverse populations. And... if you are more of a podcast person, Becca and Dominic-Madori investigate what it takes to mitigate AI biases in healthcare, featuring Amy Brown from .

Data

To effectively harness the potential of AI, obtaining unbiased and high-quality health data without compromising personally identifiable information remains a significant challenge. Accessing reliable clinical data is crucial for training AI models, and various sources such as patient-shared data, real-world evidence data, and synthetic data hold immense promise in this regard. However, significant efforts are still required to ensure appropriate access and utilization of these diverse data sources.

1. Clinical data: The European Union is currently soliciting public input to define the boundaries of clinical data sharing : This presents an opportunity for you to express your views on achieving a delicate equilibrium between transparency, safeguarding personal data, and protecting commercial interests.
2. Real world data: has?released?a Community Health Data Hub to provide public health stakeholders, community organizations, and residents access to county-level health data, according to reporting from?The Sentinel-Tribune.
3. Generating synthetic data could offer scalability and privacy benefits. However, as highlighted in this David Talby article, it currently faces limitations such as data leakage and challenges in generating diverse patient cohorts, which impede its utilization for training medical machine learning models and conducting analytics.
4. Patient-shared data: Another option would be to build confidence and trust in the population in sharing their health data and to encourage data donation and consent. This raises issues such as privacy and security and has been discussed during a session at HIMSS23 Europe: Promoting patient trust in health data sharing.

In a fragmented health data world, the future lies probably in data sharing and the ONC is Creating New Frameworks, Standards to Improve Data Sharing to improve data sharing and interoperability in healthcare through initiatives such as the Trusted Exchange Framework and Common Agreement (TEFCA) and the United States Core Data for Interoperability (USCDI). These efforts are in line with the 21st Century Cures Act, which aims to promote cross-government interoperability and data exchange.

Lastly this Massive medical breach involved the theft of personal and health information belonging to half a million individuals by hackers serves as a stark reminder that prioritizing health data security and protection should always remain our utmost concern.

Platform

As most pharmaceutical companies strive to move their clinical data management to the cloud and to find the best cloud data and analytics platform to manage their end-to-end clinical trials in a fragmented offering, many technology providers are capitalizing on the wave of generative AI to offer their platforms. As an example has launched its Artificial Intelligence Platform (AIP), and launched a number of demos.

Two niche platforms have been making headlines over the past two weeks:

1. Beaconcure Raises $14 Million Series B to Expand AI-powered Clinical Technology Platform. Developed in conjunction with its long-term partner,?, Beaconcure’s artificial intelligence platform,?Verify, streamlines and automates the clinical trial data statistical analysis process
2. Docus.ai Is Powering A Groundbreaking AI Powered Health Platform, aims to revolutionize healthcare by providing access to top medical expertise and cutting-edge AI technology. Their vision is to create a world where everyone has access to the best medical care, addressing the issue of misdiagnosis and prioritizing user privacy.

Open data Platform is also a topic and The World Health Organization () has launched data.who.int, a comprehensive digital platform aimed at providing accessible and reliable health data as a public good. The platform, supported by innovative technologies and collaborations, offers harmonized and user-friendly visualizations, multilingual accessibility, and ongoing efforts to enhance inclusivity and usability, making it a crucial resource for governments, policymakers, researchers, and the public worldwide.

And...If you have plans to develop your own platform, reminds reminds "What It Takes to Build A Platform Engineering Team". “Funding a dedicated platform engineering team without sufficient proof of the benefits is often not feasible,” says Gartner VP Analyst George. Spafford who adds to this:“Even when software engineering leaders manage to get started with platform engineering, they encounter barriers to scaling the practice. These barriers include a continued lack of buy-in and funding from executive leaders, a shortage of skills to build platforms, and insufficient staff to handle the volume of work. As a result, their early efforts struggle to gain momentum.”