In Preview: Bringing generative AI to Azure AI Health Bot
In the current era of Large Language Models (LLMs), there is a growing demand for AI in healthcare. Healthcare organizations are actively exploring ways to leverage these advanced technologies to develop their own GPT-powered copilot experiences for doctors or virtual health assistants for patients. It's important to understand that healthcare organizations will only use these tools when they adhere to the highest security and compliance standards required for healthcare purposes The escalating demand is driven by the recognition that AI systems can significantly enhance healthcare experiences with many different tasks, such as assisting with administrative or clinical tasks. The goal is to build intelligent and engaging chat experiences that utilize generative AI, providing accurate, relevant, and consistent information based on credible healthcare information or validated customer sources.*Updated on 1/06/2021* Quick Start: Setting Up Your COVID-19 Healthcare Bot
The Healthcare Bot Service is an AI-powered, extensible, secure and compliant healthcare experience. Microsoft now has a specific template pre-built for COVID-19. It takes inbound requests, asks about the patient’s symptoms, and assists in getting people access to trusted and relevant healthcare services and information based on the CDC recommendation.
Modernize DevSecOps and GitOps journey with Microsoft’s Unified solution (Azure DevOps + GitHub)
Modernize your DevSecOps and GitOps journey with Microsoft’s Unified solution and best-in-class tools (Azure DevOps + GitHub) to Simplify, Automate, Secure entire software supply chain including containers, and Govern each phase with shift-left approach.Advancements in Healthcare AI models
In 2024 we were excited to introduce the healthcare AI models in Azure AI Foundry model catalog. The collection included 18 open-source models for healthcare and life sciences spanning across first party, Microsoft Research, partner and third-party models from NVIDIA and Hugging Face. This launch represented a significant milestone, showcasing our commitment to providing accessible and advanced AI models to healthcare developers around the globe. Since that pivotal moment, there have been substantial advancements. Our team has been diligently enhancing the Managed Compute offering in Foundry models, ensuring that we consistently expand our catalog for the benefit of our partners and customers.Now, we have more than 30 curated healthcare and life sciences models in the catalog — and many more are in the pipeline. Asset protection for AI models is now available The current Foundry model catalog, enriched with open source models, stands as a testament to our dedication to democratizing AI for healthcare. These models encourage collaboration and innovation within the developer community. Managed Compute now provides asset protection and a comprehensive monetization framework that brings significant benefits to model providers, driving both security and economic advantages. Enhanced security ensures that proprietary models are safeguarded against unauthorized access, maintaining the confidentiality of valuable assets and security. Conversely, open-source models provide benefits such as transparency, flexibility, and accelerated innovation. Dimension IP-Protected Models Open-Source Models Access Control Strict access policies; hosted in secure environments; assets are encrypted or containerized Fully open; accessible to anyone unless explicitly licensed Code and weights Protection Code is not exposed; inference can be done on managed compute without asset leakage Source code and model weights are fully exposed; no protection from asset theft Customization Limited; requires collaboration with provider Highly customizable; developers can retrain or modify Support & SLAs Bundled with Azure SLAs and support structures Community-driven with no guaranteed support or uptime Innovation Cycle Focused iteration with added security reviews and controlled environments Rapid experimentation and evolution due to global developer contributions Security & Compliance Fully auditable; can meet ISO, HIPAA, GDPR, etc., through managed infrastructure Hard to certify or audit for compliance in real-world scenarios Microsoft collaborates closely with partners such as Independent Software Vendors (ISVs), Academic Medical Centers (AMC), providers, payors and pharma companies to build high-quality, sustainable solutions including development and publishing AI models. Understanding Asset Protection for Models Asset protection remains a paramount concern for model providers who offer unique and proprietary AI solutions. To address this, we have implemented a deployment template feature, preventing unauthorized access to model assets, such as model weights and inference runtimes. Asset protection is ensured on two fronts: Model assets and runtime containers are not directly accessible by end users. Model Metadata and the actual data are separated. Public storage container registries will allow users to read metadata of models, deployment templates, and environments (with container information), but actual data such as model weights and containers are behind the network boundary and protected with separate permission control. Access to protected registries in production tenant is strictly governed and controlled. End users cannot tweak the behavior of the container that serves the models. Deployment template is authored and validated by model providers. Only validated deployment templates are allowed for deployment. Closed deployment process ensures that the container serving the model is not tweaked by end users and only behave as intended/validated. If an end user tries to use an arbitrary container to serve the model, the request is rejected because the model is enforced to use the approved deployment template only. This method keeps model assets proprietary while enabling users to deploy models within their subscription. Expanding models into pharma and life sciences AI's importance in drug discovery cannot be overstated, as it revolutionizes the way researchers tackle complex biological challenges. The Microsoft Research (MSR) team has been actively developing innovative protein models, which are now being integrated into the Azure AI Foundry catalog. EvoDiff: A diffusion based generative model of protein sequences that can be used to design novel proteins with desirable properties otherwise inaccessible to structure-based models. BioEmu: The first model that generates structural ensembles with experimental accuracy capturing the dynamic flexibility of proteins that underpins protein function and revealing insights that static models miss. Current research stages showcase AI's capability to accelerate protein folding, sequence generation, and molecular design, significantly reducing the time and cost associated with drug development. AI-driven models, such as NVIDIA's BioNeMo blueprint, exemplify the power of machine learning in predicting high-stability proteins and identifying effective drug targets. These models are available in Foundry Models catalog for Biology customers now, with additional models from their Generative Virtual Screening blueprint coming soon. ProteinMPNN generates and optimizes protein sequences, predicting high-stability, functional proteins. It supports innovation in protein-based drug development and synthetic biology. OpenFold2 predicts three-dimensional protein structures from amino acid sequences. It helps identify drug targets and design effective pharmaceuticals. RfDiffusion simulates molecular diffusion across cellular environments, offering insights into transport mechanisms and interactions. It's essential for studying signal transduction, metabolic pathways, and drug delivery systems, aiding in therapeutic strategy development. MSA-search (Multi Sequence Alignment) aligns multiple protein sequences to identify similarities and differences, crucial for comparative genomics and evolutionary biology. It helps understand evolutionary relationships and functional conservation, advancing genetic research and evolutionary studies. Pathology Foundation models to improve diagnostic accuracy Paige.ai in partnership with Microsoft Research has developed state-of-the-art digital pathology foundation models over the years. Virchow : The first million-slide-level foundation model proven to boost diagnostic accuracy across pan-cancer pathology applications. Virchow2 : Builds on Virchow with enhanced performance, striking the optimal balance between computational efficiency and diagnostic precision. Virchow2G : A large-scale Virchow-2 variant optimized for maximum downstream application accuracy. Virchow2G-mini : A compact Virchow-2 variant tuned for high throughput with minimal compromise to diagnostic performance. Prism : One of the first multi-modal, slide-level foundation models in digital pathology—excelling in both diagnostic classification and biomarker prediction. While bringing models in the catalog is a significant achievement, the true potential is unlocked by connecting multiple models with agents using the Model Context Protocol (MCP) and Model Agents. This integration allows for seamless collaboration between diverse models, enhancing their collective capabilities and providing more comprehensive insights. By leveraging MCP and Model Agents, users can maximize the functionality of each model, leading to more accurate predictions, improved diagnostics, and optimized therapeutic strategies. Starting MSBuild’25, we begin our agentic journey with a recently launched healthcare agent orchestrator in Azure AI Foundry.Building Healthcare Research Data Platform using Microsoft Fabric
Co-Authors: Manoj Kumar, Mustafa Al-Durra PhD, Kemal Kepenek, Matt Dearing, Praneeth Sanapathi, Naveen Valluri Overview Research data platforms in healthcare providers, academic medical centers (AMCs), and research institutes support research, clinical decision making, and innovation. They consolidate data from various sources, making it accessible for comprehensive analysis and fostering collaboration among research teams. These platforms automate data collection, processing, and delivery, reducing time and effort needed for data management. This allows researchers to focus on their core activities while ensuring data security and regulatory compliance. The ability to work with multimodal data encourages interdisciplinary and interorganizational collaboration, uniting experts to address complex healthcare challenges. Current challenges Researchers face many common challenges as they work with multimodal healthcare data: Data integration and curation: The process of integrating various data types, such as clinical notes, imaging data, genomic information, and sensor data, presents significant challenges due to differences in formats, standards, and sources. Each AMC employs unique methods for data curation, with some utilizing on-premises solutions and others adopting hybrid cloud systems. No standardized approach currently exists for data curation, necessitating considerable organizational efforts to ensure data consistency and quality. Furthermore, data deidentification is often required to safeguard patient privacy. Data discovery and building cohorts: The lack of a unified multimodal data platform leads to the segregation of data across different modalities. Cohort discovery for each modality is performed separately and often lacks a self-service option, necessitating additional human resources to assist researchers in the data discovery process. This issue is particularly significant because researchers who require Institutional Review Board (IRB) approval cannot access the data beforehand but still need an effective method to identify and explore cohorts. Data delivery: Sensitive patient data, after institutional review board approval, must comply with privacy regulations like the Health Insurance Portability and Accountability Act (HIPAA) and the General Data Protection Regulation (GDPR), requiring secure transfer to prevent breaches. The data, sourced from various systems, needs processing for research readiness. Delivering unified data from modalities like imaging, genomics, and health records is challenging. Typically, research IT teams curate cohort data and deliver it to an SQL database or a file share, accessed by researchers via secure virtual machines. This method often leads to data duplication, creating significant overhead due to numerous ongoing research projects. Cost management: Research projects are funded by government grants and private organizations. Managing these costs is challenging. Research IT departments often implement chargebacks for transparency and accountability in resource use. However, there is a disconnect between funding models and operations. Research teams favor capital expenditure (CapEx) with upfront funding for long-term resources, while cloud platforms operate on operational expenditure (OpEx), incurring ongoing costs based on usage. This shift can lead to concerns about unpredictable costs and budgeting difficulties. Bridging this gap requires careful planning, communication, and hybrid financial strategies to align research needs with cloud-based systems. Compliance with regulations: Healthcare research uses sensitive patient data, requiring strict adherence to HIPAA and GDPR. Transparency in data handling is essential but complex. Researchers must document disclosures thoroughly, detailing who accessed the data and for what purpose. However, tracking and auditing are often fragmented due to inconsistent systems. Variability in disclosure requirements from different agencies adds to compliance challenges. Balancing an auditable trail with privacy and manageable administrative tasks is crucial. Research data platform requirements Ability to curate multi modal data into the research data platform Ability for researchers to identify cohorts (without seeing data) to submit data requests to IRB Automated data delivery after IRB workflow approves the request to access relevant data Tools for researchers as part of the same platform Secure and regulatory-compliant environment for research. An approach to building a research data platform using Microsoft Fabric This article serves as a guide to healthcare organizations, offering a point of view and a prescriptive guidance on building a research data platform using Microsoft Fabric. The solution uses several features from healthcare data solutions in Microsoft Fabric, including its discover and build cohorts capability, and features from the Fabric platform. Microsoft Fabric: is a unified, AI-powered data platform designed to simplify data management and analytics. It integrates various tools and services to handle every stage of the data lifecycle, including ingestion, preparation, storage, analysis, and visualization. Fabric is built on a Software as a Service (SaaS) foundation, offering seamless experience for organizations to make data-driven decisions. For additional details, refer to the following link: What is Microsoft Fabric - Microsoft Fabric | Microsoft Learn Healthcare data solutions in Fabric: Healthcare data solutions in Fabric help you accelerate time to value by addressing the critical need to efficiently transform healthcare data into a suitable format for analysis. With these solutions, you can conduct exploratory analysis, run large-scale analytics, and power generative AI with your healthcare data. By using intuitive tools such as data pipelines and transformations, you can easily navigate and process complex datasets, overcoming the inherent challenges associated with unstructured data formats. For additional details, refer to the following links: Healthcare data solutions in Microsoft Fabric - Microsoft Cloud for Healthcare | Microsoft Learn Discover and build cohorts: Discover and build cohorts (preview) capability in healthcare data solutions enables healthcare organizations to efficiently analyze and query healthcare data from multiple sources and formats. It simplifies the preparation of data for health trend studies, clinical trials, quality assessments, historical research, and AI development. It supports natural language queries for multimodal data exploration and cohort building, making it ideal for research and AI-driven projects. For additional details, refer to the following link: Overview of discover and build cohorts (preview) - Microsoft Cloud for Healthcare | Microsoft Learn The proposal for research data platform architecture builds upon the following foundational premises: Recognition of Fabric as the all-in-one data storage, processing, management and analytics platform with enterprise-level features around security, availability and self-service. Adoption of Fabric Workspace(s) as the security boundary (a secure logical container) for maintaining data platform items (data storage and processing assets). Fabric workspaces may be provisioned for and used by different research data platform stakeholders (groups of users) with different requirements around use cases, data privacy, data sensitivity and access security. Use of healthcare data solutions in Fabric, as the core capability to maintain healthcare data assets in a standard (interoperable) manner. Healthcare data solutions enables the storage and processing of several healthcare data modalities and formats that follow industry standards (for example, clinical modality in FHIR® NDJSON format and Clinical-Imaging modality’s DICOM® format). Industry standards make it easier for research data platform stakeholders to share (exchange) data and insights within their own organization as well as (when needed) with other organizations that they collaborate with. Use of Fabric native capabilities to address requirements that may not (yet) have been implemented for healthcare specific needs. This provides the research data platform stakeholders with the flexibility to develop various data storage and processing workloads easily in a low (or no) code manner. Fig – Conceptual architecture of research data platform in Microsoft Fabric Note: This diagram is an architectural pattern and does not constitute one to one mapping of existing Microsoft products. Organizing source data in data workspace (One Data Hub in the above diagram) Organize your enterprise data into a data workspace that could be leveraged for research purposes. This acts as a ‘One Data Hub’ for the research data platform. Multiple Lakehouse can be present in this workspace. There should be at least one Lakehouse that organizes data using ‘unified folder structure’ best practice. Convert data from non-supported format to healthcare data solutions supported format to leverage out of the box transformation for multimodal data: For healthcare data solutions supported modalities: Implement custom transformations to convert data to supported modalities/format. For unsupported modalities: Implement extensions to bronze Lakehouse to accommodate additional data modalities. Epic data availability: Epic supports FHIR data export using Bulk FHIR APIs. If your dataset meets the use cases of Epic Bulk Data, you can store the resulting FHIR resources into One Data Hub for further transformation. Avoid data content duplication: Data duplication cannot be totally avoided. However, the same file and same content are never duplicated. There will be situations when data needs to be transformed to suit the needs of existing transformation pipelines for accelerating research data platform development. Additionally, OneLake in Fabric storage, where Lakehouse is maintained, uses file compression. Healthcare data solutions in Fabric has functionality to compress raw files to zip and always writes structured data to delta parquet which is a higher compressed format. More information can be found here - Data architecture and management in healthcare data solutions - Microsoft Cloud for Healthcare | Microsoft Learn Curating data for research (One Analytics workspace in the above diagram) Implement and extend Silver Lakehouse: A flattened FHIR® data model is provided by healthcare data solutions out of the box within the Silver Lakehouse. Extending the existing data model is possible through adding new columns to existing tables or through adding new tables in the Silver Lakehouse. If there is a need to introduce a different data model altogether, it is best to implement it using a different Lakehouse. Implement and extend Gold Lakehouse: Deploy and extend Observational Medical Outcomes Partnership Common Data Model (OMOP CDM): Deploy OMOP CDM 5.4 out of the box with healthcare data solutions deployment. Extend OMOP CDM to accommodate additional modalities. For example, implement Gene sequencing, Variant occurrence and Variant annotation tables to add genomics modality into OMOP CDM or implement medical imaging data on OMOP CDM as described here - Development of Medical Imaging Data Standardization for Imaging-Based Observational Research: OMOP Common Data Model Extension - PubMed Implement custom Gold Lakehouse(s): Implement other custom Gold Lakehouse using Fabric tools that run your transformation logic from Silver to Gold. These Lakehouse cannot be connected to discover and build cohorts capability within healthcare data solutions. Customers that need access to custom data can connect their custom cohort browsers to the SQL Analytics Endpoint(s) of their custom Gold Lakehouse(s). Enable data de-identification: Microsoft provides several solutions that can be used to implement a comprehensive de-identification solution that customers expect. Refer to the articles below for details. Dynamic data masking in Fabric Data Warehouse - Microsoft Fabric | Microsoft Learn Row-level security in Fabric data warehousing - Microsoft Fabric | Microsoft Learn Column-level security in Fabric data warehousing - Microsoft Fabric | Microsoft Learn Announcing a de-identification service for Health and Life Sciences | Microsoft Community Hub Cohort discovery using cohort builder tool Microsoft’s cohort browser: Today Discovery and Build Cohort supports eyes-on cohort discovery. This is an out of the box solution that is part of healthcare data solutions in Fabric. When eyes off discovery is supported, researchers as well as research IT can benefit from both eyes off and eyes on discovery and cohort building. 3rd-party cohort browser (e.g., OHDSI Atlas): Most 3rd party cohort browsers (E.g. OHDSI Atlas) and home-grown cohort browsers typically support connection to a SQL endpoint. Microsoft Fabric platform provides the capability of exposing SQL endpoint from a Lakehouse that can be connected to a 3rd party cohort browser to perform cohort discovery. Automated data delivery Creating research workspaces with cohort needed for research: Create separate workspaces for different research projects to keep Fabric items distinct and project specific using Fabric APIs. Assign workspaces to a Fabric capacity: Note: When needed, and if the organization has more than one Fabric capacity provisioned, workspace assignment can be spread across different capacities to help manage cost and performance. Next, set up a Lakehouse and provide access for team members (as per IRB approval list). This ensures both access and security at the workspace level. Export data to research workspace (format desired by researchers): Currently, DBC exports data as CSV/JSON files stored in a Lakehouse within the same workspace. Shortcut the destination Lakehouse into research workspace to keep the sanity of cohort data. Tools for researchers: Fabric provides several data engineering and data science tools out of the box that researchers can leverage to perform research. The following are some of the documents that customers can use to enable researchers with the tools of choice. Data science in Microsoft Fabric - Microsoft Fabric | Microsoft Learn Create, configure, and use an environment in Fabric - Microsoft Fabric | Microsoft Learn Migrate libraries and properties to a default environment - Microsoft Fabric | Microsoft Learn Charge back: Fabric compute pricing depends on the chosen Fabric capacity SKU. Assigning different Fabric capacities to different projects or groups within the same cost center can facilitate chargeback. See the step mentioned above on assigning a workspace to a Fabric capacity during workspace creation. Manage historic data migration to the research data platform on Fabric In most instances, customers already possess a research data platform. They seek to transition to this proposed solution without disrupting their current research data flow and obligations. Follow this approach to migrate or use data from the existing platform to the new one: Use your current research data platform as a Lakehouse or a Data Warehouse in Fabric (take advantage of Shortcut and Mirroring features available in Fabric). Fabric offers cross-database query, i.e. allowing to query and join multiple Lakehouse and data warehouses in a single query. Customers can choose how and where to implement such queries to augment the healthcare data solutions datasets with their existing datasets, all natively in Fabric. A bridge/mapping layer can be built to link the old and the new in a cross-relational way. Conceptually, such an approach has also ties to Bring Your Own Database (BYO-DB) requirement, which is the ability to bring custom defined format and still be able to easily convert to healthcare data solutions specific format. Other workflow integration Integrate research data platform with IRB workflow: IRB workflows are dependent on the tools utilized. For instance, eIRB solution from Huron. While there is currently no direct integration between IRB workflows and the research data platform on Fabric, it is possible to develop a connector using Power Platform integration with Fabric. Specific details are not available at this time as this remains an exploratory initiative. Another approach will be to use Fabric REST APIs (as a pro-code method) that can enable richer integration between Fabric and the 3 rd -party system, and a better consuming user experience at the end. Capture logs necessary for “accounting of disclosures”: Logs in Fabric can be captured at event level. It’s up to the customer to decide the level and type of logs that need to be captured for accounting of disclosure. This will need some custom implementation. One such capability of Fabric that can be used is: Track user activities in Microsoft Fabric - Microsoft Fabric | Microsoft Learn FHIR® is a registered trademark of Health Level Seven International, registered in the U.S. Trademark Office and is used with their permission. DICOM® is the registered trademark of the National Electrical Manufacturers Association (NEMA) for its Standards publications relating to digital communications of medical information. If you are a Microsoft customer needing further information, support, or guidance related to the content in this blog, we recommend you reach out to your Microsoft account team in order to set up a discussion with the authors.Webinar: Azure OpenAI & a preview of Microsoft 365 Copilot, specifically for HLS Organizations!
Please join us for an introduction to Azure OpenAI and a preview of Microsoft 365 Copilot, specifically for healthcare and life sciences organizations. These technologies have the potential to revolutionize productivity and remove some of the drudgery of work.Tracking Azure History with Azure Resource Graph
When administrating an Azure environment, or any environment really, one will most likely find a way to track changes that were introduced. There are a number of ways to do this. Within Azure can query the Subscription or Resource Group Deployment, the downside though is this approach is limited to just the scope you are querying on. What if this is a larger organization with multiple subscriptions? You could also rely on a well-established CI/CD pipeline, a third-party governance tool, or in this case query Azure directly via the Resource Graph Explorer. For this blog will focus on using the Azure Portal offering of the tool; however, want to note that since this is API driven there are numerous offerings such as Azure PowerShell, Azure CLI, .NET, even Ruby.
