BOSTON (PRWEB) November 06, 2020

PathAI, a global provider of AI-powered technology applied to pathology research, today announced the results of a research collaboration with Gilead that retrospectively analyzed liver biopsies from participants in clinical trials evaluating treatments for NASH or CHB (1). Using digitized hematoxylin and eosin (H&E)-, picrosirius red-, and trichrome-stained biopsy slides, PathAIs machine learning (ML) models were able to accurately predict changes in features traditionally used as markers for liver disease progression in clinical practice and clinical trials, including fibrosis, steatosis, hepatocellular ballooning, and inflammation. The new results will be presented in an oral presentation and 4 poster sessions at The Liver Meeting Digital Experience (TLMdX) that will be held from November 13-16, 2020.

The data builds upon PathAIs previous success in retrospectively staging liver biopsies from clinical trials by showing that ML models may uncover patterns of histological features that correlate with disease progression or treatment response. Furthermore, ML models were able to estimate the hepatic venous pressure gradient (HVPG) in study subjects with NASH related cirrhosis and quantify fibrosis heterogeneity from digitized slides, which are measures that are not reliably captured by traditional pathology methods. After appropriate clinical validation, these new tools could be useful in staging disease more accurately than can be done with current approaches.

"We continue to use machine learning to advance our understanding of liver diseases, including NASH and hepatitis B, as a foundation for developing new methods to track disease progression and assess response to therapeutics, said PathAI co-founder and Chief Executive Officer Andy Beck MD, PhD. Our long-standing partnership with Gilead continues to demonstrate the power of AI-based pathology to support development efforts to bring new therapies to patients."

Highlights include:

Data presented at AASLD demonstrate the potential of machine learning approaches to improve our assessment of liver disease severity, reduce the variability of human interpretation of liver biopsies, and identify novel features associated with disease progression, said Rob Myers, MD, Vice President, Liver Inflammation/Fibrosis, Gilead Sciences. We are proud of our ongoing partnership with PathAI and look forward to continued collaboration toward our shared goals of enhancing research efforts and improving outcomes of patients with liver disease.

The antiviral drug TDF effectively suppresses hepatitis B virus in patients with CHB, but a small subset of patients have persistently elevated serum ALT despite virologic suppression. ML-models were applied to biopsy data from registrational studies of TDF to examine this small subgroup of non-responders. Analyses of the ML-model predicted histologic features showed that persistently elevated ALT after five years of TDF treatment is associated with a higher steatosis score at BL and increases in steatosis during follow-up. These data suggest that subjects with elevated ALT despite TDF treatment may have underlying fatty liver disease that impacts biochemical response.Machine Learning Enables Quantitative Assessment of Histopathologic Signatures Associated with ALT Normalization in Chronic Hepatitis B Patients Treated with Tenofovir Disoproxil Fumarate (TDF) Oral Abstract #18

ML-models were deployed on biopsies from registrational trials of TDF in CHB to identify cellular and tissue-based phenotypes associated with HBV DNA and hepatitis B e-antigen (HBeAg). The study demonstrated that proportionate areas of ML-model-predicted hepatocellular ballooning at BL and Yr 5, and lobular inflammation at Yr 5 were higher in subjects that did not achieve virologic suppression. In addition, lymphocyte density across the tissue and within regions of lobular inflammation correlated with HBeAg loss, supporting the importance of an early immune response for viral clearance.Machine Learning Based Quantification of Histology Features from Patients Treated for Chronic Hepatitis B Identifies Features Associated with Viral DNA Suppression and dHBeAg Loss Poster Number #0848

Standard manual methods for staging liver fibrosis have limited sensitivity and reproducibility. Application of a ML-model to evaluate changes in fibrosis in response to treatment in the STELLAR and ATLAS trials enabled development of the DELTA (Deep Learning Treatment Assessment) Liver Fibrosis Score. This scoring method accounts for the heterogeneity in fibrosis severity that can be detected by ML-models and reflects changes in fibrotic patterns that occur in response to treatment. Application of the DELTA Liver Fibrosis Score to biopsies from the Phase 2b ATLAS trial demonstrated a reduction in fibrosis in response to treatment with the investigational combination of cilofexor and firsocostat that was not detected by standard staging methods. Validation of a Machine Learning-Based Approach (DELTA Liver Fibrosis Score) for the Assessment of Histologic Response in Patients with Advanced Fibrosis Due to NASH Poster Number #1562

Integration of tissue transcriptomic data with histologic information is likely to reveal new insights into disease. Using liver tissue obtained during the STELLAR trials evaluating NASH subjects with advanced fibrosis, RNA-seq-generated, tissue-level gene expression profiles were integrated with ML-predicted histology. This analysis revealed five key genes strongly correlated with proportionate areas of portal inflammation and bile ducts, features that are themselves predictive of disease progression in NASH. High levels of expression of these genes was associated with an increased risk of progression to cirrhosis in subjects with bridging (F3) fibrosis (hazard ratio [HR] 2.1; 95% CI 1.25, 3.49) and liver-related clinical events among those with cirrhosis (HR 4.05; 95% CI 1.4, 14.36). Integration of Machine Learning-Based Histopathology and Hepatic Transcriptomic Data Identifies Genes Associated with Portal Inflammation and Ductular Proliferation as Predictors of Disease progression in Advanced Fibrosis Due to NASH Poster Number #595

The severity of portal hypertension as assessed by HPVG predicts the risk of hepatic complications in patients with liver disease but is not simple to measure. ML-models were trained on images of 320 trichrome-stained liver biopsies from a phase 2b trial of investigational simtuzumab in subjects with compensated cirrhosis due to NASH to recognize patterns of fibrosis that correlate with centrally-read HVPG measurements. Deployed on a test set of slides, ML-calculated HVPG scores strongly correlated with measured HVPG and could discriminate subjects with clinically-significant portal hypertension (HVPG 10 mm Hg).A Machine Learning Model Based on Liver Histology Predicts the Hepatic Venous Pressure Gradient (HVPG) in Patients with Compensated Cirrhosis Due to Nonalcoholic Steatohepatitis (NASH) Poster Number #1471

(1) Trials include STELLAR, ATLAS, and NCT01672879 for investigation of NASH therapies, and registrational studies GS-US-174-102/103 for tenofovir disoproxil fumarate [TDF] for CHB.

About PathAIPathAI is a leading provider of AI-powered research tools and services for pathology. PathAIs platform promises substantial improvements to the accuracy of diagnosis and the efficacy of treatment of diseases like cancer, leveraging modern approaches in machine and deep learning. Based in Boston, PathAI works with leading life sciences companies and researchers to advance precision medicine. To learn more, visit pathai.com.

Share article on social media or email:

Original post:
PathAI and Gilead Report Data from Machine Learning Model Predictions of Liver Disease Progression and Treatment Response at AASLD's The Liver Meeting...

Select your countryUnited StatesCanadaAfghanistanAlbaniaAlgeriaAmerican SamoaAndorraAngolaAntigua and BarbudaArgentinaArmeniaAustraliaAustriaAzerbaijanBahamasBahrainBangladeshBarbadosBelarusBelgiumBelizeBeninBermudaBhutanBoliviaBosnia and HerzegovinaBotswanaBrazilBruneiBulgariaBurkina FasoBurundiCambodiaCameroonCanadaCape VerdeCayman IslandsCentral African RepublicChadChileChinaColombiaComorosCongo, Democratic Republic of theCongo, Republic of theCosta RicaCte d'IvoireCroatiaCubaCyprusCzech RepublicDenmarkDjiboutiDominicaDominican RepublicEast TimorEcuadorEgyptEl SalvadorEquatorial GuineaEritreaEstoniaEthiopiaFaroe IslandsFijiFinlandFranceFrench PolynesiaGabonGambiaGeorgiaGermanyGhanaGreeceGreenlandGrenadaGuamGuatemalaGuineaGuinea-BissauGuyanaHaitiHondurasHong KongHungaryIcelandIndiaIndonesiaIranIraqIrelandIsraelItalyJamaicaJapanJordanKazakhstanKenyaKiribatiNorth KoreaSouth KoreaKosovoKuwaitKyrgyzstanLaosLatviaLebanonLesothoLiberiaLibyaLiechtensteinLithuaniaLuxembourgMacedoniaMadagascarMalawiMalaysiaMaldivesMaliMaltaMarshall IslandsMauritaniaMauritiusMexicoMicronesiaMoldovaMonacoMongoliaMontenegroMoroccoMozambiqueMyanmarNamibiaNauruNepalNetherlandsNew ZealandNicaraguaNigerNigeriaNorthern Mariana IslandsNorwayOmanPakistanPalauPalestine, State ofPanamaPapua New GuineaParaguayPeruPhilippinesPolandPortugalPuerto RicoQatarRomaniaRussiaRwandaSaint Kitts and NevisSaint LuciaSaint Vincent and the GrenadinesSamoaSan MarinoSao Tome and PrincipeSaudi ArabiaSenegalSerbiaSeychellesSierra LeoneSingaporeSint MaartenSlovakiaSloveniaSolomon IslandsSomaliaSouth AfricaSpainSri LankaSudanSudan, SouthSurinameSwazilandSwedenSwitzerlandSyriaTaiwanTajikistanTanzaniaThailandTogoTongaTrinidad and TobagoTunisiaTurkeyTurkmenistanTuvaluUgandaUkraineUnited Arab EmiratesUnited KingdomUnited StatesUruguayUzbekistanVanuatuVatican CityVenezuelaVietnamVirgin Islands, BritishVirgin Islands, U.S.YemenZambiaZimbabwe

Select your industryAd AgenciesAgricultureApparelAutomotiveBiotechnologyChemicalsConstructionConsultingConsumer GoodsEducationEnergyEngineeringEntertainmentEnvironmentalFinance & BankingFood & BeverageGovernmentHealth CareHospitalityInsuranceManufacturingMediaNot For ProfitRecreationRetailSecurityServicesTechnologyTelecommunicationsTransportationTravel and LeisureUtilitiesWholesaleOther (please specify)

Privacy Policy

By submitting this form to MIT SMR, you acknowledge that your name and contact information will be shared with SAS Institute Inc., which may contact you regarding the content.

By submitting this form to MIT SMR, you acknowledge that your name and contact information will be shared with SAS Institute Inc., which may contact you regarding the content.

This field is for validation purposes and should be left unchanged.

Here is the original post:
Free Webinar | Machine Learning and Data Analytics in the Pandemic Era - MIT Sloan

Microsoft has releasedLobe, a free desktop application that lets Windows and Mac users create customized AI models without writing any code. Several customers are already using the app for tracking tourist activity around coral reefs, the company said.

Lobeis available on Windows and Mac as a desktop app. Presently it only supports image classification by categorizing the image to a single label overall. Microsoft says that there will be new releases supporting other neural networks in the near future.

To create an AI in Lobe, a user first needs to import a collection of images. These images are used as a dataset to train the application. Lobe analyzes the input images and sifts through a built-in library of neural network architectures to find the most suitable model for processing the dataset. Then it trains the model on the provided data, creating an AI model optimized to scan images for the users specific object or action.

AutoML is a technology that can automate parts and most of the machine learning creation workflow, reducing the advancement costs. Microsoft has made AutoML features available to enterprises in its Azure public cloud. The existing AI tools in Azure target only advanced projects. Lobe being free, easy to access, and convenient to use can now support even simple use cases that were not adequately addressed by the existing AI tools.

The Nature Conservancy is a nonprofit environmental organization that used Lobe to create an AI. This model analyzes the pictures taken by tourists in the Caribbean to identify where and when visitors interact with coral reefs. A Seattle auto marketing firm,Sincro LLC,has developed an AI model that scans vehicle images in online ads to filter out pictures that are less appealing to the customers.

GitHub: https://github.com/lobe

Website: https://lobe.ai/

Related

Read more here:
Microsoft Introduces Lobe: A Free Machine Learning Application That Allows You To Create AI Models Without Coding - MarkTechPost

Amwell is looking to evolve virtual care beyond just imitating in-person care.

To do that, the telehealth companyexpects to use its latestpartnership with Google Cloud toenable it to tap into artificial intelligence and machine learning technologies to create a better healthcare experience, according to Peter Antall, M.D., Amwell's chief medical officer.

"We have a shared vision to advance universal access to care thats cost-effective. We have a shared vision to expand beyond our borders to look at other markets. Ultimately, its a strategic technology collaboration that were most interested in," Antall said of the company's partnership with the tech giant during a STATvirtual event Tuesday.

"What we bring to the table is that we can help provide applications for those technologiesthat will have meaningful effects on consumers and providers," he said.

The use of AI and machine learning can improve bot-based interactions or decision support for providers, he said. The two companies also want to explore the use of natural language processing and automated translation to provide more "value to clients and consumers," he said.

Joining a rush of healthcare technology IPOs in 2020, Amwell went public in August, raising$742 million. Google Cloud and Amwell also announced amultiyear strategic partnership aimed at expanding access to virtual care, accompanied by a$100 million investmentfrom Google.

During an HLTH virtual event earlier this month, Google Cloud director of healthcare solutions Aashima Gupta said cloud and artificial intelligence will "revolutionize telemedicine as we know it."

RELATED:Amwell files to go public with $100M boost from Google

"There's a collective realization in the industry that the future will not look like the past," said Gupta during the HTLH panel.

During the STAT event, Antall said Amwellis putting a big focus onvirtual primary care, which has become an area of interest for health plans and employers.

"It seems to be the next big frontier. Weve been working on it for three years, and were very excited. So much of healthcare is ongoing chronic conditions and so much of the healthcare spend is taking care ofchronic conditionsandtaking care of those conditions in the right care setting and not in the emergency department," he said.

The companyworks with 55 health plans, which support over 36,000 employers and collectively represent more than 80million covered lives, as well as 150 of the nations largest health systems. To date, Amwell says it has powered over 5.6million telehealth visits for its clients, including more than 2.9million in the six months ended June 30, 2020.

Amwell is interested in interacting with patients beyond telehealth visits through what Antall called "nudges" and synchronous communication to encouragecompliance with healthy behaviors, he said.

RELATED:Amwell CEOs on the telehealth boom and why it will 'democratize' healthcare

It's an area where Livongo, recently acquired by Amwell competitor Teladoc,has become the category leader by using digital health tools to help with chronic condition management.

"Were moving into similar areas, but doing it in a slightly different matter interms of how we address ongoing continuity of care and how we address certain disease states and overall wellness," Antallsaid, in reference to Livongo's capabilities.

The telehealth company also wants to expand into home healthcare through the integration of telehealth and remote care devices.

Virtual care companies have been actively pursuing deals to build out their service and product lines as the use of telehealth soars. To this end, Amwell recently deepened its relationship with remote device company Tyto Care. Through the partnership, the TytoHome handheld examination device that allows patients to exam their heart, lungs, skin, ears, abdomen, and throat at home, is nowpaired withAmwells telehealth platform.

Looking forward, there is the potential for patients to getlab testing, diagnostic testing, and virtual visits with physicians all at home, Antall said.

"I think were going to see a real revolution in terms ofhow much more we can do in the home going forward," he said.

RELATED:Amwell's stock jumps on speculation of potential UnitedHealth deal: media report

Amwell also is exploring the use of televisions in the home to interact with patients, he said.

"We've done work with some partners and we're working toward a future where, if it's easier for you to click your remote and initiate a telehealth visit that way, thats one option. In some populations, particularly the elderly, a TV could serve as a remote patient device where a doctor or nurse could proactively 'ring the doorbell' on the TV and askto check on the patient," Antall said.

"Its video technology that'salready there in most homes, you just need a camera to go with it and a little bit of software.Its one part of our strategy to be available for the whole spectrum of care and be able to interact in a variety of ways," he said.

Go here to read the rest:
Amwell CMO: Google partnership will focus on AI, machine learning to expand into new markets - FierceHealthcare

(Pixabay)

The extraordinary advances in machine learning that drive the increasing accuracy and reliability of artificial intelligence systems have been matched by a corresponding growth in malicious attacks by bad actors seeking to exploit a new breed of vulnerabilities designed to distort the results.

Microsoft reports it has seen a notable increase in attacks on commercial ML systems over the past four years. Other reports have also brought attention to this problem.Gartner's Top 10 Strategic Technology Trends for 2020, published in October 2019, predicts that:

Through 2022, 30% of all AI cyberattacks will leverage training-data poisoning, AI model theft, or adversarial samples to attack AI-powered systems.

Training data poisoning happens when an adversary is able to introduce bad data into your model's training pool, and hence get it to learn things that are wrong. One approach is to target your ML's availability; the other targets its integrity (commonly known as "backdoor" attacks). Availability attacks aim to inject so much bad data into your system that whatever boundaries your model learns are basically worthless. Integrity attacks are more insidious because the developer isn't aware of them so attackers can sneak in and get the system to do what they want.

Model theft techniques are used to recover models or information about data used during training which is a major concern because AI models represent valuable intellectual property trained on potentially sensitive data including financial trades, medical records, or user transactions.The aim of adversaries is to recreate AI models by utilizing the public API and refining their own model using it as a guide.

Adversarial examples are inputs to machine learning models that attackers haveintentionally designed to cause the model to make a mistake.Basically, they are like optical illusions for machines.

All of these methods are dangerous and growing in both volume and sophistication. As Ann Johnson Corporate Vice President, SCI Business Development at Microsoft wrote in ablog post:

Despite the compelling reasons to secure ML systems, Microsoft's survey spanning 28 businesses found that most industry practitioners have yet to come to terms with adversarial machine learning. Twenty-five out of the 28 businesses indicated that they don't have the right tools in place to secure their ML systems. What's more, they are explicitly looking for guidance. We found that preparation is not just limited to smaller organizations. We spoke to Fortune 500 companies, governments, non-profits, and small and mid-sized organizations.

Responding to the growing threat, last week, Microsoft, the nonprofit MITRE Corporation, and 11 organizations including IBM, Nvidia, Airbus, and Bosch released theAdversarial ML Threat Matrix, an industry-focused open framework designed to help security analysts to detect, respond to, and remediate threats against machine learning systems. Microsoft says it worked with MITRE to build a schema that organizes the approaches employed by malicious actors in subverting machine learning models, bolstering monitoring strategies around organizations' mission-critical systems.Said Johnson:

Microsoft worked with MITRE to create the Adversarial ML Threat Matrix, because we believe the first step in empowering security teams to defend against attacks on ML systems, is to have a framework that systematically organizes the techniques employed by malicious adversaries in subverting ML systems. We hope that the security community can use the tabulated tactics and techniques to bolster their monitoring strategies around their organization's mission critical ML systems.

The Adversarial ML Threat, modeled after the MITRE ATT&CK Framework, aims to address the problem with a curated set of vulnerabilities and adversary behaviors that Microsoft and MITRE vetted to be effective against production systems. With input from researchers at the University of Toronto, Cardiff University, and the Software Engineering Institute at Carnegie Mellon University, Microsoft and MITRE created a list of tactics that correspond to broad categories of adversary action.

Techniques in the schema fall within one tactic and are illustrated by a series of case studies covering how well-known attacks such as the Microsoft Tay poisoning, the Proofpoint evasion attack, and other attacks could be analyzed using the Threat Matrix. Noted Charles Clancy, MITRE's chief futurist, senior vice president, and general manager of MITRE Labs:

Unlike traditional cybersecurity vulnerabilities that are tied to specific software and hardware systems, adversarial ML vulnerabilities are enabled by inherent limitations underlying ML algorithms. Data can be weaponized in new ways which requires an extension of how we model cyber adversary behavior, to reflect emerging threat vectors and the rapidly evolving adversarial machine learning attack lifecycle.

Mikel Rodriguez, a machine learning researcher at MITRE who also oversees MITRE's Decision Science research programs, said that AI is now at the same stage now where the internet was in the late 1980s when people were focused on getting the technology to work and not thinking that much about longer term implications for security and privacy. That, he says, was a mistake that we can learn from.

The Adversarial ML Threat Matrix will allow security analysts to work with threat models that are grounded in real-world incidents that emulate adversary behavior with machine learning and to develop a common language that allows for better communications and collaboration.

Read more from the original source:
Microsoft/MITRE group declares war on machine learning vulnerabilities with Adversarial ML Threat Matrix - Diginomica