Duration 18:54
16+
Play
Video

Driving business transformation in healthcare using Google Cloud and AI/ML

Joe Corkery
Director of Product at Google
+ 1 speaker
  • Video
  • Table of contents
  • Video
Google Cloud Next 2020
July 14, 2020, Online, San Francisco, CA, USA
Google Cloud Next 2020
Request Q&A
Video
Driving business transformation in healthcare using Google Cloud and AI/ML
Available
In cart
Free
Free
Free
Free
Free
Free
Free
Free
Add to favorites
1.55 K
I like 0
I dislike 0
Available
In cart
Free
Free
Free
Free
Free
Free
Free
Free
  • Description
  • Transcript
  • Discussion

About speakers

Joe Corkery
Director of Product at Google
Thomas Tsai
Executive Director of Global Strategy at DUNU

Experienced product leader with a strong focus on healthcare & life sciences technology. Key areas of interest include security, compliance, privacy, clinical informatics, interoperability, medical imaging, digital health, and biomedical analytics. Graduate of Harvard Medical School and Princeton University.

View the profile

About the talk

This session covers how Google Cloud partners closely with healthcare organizations to drive business transformation. Watch highlights of customer stories as well as Google Cloud’s differentiated products and solutions for healthcare now.

Speakers: Joe Corkery, Thomas C. Tsai

Watch more:

Google Cloud Next ’20: OnAir → https://goo.gle/next2020

Subscribe to the GCP Channel → https://goo.gle/GCP

#GoogleCloudNext

AI118

product: Cloud AI & Industry Solutions; fullname: Joe Corkery;

event: Google Cloud Next 2020; re_ty: Publish;

Share

I'm joke or cream and I leave the house trained Lifesciences product team in Google Cloud. When we first started planning for next week for this, talk to me about how Healthcare organizations are transforming their businesses. Using a, i n m l, Ever since that time the world, especially as it relates to healthcare. I seen a completely different type of transformation as a result of covid-19. With that, in mind, we decided to focus. This talk more about how Google responds to covid-19 as well as some of the works at

our partners are doing. I want to start by talking through Google, Cloud Healthcare and Life Sciences team's Mission which is to create solutions to empower provider are in Pharma organizations to transform their business in the healthcare. Are no men efforts to deliver on that mission are primarily organized around four main pillars organized. We want to make it easy to ingest normalize and join data to unlock the potential of your data accessible. We want to support Open, Standards, apis interoperability and Discovery to ensure that the data can be made accessible for use

secure. We leave the industry, and security privacy and compliance for endpoint a Datacenter. Everything we do here is core to our mission at the heart of our work. What's secure accessible? Well, organized data, we aim to help organizations generate insights and new capabilities in advance to quadruple. Aim of healthcare Healthcare is a framework of rules designed to improve the patient screens of care of populations producer per capita cost of healthcare and improve. The experience of commission staff useful Killers, especially critical today during

covid-19 as organizations take action on a wide range of data available to them. Your overall agenda for the session we divided into three distinct Parts. In the first session, we will cover the application of AI to do before casting and will be joined by dr. Thomas eye surgeon and health policy researcher in the department of surgery at Brigham and Women's Hospital and the Department of Health policy and management. At the Harvard, th Chan School of Public Health is affiliated with the Harvard Global Health Institute with him. Google Cloud Kart 8, on the

development of the covid-19 Pella forecast in to how Google Cloud Healthcare specific product offerings are being used to address covid-19. Finally, in our last session will highlight work being done by one of our partners trading her and how it's being used to back covid-19, let's get started. Now, I'd like to welcome, dr. Thomas, I'd share his thoughts on our collaboration to deliver the covid-19 public forecast. Thank you, Joe for that introduction. As just said, my name is Thomas. I am a surgeon and assistant professor of Health policy and management. At the

Harvard School of Public Health and a faculty member of the Harvard Global Health Institute, the goal of the Harvard Global Health Institute has to provide actionable data to guide policy-making to improve health in both the United States and a global context during covid-19 or efforts are focused around three core areas, developing models of us Hospital capacity to help plan for storage capacity. To meet the demand of covid-19 cases, supporting federal and state policymakers and developing testing Targets in the contact tracing strategy, and 3 providing

scientific communication to combat misinformation. We are public health researchers, Health policy, experts, and clinicians weave. Bring all those perspectives to Bear when informing policy response, Georgia Health Care needs such as a covid-19 pandemic. Over the last few months, we have embarked on a national conversation in conjunction with the Harvard, Safra Center for ethics and multiple academic groups, think-tanks foundations, and policy groups. Converging Serenity, set of metrics for the suppression of covid-19. This has resulted in her covid-19 risk, level

dashboard and test Target dashboard globalepidemic.org. Our goal has been to not just show the underlying risk for Lee Tae. The community level risk. The corresponding set of Public Health actions around testing nonpharmacologic interventions such as social distancing and even shut down for the dashboard. Shows us where we are, but doesn't do is predict. What will happen in the future because of this States and even counties are unable to fully prepare for. What is to come. It's Hospital policy makers and public health officials. The predicted low reliability future

covid-19 outbreak so they could be better. Prepared. This is where artificial intelligence and machine learning come into play through a. I and machine learning. We can creep forecast models that are able to predict cases deaths and other important metrics over the next 14 days. I bought the u.s. state and county level and this is where Google Cloud comes in. Cloud. AI partnership. With the Harvard Global Health Institute has developed a novel covid-19, forecasting model as able to predict cases. Deaths, another important metrics of his hospitalization in the

14-day period at the US state and county level. These forecasts are powered by Machine, learning informed, epidemiological models, a continuously, learn from underline, public data, sets the forecast for available, free to the public. The dinner with the covid-19 public forecast. Google Cloud, researchers developed, a novel X Series machine. Learning approach that combines artificial intelligence with a robust epidemiological Foundation by Design. This new model exchange on public data and leverages an architecture allows researchers to dive into their different

relationship with the model, has learned the better interpret, why it made certain forecast won the challenges of a novel pandemic. Like covid-19 is a traditional epidemiological modeling, depends on assumptions of certain parameters such as a transmission rate, incubation period and the risk of mortality. But the AI machine learning approach enables is an opportunity to learn the data in real time. As you're progressing through this pandemic, we hope that these measures not only help the public understand how the model works. But more

importantly can enable further Innovations in the face. Disease modeling recording of Public Health officials to prioritize data quality and transparency as he underlined either wooden forms and enables. The AI models, make accurate predictions of healthcare professionals. Are these data are public with open end? Users will use these data to create their own dashboards, who died in The Forum local decision-making. There's a saying that all models are wrong, but some models are useful. The covid-19, Publix forecast model is uniquely positioned to provide, not just useful information

of more importantly actionable insights that can be used to help suppress the covid-19 pandemic, it's been a privilege to Harvard Global Health Institute to partner with Google Cloud. On this very important effort. We look forward to continuing to the ongoing collaboration Thank you, I really appreciate you participating in the session. The covid-19 public forecast model Jared's a 14-day forecast of the development of covid-19 in each US, state and County, many signals to help First Responders in the public sector and other impacts of organizations to be better prepared for what lies

ahead today is available. And we were used to interact with it directly and without cost as is included in the covid-19 virus that program that makes a large number of covid related data. Sets really searchable on bigquery Can also be downloaded directly as a CSV for your analysis and your own tool. We also created a high-level dashboard view of the data using Google data Studio, which is what you're seeing here. Let's take a closer look at that dashboard. Now, to get a better understanding of the results. In the covid-19 public forecast data Studio dashboard, you will see data

projections over the next 14 days for the United States. The forecasts are generated roughly every day in are available for all US states and the Sasquatches historical data to side by side with the forecast, Tuesday, which states are expected to have the highest increase in the number of deaths and confirmed cases. You can also quickly navigate between states, if you the forecast, for each one looks like Arizona. Additional metrics such as hospitalizations. I see you and dental. Your forecasts are also available for us, counties in historical, and forecast data

is available for cases, deaths, and hospitalizations You can use the drop-down menu to select one or more counties in due time, series of the original station for that County. Additional data at the county and state level is available in the big parade table ncsc. Addition to our partnership. With the Harvard Global Health Institute will work closely with Sada. And you see a health care to make the covid-19 pellets forecast available in the National Response, portal with the goal of ensuring, his many people can have access to the data as possible. Many

organizations have found value from these types of data, including ACA has our hope is that other organizations will also be able to take the stage and use it to plan. Accordingly to address covid-19 needs at the hospital level products and how they're being used to help with the covid-19 crisis. We just talked about how organizations like HCA Healthcare are benefiting from a i m l recovery forecasting is what powers he's a male models. However, the day is challenging times,

easier and faster to accelerate Healthcare data interoperability and enable you to unlock the power of your data to create and or consume a high models to run large-scale. Cross modality, analytics, the building deploy novel Health, Care applications Let's continue with the, team and walk through an example of how a large provider system was able to quickly gain insights into covered admissions. Using the healthcare Epi systems, the disease was moving faster than the day

that they were using to make their decision. Some of the challenges, they faced included data wait and see where their existing data analytics branoff epic Clarity in sports. We can offer me more than twenty-four hours old data continuity Regional Hospitals and Clinics each had their own satellite view of Just Around data. I'm finally available it skills, a real lack of local, IT staff made rollout of the new technical solution, impossible. What they needed was a way to tap into real-time events and in particular, those that are broadcast on the hl7, V2 message stream. The solution that

they developed Leverage, The real-time feed by consuming the hl7 me to message feed directly into the healthcare API transforming into Bakery for Analytics. The data feeds were then combined across Regional Hospitals and Clinics to create a single aggregated data set for the entire system. It was an able to use pre-build look for dashboards that can buy National Data with their local data as well as their own data to gain better insight into the specifics of the Google. Cloud. Healthcare API was the key to moving from some time to real-time analytics.

We're able to plug into the Rhapsody, interface engine and land, the hl7 messages directly Healthcare API message to or we can leverage our data, harmonization sweet to transform the hl7, be two messages into a fire representation, real time. Realtor. If you can automatically reject the data through to bigquery and can be visualized using a liquid at 4, traditionally standing up with this kind of a solution within the Healthcare System would take months, but we needed to act at the speed of covid. And by using the healthcare API, where you able to do this result,

was that the Seas visibility into real-time data about the number of cases, resources and utilization and all this was delivered from start to finish in just three weeks before we move on, I want to take a closer look at the architecture behind the solution. Solution was setting up the cloud environment, using the data protection tool kit, which automates the deployment of cloud projects and resources. The best practices automating permissions using the principle of least privilege and setting up log retention, admitted transferred or discharged HR or the electronic

health-record, sends out an hl7 message, which are Pub sub adapter listens for captures and delivers to healthcare a b. I h, l 72 message store, Rachel 72 data was converted to fire your transformation pipeline running cloud data Fusion. However that wasn't the only source of data brought into the solution. We also brought in data from operational systems and see if me for me, work the system to create a whistle mapping. Config to harmonize the data into fire. I need different sources which one different schemas terminologies in value since, as a result of operational and clinical data or

harmonized real-time to the fire store which sends dreams data directly into degree. So that the liquor dashboards reflect up-to-date information for decision-making, As we discussed on the previous slide is a lot of value in the structured data contained in the hl7 messages and the operational did. However there's a significant amount of value locked up. An unstructured data contained in the chemical Narrative of the progress, notes requires a different approach. But I'm a little bit deeper into clinical entity extraction. What you're seeing here is a small part of a typical. I see, you

know, it's very dense there, lots of unintuitive abbreviations is weird, punctuation, because large sections are often generated by the HR and their two languages here was English and not, but there is often more detail and richness of context buried in a clinical note, understand is not available in the structure thr data. So what can we do about this natural language ATI, which is coming soon in beta, will help you find a SAS in length, and in particular it allows you to extract clinically relevant entities from a text to understand the relationship between those entities to

understand the context in which those entities appear in the day is also associated with no knowledge. He's In addition to the capabilities offered by the healthcare natural language API for customer information, extraction models that have high fidelity for healthcare and Life. Sciences applications is significant extraction for healthcare which is also coming soon. Provide you with the ability to generate high quality custom models for Health Care applications, with no coding skills, require just uploaded label, the training data, train the model with the simple, click and then deploy that

model scale. major US Health Center was seeing patients presenting to them with a wide variety of different symptoms through multiple clinical entry points, including their call centers, chatbots, telemedicine, visits and of course, in person care The data has been collected at these different entry points was not connected together and it was not an easy way to pull all this data together to identify common patterns and their population and response that I decided to kick off a pilot to automate the extraction of clinical information. From these unstructured data sources to better, make

more accurate predictions in better triage to covid-19 patients. The early results of this infernal promising, they pulled all the disparate data sets together into a single store and also enhance their chatbots to ask covid-19 specific questions as well. As a result of their ability to better extract information from multiple sources. They seem improvements in the accuracy of their diagnosis, as well as a reduction in their time, to treat their also exploring, how this can be used in the future to better identify. Chronic conditions related to covid. The clothes, I want to highlight one

of our partners that's working on finding new treatments for covid-19 near Discovery company that has built a platform that uses physics and chemistry principles to create computational models. That predict how different substances will interact with molecular level simulations enormous scale to assess billions of molecules in silico before selecting candidates to take him to the lab. Renounce, our initial partnership in May when Google Cloud was providing the Schrodinger with access to powerful Computing,

capacity, to accelerate drug Discovery for its commercial partners and its own internal pipeline. Order more recently and in response to the covid-19 pandemic. Google cloud is providing Schrodinger with 16 million hours of GPU time through our researcher credit program. In this competition every for covid-19 which fuse consecutively would equate to 1826 years of round-the-clock computing. Accelerate, the drug Discovery, process Schrodinger's teamed up with Takeda. Novartis Gilead Sciences in wuxi apptec in a philanthropic initiative to share ideas, resources and data.

With the goal of developing novel, Therapeutics covid-19 We're very excited to be able to support this initiative. We wish them the very best of luck. We don't say you found this useful and informative and that this will be helpful for your organization as you can about the covid-19 crisis. We're now in the final months of next week of several Healthcare and Life Sciences sessions that you might find useful. You can usually find this information link from the Google Cloud Healthcare and Life Sciences website or from our next home page. Thank you very much.

Cackle comments for the website

Buy this talk

Access to the talk “Driving business transformation in healthcare using Google Cloud and AI/ML”
Available
In cart
Free
Free
Free
Free
Free
Free
Free
Free

Ticket

Get access to all videos “Google Cloud Next 2020”
Available
In cart
Free
Free
Free
Free
Free
Free
Free
Free
Ticket

Similar talks

Aashima Gupta
Director, Global Healthcare Solutions at Google Cloud
+ 1 speaker
John Halamka
President at Mayo Clinic Platform
+ 1 speaker
Available
In cart
Free
Free
Free
Free
Free
Free
Free
Free
Denise Winkler
Strategic Business Executive, Global Public Sector at Google
+ 1 speaker
Jennifer Ricker
Acting Secretary / CIO at Illinois Department of Innovation & Technology (DoIT)
+ 1 speaker
Available
In cart
Free
Free
Free
Free
Free
Free
Free
Free
Mike Daniels
VP, Global Public Sector, Google Cloud at Google
+ 1 speaker
Roberta Reardon
Commissioner at New York State Dept. of Labor
+ 1 speaker
Available
In cart
Free
Free
Free
Free
Free
Free
Free
Free

Buy this video

Video

Access to the talk “Driving business transformation in healthcare using Google Cloud and AI/ML”
Available
In cart
Free
Free
Free
Free
Free
Free
Free
Free

Conference Cast

With ConferenceCast.tv, you get access to our library of the world's best conference talks.

Conference Cast
635 conferences
26170 speakers
9693 hours of content
Joe Corkery
Thomas Tsai