August 2024
|
00 issues in this vol.

XEUREKA, FORTANIX AND NVIDIA PARTNER TO ENHANCE ACCURACY OF AI-POWERED DRUG DISCOVERY

INNOVATION IN DRUG DISCOVERY

Xeureka, a subsidiary of Mitsui & Co., recently announced improvements in its artificial intelligence (AI) models for drug discovery, boosting accuracy from 65% to 74% by leveraging Fortanix and NVIDIA's Confidential Computing technologies. Utilizing the latest advancements in data science and computer hardware, Xeureka effectivelyremoved sensitive and confidential information from a multi-source dataset,overcoming one of the largest barriers to the development of AI in healthcare.

Drug discovery carries a high cost burden, typically between $1-2 billion per new drug, and can take between 10 and 20 years, both of which Xeureka aims to reduce through the introduction of their improved models, decreasing the barrier for new treatments to reach patients

PAIGE & MICROSOFT LAUNCH NEW AI MODEL FOR CANCER PATHOLOGY

MEDICAL APPLICATION OF EMERGING TECHNOLOGY

Partners Paige & Microsoft have announced the launch of Virchow2, the second largest cancer pathology AI model to date. Utilizing over 3 million path slides and trained using de-identified data from over 225,000patients, these models set a new record in AI trainingscale, surpassing previous performance standards, as showcased in a recent technical report. This enormousdataset, created in collaboration with Microsoft, offersunparalleled diversity and depth, including over 40tissue types stained with H& E and diverse immunestains (IHC), making it suited to a wider variety of applications.

As computational pathology solutions near the stage of practical use, they bring the potential of increasingly accurate diagnoses, increased chance of detecting cancer early in its progression, and ultimately improved treatment outcomes for patients.

FDA CLEARS AI-POWERED DIAGNOSTIC TOOLS FOR CARDIAC AND BRAIN IMAGING

REGULATION OF EMERGING TECHNOLOGY

The FDA has recently provided 510(k) clearances for several AI-powered diagnostic tools, including products from AISAP and MRIMath. AISAP's CARDIO software can diagnose up to 90% of common cardiacstructural and functional parameters at the bedsidewith high accuracy, demonstrating 93% sensitivity and specificity in clinical trials. MRIMath's i2Contour device for glioblastomamanagement has markedly increased efficiency by saving up to an hour per imaging study and achieved95% accuracy for T1C sequences, offering substantial improvements in patient care and reducing inter-uservariability.

The clearances of AISAP's CARDIO and MRIMath’s i2Contour indicate the FDA's willingness to explore innovative solutions to improving accuracy in medicine, and pave the way for future AI-powered tools to reach the market and improve outcomes across patient populations.

FDA CONTINUES TO DELIBERATE ON RESPONSIBILITY TO REGULATE AI

REGULATION OF EMERGING TECHNOLOGY

Pharmaceutical companies are increasingly integrating AI into their research, but the FDA is still determining its regulatory role, aiming to issue guidance later this year.At a recent meeting with industry executives andacademics, data quality, transparency and explicabilitywere identified as major challenges to creating apredictable regulatory environment for emerging technologies. To expedite the process of regulatory adoption, FDA officials have encouraged companies to bring AI pilot programs to the agency to ensure new solutions meet standards throughout the development process.

While the FDA works to determine its role in the landscape of emerging technology, the lack of clarity in current guidance may limit industry investment the technology's manufacturing uses, and delay the integration of AI-based solutions in practical treatment settings.

EVALUATING & ENHANCING TRUST IN MEDICAL AI: THE METRIC-FRAMEWORK

REGULATION OF EMERGING TECHNOLOGY

Inspired by the influx of AI and ML applications in medicine, a systematic review of academic studies on emerging technology recently identified theimportance of evaluating the data quality within new tools as part of the regulatory approval process. From an analysis of 5408 studies, 120 eligible records weresynthesized to develop the METRIC-framework, aspecialized data quality framework designed to improve the trustworthiness of AI in medicine,incorporating 15 dimensions to evaluate medical ML training datasets.

As regulatory bodies like the FDA continue to explore their role in allowing emerging technologies like AI and ML to reach the medical landscape, standardized criteria like the METRIC-framework have the potential to guide faster approval decisions and expedite the time required for new solutions to reach patients.

ACCELERATING DRUG DISCOVERY WITH AI: SANOFI COLLABORATES WITH FORMATION BIO AND OPENAI

INNOVATION IN DRUG DISCOVERY

Sanofi, Formation Bio, and OpenAI recently announced their partnership in developing AI-powered software for the pharmaceutical industry, using Sanofi’s proprietary data in the creation of industry-tuned AI models to optimize drug discovery. Initial focus areas for collaboration will include the design of new drug compounds, identifying existing drugs with repurposing potential, and optimizing clinical trial processes. In 2023, AI applications enabled Sanofi to identify 90 new drug targets, advance seven in its pipeline,and expedite molecule progression in 12 therapeutic indications, which the new partnership will continue to build upon.

These advances are expected to reduce the typical dozen-year timeline between molecule discovery and patient availability, with AI interventions applied to 75% of Sanofi's small-molecule portfolio.
contact_imgcontact_img_mob

contact us

Get In Touch With Us

Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.

Stay Informed with the Latest Insights

Get expert analysis, industry trends, and exclusive updates delivered straight to your inbox. Stay ahead of the curve with valuable insights that drive success.

Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.