AI and IVD Links We Are Tracking

Developments in artificial intelligence and in vitro diagnostics including: regulations, new products, and expert viewpoints are constant. Here we list some notable links in AI and IVD we are tracking.

TEMPUS AI WILL BUY AMBRY: Tempus AI has announced an agreement to acquire Ambry Genetics for $375 million in cash and $225 million in shares. Ambry is Tempus’ primary reference lab. The deal will combine Ambry’s extensive genetic testing capabilities with its AI-driven data analytics platform.
[READ MORE] https://www.cooley.com/news/coverage/2024/2024-11-04-tempus-announces-$600-million-acquisition-of-ambry-genetics

ROCHE BETS ON AI-DRIVEN PATHOLOGY: Roche announced it had integrated over 20 AI algorithms from eight collaborators into its digital pathology open environment. This initiative aims to enhance pathology insights, benefiting cancer patients through precision medicine and enabling targeted treatments.
[READ MORE] https://diagnostics.roche.com/global/en/news-listing/2024/roche-advances-ai-driven-cancer-diagnostics-by-expanding-its-digital-pathology-open-environment.html

NEED MARKET INTELLIGENCE IN AI AND IN VITRO DIAGNOSTICS, UPDATED REGULARLY, WITH AFTER-PURCHASE SUPPORT?

ARTIFICIAL INTELLIGENCE IN IVD MARKETS

NOT ENOUGH VALIDATION, PEER-REVIEW ON AI, SAYS STUDY: According to a Nature study – Novel products applying artificial intelligence (AI)-based methods to digital pathology images”* examined public evidence for AI-based digital pathology products, focusing on their regulatory approval, intended use, and validation studies. The study identified 26 products available in the European Economic Area and Great Britain markets, 24 of which gained regulatory approval as General IVD medical devices through self-certification. However, only 38% of these products had peer-reviewed internal validation studies, and 42% had peer-reviewed external validation studies. To promote transparency, the authors developed an online register consolidating public evidence, accessible at [https://osf.io/gb84r/], which aims to serve as a resource for decision-making regarding novel devices.

ADVAMED: REGULATE AI, BUT LET US HAVE IT: AdvaMed, the Medtech Association, commended the House of Representatives’ bipartisan Task Force on Artificial Intelligence for urging the Centers for Medicare & Medicaid Services (CMS) to develop a formalized payment pathway for AI-enabled medical devices. Sad the organization “AI tools streamline these practices and reduce time spent on services, current payment mechanisms cannot adequately reimburse these tools.” [READ MORE] https://www.advamed.org/industry-updates/news/advamed-commends-house-ai-taskforces-recommendation-to-ensure-reimbursement-expand-access-to-ai-enabled-medical-devices/

BECKMAN EXEC: “AI ESSENTIAL FOR LAB OPERATIONS”: While there’s a lot of focus on whether AI will replace doctor diagnosis, or enhance them, Tom Neufelder, described by CAP TODAY as a AI virtuoso, said in a recent panel conducted by that publication: AI is already becoming integral for – Error Reduction (Using AI and computer vision for barcode reading and tube identification). For Device Monitoring (AI can detect anomalies, notify operators, and integrate with IoT systems for proactive service). For Predictive Maintenance : Optimizes service schedules based on device performance metrics, minimizing downtime. Full panel discussion: [READ MORE] https://www.captodayonline.com/ai-virtuosos-reveal-ins-outs-hopes-doubts/

QIAGEN: 14 AI APPLICATIONS BY 2028: Qiagen, Hilden, Germany-based life-sciences and diagnostics company, plans to increase its AI-enabled applications from five to at least 14 by 2028.
WSJ [READ MORE] https://www.wsj.com/articles/qiagen-targets-higher-profitability-more-ai-enabled-applications-by-2028-3adb647c?

EU GETS TOUGH ON AI IN DEVICES: Regulatory Developments in the EU: The European Commission published the finalized text of the Artificial Intelligence Act (AIA), establishing a legal framework to promote the uptake of “human-centric and trustworthy” AI for medical devices and IVDs. [READ MORE] https://www.raps.org/News-and-Articles/News-Articles/2024/7/EU-publishes-regulation-governing-use-of-AI-in-med

WHAT DOES THE EU MEAN? “ALMOST ALL” IVDS IMPACTED, SAYS IQVIA – “For IVD Medical Devices, the vast majority require conformity assessment and would be classified as high-risk.” With this, the EU intends that all AI used within Devices (including IVDs) that require conformity assessment will be classified as high-risk. This would be in addition to any that have an AI system that is intended to be used as a safety component, In life sciences, AI investment is growing at an astonishing rate. In 2023, over $12bn worth of investment in life sciences deals with AI, machine learning or advanced analytics were announced, more than double the level in the prior two years Organisations should develop an AI governance and compliance strategy, taking advantage of existing risk management processes.” Read IQVIA’s document: [READ MORE] https://www.iqvia.com/locations/emea/blogs/2024/10/eu-ai-act-heres-how-this-will-affect-your-organisation

AI AND MASS SPEC FOR COLONSCOPY: CanSense, a Swansea University spinout, is developing AI and spectroscopy-based blood tests for detecting colorectal cancer as a supplement to or to be used in-between colonoscopies. The non-invasive, cost-effective solution is designed to improve diagnostic efficiency and quality. Partnering with Informed Genomics, CanSense aims to bring its technology to market and secure regulatory approval.
( [READ MORE] https://www.ibm.com/case-studies/informed-genomics-cansense)

KEEP AI TO “PROVEN USES” RIGHT NOW, SAYS SURVEY: Americans are most comfortable in AI’s more proven areas – analyzing data and looking at images. In other areas, less so. [READ MORE] https://www.beckershospitalreview.com/innovation/where-americans-are-ok-with-ai-in-their-healthcare.htm l)

MDIC RELEASES MODEL PCCP FOR AI DEVICE COS: The U.S. Food and Drug Administration (FDA), Health Canada, and the U.K.’s Medicines and Healthcare products Regulatory Agency (MHRA) jointly identified 10 guiding principles that can inform the development of Good Machine Learning Practice (GMLP). Based on those frameworks, the Medical Device Industry Consortium AI Work Group released a Predetermined Change Control Plan (PCCP) is essential for managing post-market changes to AI/ML-enabled medical devices, ensuring future updates are proactively communicated, validated, and maintain the safety and effectiveness of the device. Medical Device Industry Consortium AI Work Group formed to formulate policies around AI, developed a Predetermined Change Control Plan document. [READ MORE] https://mdic.org/resources/predetermined-change-control-plans-for-ai-and-ml-enabled-medical-devices/

CE MARK FOR AI-BASED SUPCIOUS LUNG NODULE FINDER: Optellum, a medical technology company based in Oxford, U.K., has obtained the CE Mark for its Virtual Nodule Clinic. This AI-powered clinical decision support software assists clinicians in identifying and monitoring patients with suspicious lung nodules detected through CT scans, facilitating early lung cancer diagnosis. [READ MORE] https://www.optellum.com/news/optellum-receives-ce-mark-for-virtual-nodule-clinic).

$25 MILLION RAISED FOR AI-BASED TEST PRODUCTS: Money is already betting on AI in IVD. One example is Viome, which secured $25 million in Series D funding, with participation from Khosla Ventures, Marc Benioff, and others. It is not the firm’s first funding either. The company appointed Kal Raman as COO, a former senior executive at Amazon, Groupon, and Samsung. Viome offers total body intelligence tests that provide personalized food recommendations and custom supplements based on microbiome and gene expression analysis.
[READ MORE] https://www.mobihealthnews.com/news/viome-secures-25m-hires-former-amazon-groupon-samsung-executive

MUTLIPLE AI COMPANIES TEAM UP ON PATHOLOGY: PathAI announced integrations with leading AI pathology companies like Paige, Visiopharm, Deep Bio, and DoMore Diagnostics. These collaborations enhance PathAI’s AISight image management system for AI-driven pathology workflows.
( [READ MORE] https://www.pathai.com/resources/pathai-announces-integrations-with-leading-ai-pathology-companies-deep-bio-domore-diagnostics-paige-and-visiopharm-through-its-aisight-image-management-system/)

Older Links:

CAN AI SOFTWARE BE REGULATED BY FDA?: “Yes, if…” is the answer. Does it perform “signal or image acquisition, processing, or analysis.” AI is regulated software if it meets conditions. Is it “intended for a healthcare professional to independently review and understand the basis of the software’s recommendation”
[READ MORE] https://www.meddeviceonline.com/doc/does-the-fda-or-eu-ivdr-consider-ivd-software-an-ivd-medical-device-0001)

SIX LAWS LAWS LAID DOWN FROM WHO: The World Health Organization’s guidelines break no new ground, but add the heft of an influential global organization to some of the same requests we’ve seen. It’s unclear if the developers will cooperate or will be made to have to.

Transparency: WHO wants documentation of the entire development of the product to be made and be made available to future users and regulators.
Risk Management: Detailed standards for construction of training models, what part is human-made and what part is machine-made. Cybersecurity – Can the AI system be compromised or misused?
Intended Use: If it says it’s a system for detecting tumors – that use should be supported with evidence and is that clearly stated.
Data Quality – Good data must go in, or bad results will come out. WHO says “it can be difficult for AI models to accurately represent the diversity of populations, leading to biases, inaccuracies or even failure.”
Privacy, in service of privacy and data protection.
Collaboration: WHO says it is essential that patients, healthcare professionals, industry representatives, and governments work as partners. For instance, even if a system is approved for a purpose, what if it strays to another, regulatory bodies must be in constant contact. [READ MORE] https://healthpolicy-watch.news/who-issues-new-diagnostics-list-and-guide-for-regulating-artificial-intelligence/

NINE ONCOLOGY TRIALS FOR IVD IN AI: GlobalData says there are 569 clinical trials focused on developing IVD devices in oncology out of 1,490 active clinical trials in oncology. Out of these – nine (9) are being conducted to test partial or full analysis by AI to improve oncology diagnosis. Small percentage of total trials, but a significant number in terms of AI’s progress. [READ MORE] https://www.latestly.com/technology/artificial-intelligence-technology-to-drive-advancements-in-vitro-diagnostic-products-for-cancer-diagnosis-5221362.html

BELFAST-BASED COMPANY GETS CERTIFIED TO MAKE AI-BASED PRODUCTS: The ISO 13485 certification validates that Sonrai has established comprehensive policies and procedures to meet the stringent requirements for developing and producing medical devices in compliance with international standards. The company says their devices “he cloud-based devices enable pathologists to upload images of resection and biopsy samples and analyze them using Sonrai’s AI algorithms. The IVD medical devices will enhance MSI (Lynch Syndrome) testing accuracy and consistency for colorectal cancer, and aid Tumor Proportion Scoring for PD-L1 in Non-Small Cell Lung Cancer samples.”

AI-BASED KIDNEY TEST: Image-based nature of this IVD test plays to AI strengths. FDA approves Minutekidney (CDK) test using AI-enhanced colometric analysis. [READ MORE] https://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfpmn/pmn.cfm?ID=K210069)

EVIDENCE OF AI USEFULNESS IN SEPSIS: AI used in Sepsis. Read more

WANT TO SELL AI TO LABS? YOU’D BETTER TALK COST-SAVING: IQVIA Survey of Clinical Labs – Cost savings emerged as the most influential factor supporting the adoption of AI in labs.[READ MORE] https://www.iqvia.com/blogs/2019/03/artificial-intelligence-and-in-vitro-diagnostics-advancing-patient-care

SHERLOCK LAUNCHING AI TEST DESIGN PLATFORM: The assay design service will allow users to access Sherlock’s novel AI algorithms to create their own tests, and is estimated to reduce assay development time from one year down to three months. [READ MORE] https://www.360dx.com/molecular-diagnostics/sherlock-bioscience-launch-free-ai-based-operating-system-crispr-dx)

AI MODEL PICKS COGNITIVE IMPAIRMENT RISK: A working ML model looked at 46 “shadow features” – features blinded to the model – for MCI (mild cognitive impairment, risk for dementia) in a high-risk stroke population and picked them. [READ MORE] https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10308219/)

IVDR COMPLIANCE FOR AI SYSTEMS: They aren’t immune and “Explainability and causability” need to be documented for artificial intelligence, just like everything else. [READ MORE] https://www.sciencedirect.com/science/article/pii/S1871678422000334)

CAMERA AUTOMATES PRE-ANALYTICAL STEPS: E-con’s AI-based image signal processor is used for debayering, color correction, tonal balancing, noise reduction, and autoexposure & auto white balance algorithms. [READ MORE] https://www.e-consystems.com/blog/camera/applications/e-cons-new-edge-ai-camera-for-automating-pre-analytic-tasks-in-in-vitro-diagnostics/)

ABBOTT’S AI-BASED MODEL FOR TROPONIN RESULTS INTERPRETATION: [READ MORE] https://emerj.com/ai-sector-overviews/artificial-intelligence-at-abbott/

THREE HEALTH SYSTEMS BACK IVD AI COMPANY: If others follow these three Midwest hospitals in saying, no we’ll design our own, thankyouvermuch. This could be a new trend of the actual providers getting into the AI company space. [READ MORE] https://www.beckershospitalreview.com/innovation/chatgpt-like-platform-hippocratic-ai-raises-15m-partners-with-3-health-systems.htm)

DEEP LEARNER PICKS HOSPITAL READMISSION BIG FACTORS: A Deep Learning-created model selected 16 selected laboratory tests out of 981 and found those tests yielded equivalent performance to predicting hospital readmission. [READ MORE] https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10148287/)

WHAT JAMA CARES ABOUT IN AI: JAMA informs potential contributors what types of AI articles it wants to see in its pages. “We seek investigations into the role of AI in addressing the challenges clinicians face in medical training and in the practice of medicine,” among others. They are also making clear they want diversity explored or lack of diversity in developing AI models, to track any problems in lack of usefulness for all patient populations.
Read more

AI CAN MAKE BETTER DOCTORS – NY TIMES: Beyond saving us time, the intelligence in A.I. — if used well — could make us better at our jobs. [READ MORE] https://www.nytimes.com/2023/07/06/opinion/artificial-intelligence-medicine-healthcare.html)

AI PREDICTS 88% OF MALARIA CASES: [READ MORE] https://www.frontiersin.org/articles/10.3389/fmala.2023.1148115/full)

NEED MARKET INTELLIGENCE IN AI AND IN VITRO DIAGNOSTICS, INCLUDING MARKET ESTIMATES, SIGNIFICANT PLAYERS, NEW TECH AND MORE, UPDATED ON A REGULAR BASIS AND WITH AFTER PURCHASE SUPPORT?