Microsoft Shares Healthcare AI, Cloud Updates

In a feature release for Microsoft Cloud for Healthcare published this week, the company shared updates for its Azure AI Services for Health offerings aimed at social determinants of health (SDOH) analytics, responsible artificial intelligence (AI) implementation, and clinical trial matching.

According to the release, these features will be previewed at the HIMSS 2023 conference as part of Microsoft’s ongoing efforts to drive technological innovation in the healthcare space.

The first update adds support for SDOH and ethnicity data within Azure’s Text Analytics for health. This addition enables users to extract insights from social, environmental, and demographics factors found in unstructured biomedical data.

This update also includes assertion detection, according to a recent Microsoft blog post. Assertion detection, such as negation of substance use by a patient, can then be captured using a combination SDOH data, substance use, and substance use amount found in unstructured clinical notes.

The second update discusses a preview of Project Health Insights, announced this week as a service to help leverage health data. The service generates insights using patient data and includes pre-built AI models to generate specific insights based on confidence scores and evidence from the input data.

The preview at HIMSS23 will demonstrate how two Project Health Insights models can be used for cancer research and care: Oncology Phenotype and Clinical Trial Matcher.

Oncology Phenotype allows clinicians to identify cancer attributes within a patient population with existing cancer diagnoses. These attributes include tumor site, histology, and clinical stage, in addition to tumor, nodes, and metastasis (TNM) categories and pathologic stage TNM categories.

Clinical Trial Matcher helps match patients to potentially suitable clinical trials based on patient data and the trial’s eligibility criteria. The tool can also find a cohort of potentially eligible patients for a list of clinical trials.

The third update previews a new Azure Health Bot template, which allows users to integrate Azure OpenAI Service into their Health Bot. According to a blog post published by Microsoft earlier this week, Azure Health Bot’s capabilities include providing patient triage and healthcare related information from clinically validated sources to users.

However, in some scenarios, the bot does not understand what the end user is trying to ask. In these cases, the integration of OpenAI is designed to help provide answers to user queries.

The feature release notes that this feature “does not aim to facilitate the bot to answer unknown queries in the medical space, rather, it enables customers to access the Azure OpenAI Service API and decide how to use the model to improve their bot built through the Azure Health Bot service.”

The preview of this feature is currently being offered for internal testing and evaluation purposes only.

Microsoft also announced the general availability of its accelerator kit for healthcare, part of the Responsible AI Dashboard in Azure Machine Learning.

The accelerator kit is designed to assist users with training and debugging AI models to address fairness, explainability, and biases prior to deployment in healthcare settings.