How Health AI Became Key in COVID-19 Response
In response to the COVID-19 pandemic, our team will be interviewing experts from across the ecosystem to bring the HLTH community timely facts and updates.
HLTH Team: Tell us a bit about Microsoft Research and Health Next?
Hadas Bitran: Healthcare NExT began as an incubation initiative within Microsoft Research, experimenting with different technologies and approaches to help meet our healthcare customers’ needs. As we’ve progressed, we’ve drawn on the wealth of expertise found across Microsoft, not just within our research organization. By combining our reputation in research with our expertise in health and life sciences, we’re delivering innovation that empowers digital transformation at scale and we’re convinced that healthcare may be artificial intelligence’s most urgent application.
HLTH Team: Where does your team fit in? What problems/issues is your team trying to solve and why?
Hadas Bitran: I lead the Microsoft Healthcare group in the Microsoft Israel R&D Center. We are focused on health artificial intelligence (AI), and our team leads the development of healthcare language services: conversational AI and natural language processing (NLP) technologies used in the healthcare and biomedical areas. Worldwide, our healthcare customers continue to be overburdened with information and challenged by inefficiencies. The healthcare industry is also overloaded with unstructured text. Understanding this text and putting it into structure is key to unlocking insights, and our team is working to help ease these burdens.
HLTH Team: We have heard about the Microsoft Healthcare Bot utilization during COVID-19. How did this bot come to be?
Hadas Bitran: During a Microsoft company hackathon, we were looking to create a multi-turn conversational technology that would empower patients with self-service assessment and symptom checking. The Microsoft Healthcare Bot service was born.
HLTH Team: How was the Microsoft Healthcare Bot service being used before COVID-19?
Hadas Bitran: Before COVID-19, we had about 4,000 customers using the Microsoft Healthcare Bot service to power their virtual health assistants. They were using it for symptom checking, finding locations and business hours, completing questionnaires, telehealth, and delivering patients with answers at scale to various health-related questions.
HLTH Team: What about after the virus hit?
Hadas Bitran: When the virus hit, we saw the overwhelming impact to frontline workers around the world. We worked with the CDC to create a CDC COVID-19 symptom checking protocol for screening people and checking symptoms, to help reduce the burden on healthcare workers. We worked 20-hour days, through nights and weekends, to ensure the service fulfilled these needs. We created templates in the product to enable quicker development of COVID-19 response bots. The first COVID-19 bot was created in just four days. Then this timeline became two days, and finally there was a customer that created a bot in just one day using our technology. We now have Microsoft Healthcare Bot instances worldwide within the government, healthcare, and retail industries in 25 countries.
HLTH Team: Besides symptom checking, how else is the Microsoft Healthcare Bot service being used?
Hadas Bitran: The U.S. Veterans Affairs Department recently deployed their bot built in under a month via the Microsoft Healthcare Bot service. This new, conversational tool marks a first for the agency and can be accessed around the clock to triage symptoms related to COVID-19, and offer targeted responses regarding testing options, stimulus payments, telehealth, scheduling, prescription refills and more. Together with alliance partners, we have also launched the CoVIg-19 Plasma Bot, a self-screening tool that anyone can use to see if they qualify to donate their plasma. Beyond this, the Microsoft Healthcare Bot Scenario Template Catalog includes templates for many other bot scenarios, for example - returning to the workplace screener, mental health questionnaires, appointment booking, filing claims, provider lookups, and more.
HLTH Team: What other exciting technologies can we look forward to from your group?
Hadas Bitran: We just launched as part of Azure Cognitive Services Text Analytics for health, now available in public preview. With the influx of healthcare data generating over a zettabyte of healthcare data every year, it is increasingly critical for providers to unlock access to the information in free text to drive greater efficiency and overall better healthcare outcomes. Using NLP techniques, the health feature of Text Analytics enables developers to process and extract insights and valuable information from unstructured clinical documents. Trained on a diverse range of medical data —covering various formats of clinical notes, clinical trials protocols, and more—the health feature is capable of processing a broad range of data types and tasks, without the need for time-intensive, manual development of custom models to extract insights from medical data.
HLTH Team: Do you have examples of Text Analytics for health in action?
Hadas Bitran: In response to the COVID-19 pandemic, Microsoft partnered with the Allen Institute of AI and leading research groups to prepare the COVID-19 Open Research Dataset, a free resource of more than 47,000 scholarly articles for use by the global research community. Using Text Analytics for health and Cognitive Search, we developed a COVID-19 search engine to enable researchers to generate new insights in support of the fight against the disease. Text Analytics for health is also at work powering our clinical trials matching technology. You can access an example implementation of it, that matches patients to COVID-19 trials, on Bing.com. People can use this to match with a COVID-19 clinical trial so they can help in the search for possible solutions to the virus. Visit Bing.com, scroll to the right, and you’ll find an icon for COVID-19 resources with options for “clinical trials” or “take the CDC self-checker.”
HLTH Team: Expand a bit more on your work on using this technology with clinical trials.
At any moment in time, there are more than 50,000 clinical trials around the world looking to recruit patients for their studies. For patients, it is challenging to find clinical trials that are a match. At the same time, researchers are struggling to recruit suitable participants for their clinical trials. Patient recruitment is the main cause of delays in trials, and 50% of the trials fail to reach their recruitment target. The Clinical Trials Matching technology uses state-of-the-art NLP technology that can understand both clinical trial protocols and patient health information– and match accordingly. In the case of COVID-19, for example, it helps narrow down and prioritize the set of potential clinical trials to a more relevant set of trials for which the patient appears to be qualified. Take a look at Bing.com for an example of this in action.
Stay up-to-date and learn more about Microsoft:
About Hadas Bitran:
Hadas Bitran is Head of the Healthcare Group in the Microsoft Israel R&D Center, as part of Microsoft Healthcare NExT in Microsoft Artificial Intelligence and Research.
Hadas and her group are focused on building AI technologies and solutions for the healthcare space, including language services, natural language processing, conversational intelligence and personal health assistants technologies and services.
During her career in Microsoft, Hadas has been developing products in the domains of Healthcare, Personal Assistants, Recommendation Systems, Online Advertising and Consumer Privacy. Her work has been shipped as part of Cortana, Bing and the Windows Operating System.
Before Microsoft, Hadas held senior leadership positions at SAP. Prior to that, she managed R&D and product teams in start-up companies.
Hadas received her B.Sc. degree in Computer Science from Tel Aviv University, and her MBA from Kellogg School of Management, Northwestern University in Chicago.