Hello, 

We would like to run additional analysis on quarantined files as part of a custom workflow. Is there a way to programmatically access quarantined files without restoring them from the quarantine. We’d like to leave the files in the quarantine, but we want to copy the files into another location within our organization outside of Defender for deeper malware analysis. Ideally, we’d like to use an MS Defender API for this post-quarantine action vs. using the MpCmdRun.exe util as we don’t want to restore the file.

Thanks!

Laurel

​Hello,  We would like to run additional analysis on quarantined files as part of a custom workflow. Is there a way to programmatically access quarantined files without restoring them from the quarantine. We’d like to leave the files in the quarantine, but we want to copy the files into another location within our organization outside of Defender for deeper malware analysis. Ideally, we’d like to use an MS Defender API for this post-quarantine action vs. using the MpCmdRun.exe util as we don’t want to restore the file. Thanks!Laurel  Read More

​

When using Outlook on mac in either Chrome or Safari, my folders are delayed in showing up when I click on “Move To”.  After several seconds or clicks the folders finally show up.  Anyone else experiencing this?

​When using Outlook on mac in either Chrome or Safari, my folders are delayed in showing up when I click on “Move To”.  After several seconds or clicks the folders finally show up.  Anyone else experiencing this?  Read More

​

Looking for some assistance, we have onboarded 2012 through to 2022 servers via MDE into Defender for Servers and the devices are all visible within the Device list on the security portal. Issue I’m have is that if we try and create a baseline assessment policy against any of the server groups, even with no filtering in place, it is only picking up a small handful of the devices.  I’m extracted all the devices for all the OS versions and ran various pivots to try and get a match on numbers to see if there was any common ground as to what was being detected. 

Any suggestions as I don’t want to just enable it without knowing what servers are being looked at.

​Looking for some assistance, we have onboarded 2012 through to 2022 servers via MDE into Defender for Servers and the devices are all visible within the Device list on the security portal. Issue I’m have is that if we try and create a baseline assessment policy against any of the server groups, even with no filtering in place, it is only picking up a small handful of the devices.  I’m extracted all the devices for all the OS versions and ran various pivots to try and get a match on numbers to see if there was any common ground as to what was being detected.  Any suggestions as I don’t want to just enable it without knowing what servers are being looked at.   Read More

​

Empowering Seamless Collaboration for Modern Workforces 

We are thrilled to announce the general availability of the Microsoft 365 Document Collaboration Partner Program (MDCPP). This new program allows eligible independent software vendors (ISVs) who provide cloud communication and collaboration platforms the opportunity to offer customers a collaboration experience inside and outside meetings. ISVs who join the program can enable customers on their platform to view, collaborate, and coauthor documents within Microsoft apps. We’re excited to expand this experience available in Microsoft Teams to new platforms.  

Enable Live Microsoft 365 Document Collaboration 

The Microsoft 365 Document Collaboration Partner Program allows eligible ISVs to include collaborative Microsoft 365 app experiences in their preferred communication and collaboration platform. And we’re bring the same quality customers expect from Microsoft for accessibility, security, and privacy. The Microsoft 365 Document Collaboration Partner Program includes: 

Web app option for real-time co-authoring 

Now eligible ISVs can provide simultaneous collaboration and editing of documents integrated into their platform. This includes the ability to integrate Word, Excel, and PowerPoint documents from SharePoint and OneDrive for Business.  Available today. 

An image of a Word document that multiple users can collaborate on

Live app option in meetings 

Soon, customers will be able to experience interactive and engaging experiences with PowerPoint Live and Excel Live in their preferred meeting solution. With PowerPoint Live and Excel Live, meeting participants can interact and collaborate with documents in real time, including editing and exploring documents directly in the meeting window. Presenters can also see rich presenter view, speaker notes, and other available presentation tools. Coming soon.  

An image of PowerPoint Live showing a slide with speaker notes in a Teams meeting.

Partnering for Success with Zoom 

We have collaborated with Zoom to leverage their expertise in communication and collaboration in the development of MDCPP.    

“We continue the work with Microsoft as announced earlier this year to bring this to market for Zoom Workplace customers,” said Brendan Ittelson, chief ecosystem officer, Zoom. 

Getting Started 

We invite all eligible ISVs to take advantage of the Microsoft 365 Document Collaboration Partner Program and transform their document collaboration process. To get started, visit our website for more information and resources. 

Join the MDCPP Today 

The Microsoft 365 Document Collaboration Partner Program is available to all eligible Microsoft partners and customers. Join us in this exciting journey and experience the future of document collaboration. 

We are excited for an easier, more collaborative future for our customers.  

​Microsoft Tech Community – Latest Blogs –Read More 

We’re excited and honored to be positioned in the Leaders Category in the IDC MarketScape: Worldwide SIEM (security information and event management) for Enterprise 2024 Vendor Assessment (doc #US51541324, September 2024)—our third major analyst report in SIEM to name Microsoft as a Leader. We were recognized in the most recent reports as a Leader in the 2024 Gartner® Magic Quadrant™ for Security Information and Event Management and as a Leader in The Forrester Wave™: Security Analytics Platforms, Q4 2022. We believe this position validates our vision and continued investments in Microsoft Sentinel, making it a best-in-class, cloud-native SIEM solution. It’s always a rewarding experience when trusted analysts recognize the continued work we’ve put into helping our customers modernize their operations, improve their security posture, and work more efficiently. 

A Leader in the market with an innovative solution for the SOC

Microsoft Sentinel provides a unique experience for customers to help them act faster and stay safer while managing the scaling costs of security. Customers choose our SIEM in order to:  

Protect everything with a comprehensive SIEM solution. Microsoft Sentinel is a cloud-native solution that supports detection, investigation, and response across multi-cloud and multi-platform data sources with 340+ out-of-the-box connectors A strength of Microsoft’s offering is its breadth, which includes user entity and behavior analytics (UEBA), threat intelligence and security orchestration, automation, and response (SOAR) capabilities, along with native integrations into Microsoft Defender threat protection products. 

Enhance security with a unified security operations platform. Customers get the best protection when pairing Microsoft Sentinel with Defender XDR in Microsoft’s unified security operations platform. The integration not only brings the two products together into one experience but combines functionalities across each to maximize efficiency and security. One example is the unified correlation engine which delivers 50% faster alerting between first- and third-party data, custom detections and threat intelligence.3 Customers can stay safer with a unified approach, with capabilities like automatic attack disruption—which contains attacks in progress, limiting their impact at machine speed.   

Address any scenario. As the first cloud-native SIEM, Microsoft Sentinel helps customers observe threats across their digital estate with the flexibility required for today’s challenges. Our content hub offerings include over 200 Microsoft- created solutions and over 280 community contributions. The ability to adapt to the unique use cases of an organization is something called out in both the Forrester and Gartner reports.  

Scale your security coverage with cloud flexibility. Compared with legacy, on-premises SIEM solutions, Microsoft Sentinel customers see up to a 234% return on investment (ROI).1 This makes it an attractive option for customers looking for a scalable offering to meet the evolving needs of their business while managing the costs of data. We’ve recently launched a new, low-cost data tier called Auxiliary Logs to help customers increase the visibility of their digital environment, while keeping their budgets in check. In addition, Microsoft’s SOC Optimizations feature, a first of its kind offering, provides targeted recommendations to users on how to better leverage their security data to manage costs and maximize their protection, based on their specific environment and using frameworks like the MITRE attack map  

Respond quickly to emergent threats with AI. Security Copilot is a GenAI tool that can help analysts increase the speed of their response, uplevel their skills, and improve the quality of their work. 92% of analysts reported using Copilot helped make them more productive and 93% reported an improvement in the quality of their work.

What’s next in Microsoft Security 

Microsoft is dedicated to continued leadership in security through ongoing investment to provide customers with the intelligence, automation, and scalability they need to protect their businesses and work efficiently. New and upcoming enhancements include more unified features across SIEM and XDR, exposure management and cloud security in the unified security operations platform, and our SIEM migration tool—which now supports conversion of Splunk detections to Microsoft Sentinel analytics rules and additional Copilot skills to help analysts do their job better.  

To learn more about Microsoft Security solutions, visit our website. Bookmark the Security blog to keep up with our expert coverage on security matters. Also, follow us on LinkedIn (Microsoft Security) and X (@MSFTSecurity) for the latest news and updates on cybersecurity. 

​Microsoft Tech Community – Latest Blogs –Read More 

实际的环境中，IoT设备可以生成大量的数据。为了减少上传的数据量或者降低控制策略的生产延迟，有时必须在设备端对数据进行实时分析或处理。Azure Stream Analytics服务就是很好的解决方案方案一，用户可以从Azure Portal中创建Azure Stream Analytics服务，然后在Azure IoTHub中将其设置为IoT Edge Module并部署到Azure IoT Edge设备上。本文将演示如何创建Azure Stream Analytics作业，并将其部署到IoT Edge设备上。

1.创建存储账户

      首先，在Azure门户中，转到“新建”，在搜索框中输入“存储”，选择“存储帐户 – Blob，文件、表、队列”。

      然后，在“创建存储帐户”中，输入存储帐户的名称，选择存储IoTHub的相同位置（这里为东亚），然后选择“创建”。请记下该名称以供稍后使用。

      接下来，转到刚才创建的存储账户，选择“Blob Service”。为Azure Stream Analytics模块创建一个用于存储数据的新容器，将访问级别设置为“Container”，选择“确定”。

2.创建Azure流分析作业

      首先，在Azure门户中，转到“创建”>“物联网”，然后选择“流分析作业”。

      然后，在“新建流分析作业”中执行以下操作：在“作业名称”框中输入作业名称；在“托管环境”下，选择“Edge”；在剩余字段中使用默认值。

      然后，在所创建作业中的“作业拓扑”下，依次选择“输入”-“添加”。在“输入别名”框，输入温度。在“源类型”框，选择“数据流”。剩余字段中使用默认值。 

       接下来，在所创建作业中的“作业拓扑”下，依次选择“输出”-“添加”，在“输出别名”框中，键入警报，在剩余字段中使用默认值。之后选择“创建”。 

      最后，在所创建作业中的“作业拓扑”下，依次选择“查询”-“添加”，加入以下SQL语句并保存

从温度时间戳中选择 
     “重置”作为命令
进入
    警报    ，按时间创建分组，按 TumblingWindow（秒，30）分组，按 Avg（machine.temperature）> 70

3. 部署Stream Analytics Job

      首先，在Azure Portal的IoTHub页面内，转到“IoT Edge”并打开IoT Edge设备的详细信息页面。

      “设置模块”，并确保已经按照文章中的步骤之前添加了tempSensor模块，因为这里的Azure Stream Analytics模块是针对选择tempSensor模块产生的数据来进行实时分析的。

      在“添加模块”页面，选择“导入 Azure Stream Analytics IoT Edge 模块”，

      在接下来的Edge Deployment页面，选择之前创建好的Stream Analytics – Edge Job，注意，这里要选择之前第一部分已经创建好的存储账户和容器，点击保存，如下图所示。

      之后，将下面的代码复制到路由，将{ moduleName }替换为复制的模块名称：

{
     “routes”：{                                                              
       “telemetryToCloud”：“从/messages/modules/tempSensor/* 到 $upstream”，
       “alertsToCloud”：“从/messages/modules/{moduleName}/* 到 $upstream”，
       “alertsToReset”：“从/messages/modules/{moduleName}/* 到 BrokeredEndpoint(“/modules/tempSensor/inputs/control”)”，
       “telemetryToAsa”：“从/messages/modules/tempSensor/* 到 BrokeredEndpoint(“/modules/{moduleName}/inputs/temperature”)”
     }
}

      选择下一步，然后提交。看到返回到“设备详细信息”页面，并选择“刷新”。应会新的流分析模块已经在列表中，但状态还是一个待部署。

      稍后，等该模块部署到设备以后，刷新列表，可以发现，EdgeASA已经处于运行状态。

      回到Putty工具，利用“docker log -f {moduleName}”指令（其中，{moduleName}用刚刚部署的流分析模块的名称代替），就可以查看流分析的日志信息。

​实际的环境中，IoT设备可以生成大量的数据。为了减少上传的数据量或者降低控制策略的生产延迟，有时必须在设备端对数据进行实时分析或处理。Azure Stream Analytics服务就是很好的解决方案方案一，用户可以从Azure Portal中创建Azure Stream Analytics服务，然后在Azure IoTHub中将其设置为IoT Edge Module并部署到Azure IoT Edge设备上。本文将演示如何创建Azure Stream Analytics作业，并将其部署到IoT Edge设备上。1.创建存储账户      首先，在Azure门户中，转到“新建”，在搜索框中输入“存储”，选择“存储帐户 – Blob，文件、表、队列”。        然后，在“创建存储帐户”中，输入存储帐户的名称，选择存储IoTHub的相同位置（这里为东亚），然后选择“创建”。请记下该名称以供稍后使用。        接下来，转到刚才创建的存储账户，选择“Blob Service”。为Azure Stream Analytics模块创建一个用于存储数据的新容器，将访问级别设置为“Container”，选择“确定”。  2.创建Azure流分析作业      首先，在Azure门户中，转到“创建”>“物联网”，然后选择“流分析作业”。        然后，在“新建流分析作业”中执行以下操作：在“作业名称”框中输入作业名称；在“托管环境”下，选择“Edge”；在剩余字段中使用默认值。      然后，在所创建作业中的“作业拓扑”下，依次选择“输入”-“添加”。在“输入别名”框，输入温度。在“源类型”框，选择“数据流”。剩余字段中使用默认值。          接下来，在所创建作业中的“作业拓扑”下，依次选择“输出”-“添加”，在“输出别名”框中，键入警报，在剩余字段中使用默认值。之后选择“创建”。         最后，在所创建作业中的“作业拓扑”下，依次选择“查询”-“添加”，加入以下SQL语句并保存从温度时间戳中选择      “重置”作为命令进入    警报    ，按时间创建分组，按 TumblingWindow（秒，30）分组，按 Avg（machine.temperature）> 703. 部署Stream Analytics Job      首先，在Azure Portal的IoTHub页面内，转到“IoT Edge”并打开IoT Edge设备的详细信息页面。        “设置模块”，并确保已经按照文章中的步骤之前添加了tempSensor模块，因为这里的Azure Stream Analytics模块是针对选择tempSensor模块产生的数据来进行实时分析的。        在“添加模块”页面，选择“导入 Azure Stream Analytics IoT Edge 模块”，        在接下来的Edge Deployment页面，选择之前创建好的Stream Analytics – Edge Job，注意，这里要选择之前第一部分已经创建好的存储账户和容器，点击保存，如下图所示。        之后，将下面的代码复制到路由，将{ moduleName }替换为复制的模块名称：{      “routes”：{                                                                      “telemetryToCloud”：“从/messages/modules/tempSensor/* 到 $upstream”，       “alertsToCloud”：“从/messages/modules/{moduleName}/* 到 $upstream”，       “alertsToReset”：“从/messages/modules/{moduleName}/* 到 BrokeredEndpoint(“/modules/tempSensor/inputs/control”)”，       “telemetryToAsa”：“从/messages/modules/tempSensor/* 到 BrokeredEndpoint(“/modules/{moduleName}/inputs/temperature”)”      } }      选择下一步，然后提交。看到返回到“设备详细信息”页面，并选择“刷新”。应会新的流分析模块已经在列表中，但状态还是一个待部署。        稍后，等该模块部署到设备以后，刷新列表，可以发现，EdgeASA已经处于运行状态。        回到Putty工具，利用“docker log -f {moduleName}”指令（其中，{moduleName}用刚刚部署的流分析模块的名称代替），就可以查看流分析的日志信息。  Read More

​

Four years post pandemic, organizations continue to refine their work arrangement policies—balancing employee feedback, industry trends, and leadership preferences across in-person, hybrid, and remote models. As they navigate this evolving landscape, one thing is clear: there’s no one-size-fits-all solution. To help our customers better navigate the world of hybrid work, Viva People Science dove into our data pool of insights from global leaders and people managers to better understand whether work arrangement policies impacted leaders’ perceptions of their organization. The Viva People Science team’s High Performing Organizations survey was used for this analysis, which studied 1,100 global leaders and people managers across 10+ industries.

For this month’s research drop, we aimed to gain a deeper understanding of how leaders viewed their ability to meet performance indicators, their evaluation of their organization’s effectiveness, and their perceived top barriers to high performance.

We grouped leader perceptions based on how they described their organization’s work arrangement policy. There were four categories:

Fully flexible – these organizations require no days onsite (21.5% of the sample)
Partially onsite – these organizations require 1-2 days onsite per week (31.3% of the sample)
Predominantly onsite – these organizations require 3-4 days onsite per week (40.7% of the sample)
Fully onsite – these organizations require 5 days onsite per week (6.9% of the sample)

It’s important to note that this categorization allows us to compare leader perceptions from organizations with various work arrangement policies. It doesn’t provide context on how employees at these organizations manage their time and location based on the policies or how co-located the teams are at these organizations.

Our data revealed an interesting trend! Leaders’ perceptions of their organization didn’t shift in a straightforward or linear way as the level of flexibility in work arrangement policies increased or decreased. In fact, we found that leaders from organizations with fully flexible or predominantly onsite reported higher organizational performance and effectiveness than leaders from organizations with partially onsite or fully onsite policies. This trend persisted throughout our analysis – suggesting that it’s not a battle of “remote vs. in-person” but that there is a level of nuance to work arrangement policies.

Top performance ratings come from leaders within fully flexible and predominantly onsite organizations

When exploring leaders’ perceptions of their organization’s performance, leaders at fully flexible and predominantly onsite organizations reported achievement of performance indicators more consistently than leaders at partially onsite and fully onsite organizations.

This pattern of leader perceptions indicates that we can’t assume that flexibility and weekly in-person time are the only variables at play in the impact of work arrangement policy on overall organizational performance. For example, more in-person time may be thought to connect to higher levels of collaboration1. However, while we do see that leaders at predominantly onsite organizations rate their collaboration the highest, this isn’t the case for all work arrangement policies with an in-person element. When crafting and implementing a work arrangement policy, consider more than just the number of days onsite.

Finding the right balance of work routines may influence perceptions of work practice effectiveness

Taking a deeper look into fully flexible and predominantly onsite organizations, we saw that these organizations may offer a sense of stability and balance in work routines and environments, driving higher leader perceptions. When asked to evaluate their organization on a set of organizational effectiveness factors, the pattern of stronger feedback from leaders at fully flexible and predominantly onsite organizations reigned consistent among leaders.

We’ve seen in previous research that being intentional with how and where you plan certain in-person work activities can facilitate effective work and productivity by increasing high-quality connections between employees2. But when we look at the above organizational effectiveness indicators, leaders from partially onsite and fully onsite organizations reported lower levels of vision, collaboration, innovation, and efficiency.

So, what is causing lower perceptions of these imperative work conditions? When considering how work arrangement policies may play a part, it may be that leaders who are onsite 1-2 days feel their work week is fragmented and leaders who are required to be onsite five days feel overwhelmed by in-person expectations. We’ve also seen in research that even as work arrangement policies evolve to include more in-person time, not all work routines are evolving accordingly. In fact, regardless of work arrangement policy, the volume of virtual meetings has not reduced3. This suggests that being in-person doesn’t reduce time spent collaborating in a virtual work environment, highlighting a need for effective collaboration habits regardless of where the work happens, such as effective meetings, intentional communications, and productive asynchronous work.

Unique strengths of fully flexible and predominantly onsite policies

Fully flexible and predominantly onsite policies both relate to higher leader perceptions and evaluations, but these policies each have certain areas where they stand out.

Perceived manager effectiveness and alignment are higher in fully flexible environments.

Leaders at fully flexible organizations rate their organizations higher on manager effectiveness and goal alignment, suggesting that those with fully flexible policies have created an environment that deepens the impact that leaders can have for their teams, such as providing great clarity and alignment in the flow of work. Research shows that virtual work can reduce the impact of traditional power dynamics, facilitating an environment where employees are more confident in bringing up sensitive topics and breaking down departmental silos4.

The perceived impact of “face time” is not consistent for in-person environments.

Between work arrangement policies with an in-person element, leaders at predominantly onsite organizations report the highest levels of confidence that their organization’s employees know what is expected of them and how to grow within the organization. It’s interesting that this benefit of being in-person doesn’t extend to partially onsite or fully onsite leaders’ perceptions, suggesting that there may be a bell curve of in-person impact.

Leader ratings suggest that being in-person only 1-2 days may not provide enough opportunities to network with cross-functional teams and identify performance goals for internal mobility, but that being fully onsite may introduce location-specific limitations on internal mobility, such as location restrictive job openings. Consider the moments that matter most to being in-person when crafting your work arrangement policy and the impact that you want those moments to have, such as generating team cohesion, supporting employees through onboarding, or facilitating effective project kick-offs2.

Overburdened managers and inefficient cultures stand in the way of high performance, regardless of work arrangement policy

Leaders also reported the barriers they feel slow their organization’s progress toward high performance. All four categories of work arrangement policies shared a common barrier in their top three: overburdened managers.

Leaders with different work arrangement policies reported the same barriers to high performance.

Inefficient work culture ranked as the top barrier for all flexible policies, suggesting that getting the processes and culture right is still a work in progress.
For predominantly onsite organizations, accumulation of tedious tasks and inefficient organizational practices ranked second. It may be that consistently being onsite comes with the greater expectation for leaders to perform routine and tedious tasks while they are in-office.
For fully onsite organizations, inefficient tools and resources ranked third. It may be that being fully onsite limits leaders’ ability to perform necessary focus work and leverage their time most effectively.

This research advances our understanding of how work arrangement policies shape leaders’ views on organizational performance and effectiveness. A key takeaway is that these policies drive noticeable differences in some areas of work but have little effect in others. This suggests that regardless of work arrangement policy, leaders should focus on getting foundational priorities right: 1) support overburden managers, 2) set up collaboration norms that are location-agnostic, and 3) invest in the moments that matter. hat matter.

While these takeaways reflect leaders’ perspectives, the employee viewpoint is also needed to shape effective work policies. Recent research shows that US employees expect their organizations to allow them to work from home around 2.3 days per week in the coming year, with employees on average wanting to be able to work up to 3 days per week from home5. Connecting leader and employee perceptions are essential to understanding the holistic experience of your organization and critical when strategizing work arrangement policies to facilitate high-quality employee experiences and expectations.

Stay tuned for our November Research Drop to keep up with what the Viva People Science team is learning!

1 – Gallup, How Important Is Time in the Office? March 2, 2023.

2 – Microsoft WorkLab, In the Changing Role of the Office, It’s All about Moments That Matter. 2024.

3 – Harvard Business Review, Hybrid Work Has Changed Meetings Forever. June 17, 2024.

4 – Deloitte, Inclusive or isolated? New DEI considerations when working from anywhere. May 25, 2023.

5 – Barrero, Jose Maria, Nicholas Bloom, & Steven J. Davis, 2021. “Why working from home will stick,” National Bureau of Economic Research Working Paper 28731.

​Microsoft Tech Community – Latest Blogs –Read More 

The healthcare industry is undergoing a transformation, driven by the power of artificial intelligence (AI). Last week at the HLTH conference, Microsoft announced advanced healthcare AI models in Azure AI Studio.

Developed in collaboration with Microsoft Research, our strategic partners, and leading healthcare institutions, these AI models are specifically designed for healthcare organizations to rapidly build and deploy AI solutions tailored to their specific needs, all while minimizing the extensive compute and data requirements typically associated with building multimodal models from scratch. With the healthcare AI models, healthcare professionals have the tools they need to harness the full potential of AI to assist patient care.

Modern medicine encompasses various data modalities, including medical imaging, genomics, clinical records, and other structured and unstructured data sources. Understanding the intricacies of this multimodal environment, Azure AI onboards specialized healthcare AI models that go beyond traditional text-based applications, providing robust solutions tailored to healthcare’s unique challenges.

Introducing the New Healthcare AI Models in the Azure AI Model Catalog:

The Azure AI model catalog has a new industry “Health and Life Sciences” filter with a new array of state-of-the-art, open source, healthcare models. This includes Microsoft’s first-party models as well as models by strategic partners:

MedImageInsight (paper) This embedding model enables sophisticated image analysis, including classification and similarity search in medical imaging. Researchers can use the model embeddings and build adapters for their specific tasks, streamlining workflows in radiology, pathology, ophthalmology, dermatology and other modalities. For example, researchers can explore how the model can be used to build tools to automatically route imaging scans to specialists, or flag potential abnormalities for further review, enabling improved efficiencies and patient outcomes. These models must be thoroughly tested, validated, and, in some cases, further fine-tuned to ensure their applicability in specific use cases. Furthermore, the model can be leveraged for responsible AI such as out-of-distribution (OOD) detection and drift monitoring, to maintain stability and reliability of AI tools and data pipelines in dynamic medical imaging environments.

Example Juypter notebooks deployable in Azure Machine Learning:
https://aka.ms/healthcare-ai-examples-mi2-deploy 
https://aka.ms/healthcare-ai-examples-mi2-zero-shot 
https://aka.ms/healthcare-ai-examples-mi2-adapter 
https://aka.ms/healthcare-ai-examples-mi2-exam-parameter

MedImageParse (paper) Designed for precise image segmentation, this model covers various imaging modalities, including X-Rays, CT scans, MRIs, ultrasounds, dermatology images, and pathology slides. It can be fine-tuned for specific applications such as tumor segmentation or organ delineation, allowing developers to test and validate the ability to leverage AI for highly targeted cancer and other disease detection, diagnostics and treatment planning.

Example Juypter notebooks deployable in Azure Machine Learning:

https://aka.ms/healthcare-ai-examples-mip-deploy 
https://aka.ms/healthcare-ai-examples-mip-examples

CXRReportGen (paper) Chest X-rays are the most common radiology procedure globally. They’re crucial because they help doctors diagnose a wide range of conditions—from lung infections to heart problems. These images are often the first step in detecting health issues that affect millions of people. By incorporating current and prior images, along with key patient information, this multimodal AI model generates report findings from chest X-rays, highlighting AI-generated findings directly on the images to align with human-in-the-loop workflows. Researchers can test this capability and the potential to accelerate turnaround times while enhancing the diagnostic precision of radiologists. This model has demonstrated exceptional performance on the industry standard MIMIC-CXR benchmark (paper):

Example Juypter notebooks deployable in Azure Machine Learning:
https://aka.ms/healthcare-ai-examples-cxr-deploy

Partner models: Paige.ai, Providence Healthcare, NVIDIA, and M42 contributed foundational models to the catalog, spanning areas including pathology, 3D medical imaging, biomedical research, and medical knowledge sharing. Developed under a core set of shared AI principles, these models provide a powerful starting point for organizations as they launch their own AI projects, while embedding responsible practices across the industry.

The open access to AI models on the catalog and modular approach allows healthcare organizations to customize solutions, maintain control over their data, and build trust through shared development and oversight. This approach aligns with our commitment to responsible AI, ensuring our technologies meet ethical standards and earn the trust of the medical community.
Microsoft is dedicated to responsibly scaling artificial intelligence and continuously improving our tools by listening and learning. Importantly, Microsoft does not use customer data to train AI models without explicit customer permission or in undisclosed ways. We collaborate with organizations to help them harness their data, enabling the development of predictive and analytical solutions that may drive their competitive advantage.

Azure AI Studio: Empowering Health and Life Sciences with Seamless AI Integration

Azure AI Studio offers healthcare professionals a comprehensive platform to develop, fine-tune, deploy, and continuously monitor AI models tailored to their unique needs. With the introduction of the new healthcare AI models, Azure AI Studio simplifies the integration of AI into healthcare workflows which allows professionals to focus on improving patient outcomes. Here’s how Azure AI Studio delivers value:

Bring your data and fine-tune models: Azure AI Studio and healthcare AI models complement the healthcare data solutions available in Microsoft Fabric, creating a unified environment to bring multimodal proprietary data to enable a wide range of use cases. Healthcare professionals can leverage the models as-is using the playground in Azure AI Studio or fine-tune pre-trained models with their data in Azure Machine Learning to adapt models for their specific clinical needs.
Rapid Development and Deployment: Azure AI Studio provides an intuitive interface and a comprehensive set of generative AI operations (GenAIOps) toolchains that enable professionals to quickly develop, test, and deploy AI applications. This streamlined process can accelerate the adoption of AI in healthcare, empowering organizations to integrate sophisticated diagnostic and analytical tools into their existing workflows. With built-in support for deploying models in cloud, on-premises, or hybrid environments, healthcare professionals can optimize their AI solutions for various clinical settings.
Supporting Safety and Compliance: Trust is crucial in healthcare, where AI can impact patient care. The model cards in the model catalog share details about the training and evaluation datasets used, including fairness testing where applicable. The platform supports hybrid deployment options for enhanced control over sensitive healthcare data. Additionally, Azure AI aligns with healthcare regulations such as HIPAA, helping organizations maintain high standards of data security, patient confidentiality, and overall compliance.

Real-world Impact: Customer Success Stories

Healthcare organizations are already leveraging these models to transform their workflows with Azure AI:

Mass General Brigham: MGB is using Microsoft’s MedImageInsight model to surface additional relevant information during clinical research and streamlining radiologist workflows, alleviating the administrative workload on clinical staff, and enhancing the speed of patient care.
University of Wisconsin-Madison: UW is targeting advanced report generation from medical imaging analysis using Microsoft’s CXRReportGen. With ever-increasing imaging volumes colliding with the ongoing combination of radiologist burnout and shortages, a state-of-the-art medical imaging model can be used to build an application that can transform a medical image into a draft note, supporting better outcomes for patients while helping clinicians focus on the hands-on components of their roles.
Sectra: Sectra is working with Microsoft to build on top of foundational models like MedImageInsight to automate the process of understanding the types of exams coming through the Sectra Vendor Neutral Archive (VNA) system for better routing and display.
Mars PETCARE: Mars PETCARE is exploring the use of the healthcare AI models for veterinary medicine applications, such as data evaluation in radiology and pathology teams. By combining veterinary expertise with advanced AI models, Mars PETCARE is setting new standards in animal health. This collaboration has the potential to transform veterinary diagnostics, improve the quality of care for pets, and showcase the versatility of healthcare AI models in non-human medical contexts.
Paige: In life sciences, Paige is working to combine radiology, pathology, and genomic insights for a more comprehensive approach to disease diagnosis, aimed at accelerating the discovery of new treatments.

Join Us in Shaping the Future of Healthcare

Join us at Microsoft Ignite to witness these models in action and learn how they can transform your healthcare practice. Visit the documentation and AI Studio to explore these cutting-edge healthcare AI models and start your journey toward a data-driven, AI-empowered future.

Medical device disclaimer: Microsoft products and services (1) are not designed, intended or made available as a medical device, and (2) are not designed or intended to be a substitute for professional medical advice, diagnosis, treatment, or judgment and should not be used to replace or as a substitute for professional medical advice, diagnosis, treatment, or judgment. Customers/partners are responsible for ensuring solutions comply with applicable laws and regulations.

Generative AI does not always provide accurate or complete information. AI outputs do not reflect the opinions of Microsoft. Customers/partners will need to thoroughly test and evaluate whether an AI tool is fit for the intended use and identify and mitigate any risks to end users associated with its use. Customers/partners should thoroughly review the product documentation for each tool.

​Microsoft Tech Community – Latest Blogs –Read More 

Welcome to our latest blog post in the AI Empowerment series! In a recent webinar, we explored the world of AI in Human Resources. The webinar was packed with insights, practical examples, and a whole lot of fun! Here’s a recap of the key points discussed and some practical tips on how you can start using AI in your everyday HR tasks.

The current state of AI in HR

We began the conversation discussing where HR as a profession is with AI adoption today. Whilst we are seeing general optimism and acknowledgement that organizations need to keep up with AI, there are also concerns around the use of responsible AI, manager enablement and data privacy. Here are some key findings from a recent study by Gartner (2024):

Optimism and Adoption: 61% of HR professionals are optimistic about AI’s potential to enhance HR practices, particularly in HR analytics, employment law, and learning and development. However, only 38% of HR leaders are currently piloting or implementing AI solutions today.

Common use cases today: Recruitment, chatbots and virtual assistants, and reducing administrative load are the top areas where AI is making an impact.

Challenges and Concerns

Data privacy and security: Ensuring robust data handling practices and protecting sensitive information are paramount.
Professional caution: HR practitioners need to navigate the complexities of using new technology while keeping up with evolving laws and maintaining a human touch.

Using AI in everyday HR Tasks using Microsoft 365 Copilot

Using the Microsoft Copilot HR Scenario Library as inspiration, we explored a few use cases to demonstrate where and how Copilot could be leveraged in various HR tasks.

Conducting Market Research for Recruitment

Example Copilot prompt: “Generate a table of key skills and experience required for software engineers at large enterprise technology companies based on market trends and leading technology companies.”

Creating Job Descriptions

Example Copilot prompt: “Draft a job description for a software engineer role, including key responsibilities, required skills, and company culture.”

Updating Policy Documents

Example Copilot prompt: “Scan the latest version of our remote work policy and create an FAQ document for employees.”

Writing effective prompts for your daily HR tasks

During the webinar, we spoke about the importance of experimenting with prompt writing and reiteration to help create prompts that would produce the desired outcome. We wrote some prompts together and reviewed the subsequent Copilot generated output! Here are a few examples that stood out:

1. Writing pre-survey communication to employees

– Example Copilot prompt:  “Help me draft an email to our employees about the upcoming global employee survey. The email should include information about when the survey is open and the closing deadline, reasons employees should participate and how they can provide their feedback.”

2. Encouraging employee participation in the survey

– Example Copilot prompt: “We are in the middle of our global employee engagement survey period and have only received 20 responses across the company. Help me write a reminder email to employees encouraging them to participate and the importance of sharing their feedback. Use a tone that is encouraging yet friendly as employees are very busy at this time of year.”​

3. Helping leaders re-draft internal communications about a Return to Office policy using appropriate style and tone

– Example Copilot prompt: “Please provide a new version of the below email draft. I’d like the tone to be professional, but ensure the language used is empathetic and compassionate.”

Looking to the future

The potential for AI to revolutionize human resources is boundless. Imagine a world where HR professionals can focus more on strategic initiatives and employee engagement, while AI handles the repetitive and time-consuming tasks. The integration of AI into HR processes not only enhances efficiency but also opens up new avenues for innovation and growth in the profession. Here are some further resources to help you navigate your HR AI journey:

Microsoft Adoption: Using Copilot in Human Resources (includes downloadable HR scenario kit, example prompts and much more)

Viva Community Call (Oct 2024): How Microsoft HR is using Viva and M365 Copilot to Empower Employees

Microsoft Copilot Academy now available to all Microsoft 365 Copilot users

Empowering responsible AI practices

Copilot FAQs

If you missed our live webinar, you also can watch the recording and access the presentation below.

​Microsoft Tech Community – Latest Blogs –Read More 

The Phi model collection represents the latest advancement in Microsoft’s series of Small Language Models (SLMs). Back in August 2024, we welcomed the latest additions, Phi-3.5-mini and Phi-3.5-MoE, a Mixture-of-Experts (MoE) model:

Phi-3.5-mini: This 3.8B parameter model enhances multi-lingual support, reasoning capability, and offers an extended context length of 128K tokens.
Phi-3.5-MoE: Featuring 16 experts and 6.6B active parameters, this model delivers high performance, reduced latency, multi-lingual support, and robust safety measures, surpassing the capabilities of larger models while maintaining the efficacy of the Phi models.

From Generalist to Custom SLMs

The results from our benchmarks underscore the remarkable efficiency and capability of Phi-3.5-mini and Phi-3.5-MOE. But even so, the models can be further customized for your unique needs to match the performance of larger models for a given task.

There are three powerful techniques that can be used to customize language models for your organization’s specific needs:

Prompt Engineering
Retrieval Augmented Generation (RAG)
Fine-tuning

Let’s delve into each of these techniques.

Prompt Engineering is about providing clear instructions directly within the prompt, often in the system prompt, to guide the model’s responses. This method falls under the category of “giving more information in the prompt” and can be particularly useful for shaping the model’s behavior and output format.

Next, we have Retrieval Augmented Generation, or RAG. This technique is employed when you want to incorporate organizational data and knowledge into the model’s responses. RAG allows you to provide the model with reliable sources for answers through additional documents. It retrieves the relevant information and augments the prompt, enhancing the model’s ability to generate informed and contextually accurate responses. RAG also belongs to the category of “giving more information in the prompt”.

Fine-tuning is the process of customizing a model using labelled training data, which often leads to better performance and reduced computational resources. In fact, fine-tuning a smaller model with the appropriate training data can lead to its performance exceeding a larger model for a specific task. Specifically, Low-Rank Adaptation (LoRA) fine-tuning is an excellent approach for adapting language models to specific use cases due to several key advantages. First, LoRA significantly reduces the number of trainable parameters, making the fine-tuning process more efficient, saving time and cost. This reduced demand on resource allows for quicker iterations, making LoRA easier to experiment with fine-tuning tasks. LoRA keeps the original model weights mostly unchanged, which helps in maintaining the general capabilities of the pre-trained model while adapting it to specific tasks.

LoRA fine-tuning for Phi-3.5 models

Today, we are proud to announce the availability for LoRA fine-tuning the Phi-3.5-mini and Phi-3.5-MoE models in Azure AI starting November 1, 2024.

Serverless fine-tuning for Phi-3.5-mini and Phi-3.5-MOE models enables developers to quickly and easily customize the models for cloud scenarios without having to manage compute. The fine-tuning experience is available in Azure AI Studio, and it adopts a pay-as-you-go approach, ensuring you only pay for the actual training time your fine-tuning requires.

Once finetuned, the models can be deployed in Azure for inference, with the option to enable Content Safety. Deploying your fine-tuned model is a streamlined process with our pay-as-you-go service. The billing for fine-tuned model deployments is based on the number of input and output tokens used, along with a nominal hosting charge for maintaining the fine-tuned model. Once deployed, you can integrate your fine-tuned model with leading LLM tools like prompt flow, LangChain, and Semantic Kernel, enhancing your AI capabilities effortlessly.

Fine-tuning with the Managed Compute option is also available. You may use the Azure Machine Learning Studio user interface or follow our notebook example to create your custom model. Using the notebook has the advantage of greater flexibility for the configurations used by the fine-tuning job. You have the option to download the fine-tuned model, and then deploy it using Azure managed compute resources, your own premises, or your edge devices. The fine-tuned model is also licensed under MIT license.

Closing remark

The Phi-3.5 family of models represents a significant advancement in the realm of SLMs. With the introduction of Phi-3.5-mini and Phi-3.5-MoE, we have pushed the boundaries of performance, efficiency, and customization. The availability of LoRA fine-tuning in Azure AI further empowers developers to tailor these models to their specific needs, ensuring optimal performance for a wide range of applications. As we continue to innovate and refine our models, we remain committed to providing cutting-edge solutions that drive progress and enhance user experiences. Thank you for joining us on this journey, and we look forward to seeing the incredible things you will achieve with the Phi-3.5 models.

​Microsoft Tech Community – Latest Blogs –Read More