How digital technology is helping the fight against COVID-19

In August, Providence’s Digital Innovation Group (DIG) leaders hosted a LinkedIn Live event to talk about how the recent accelerated advances in digital healthcare have been supporting the challenges brought about by the COVID pandemic. The broadcast drew in several thoughtful audience questions that were later answered in detail. These are shared below.

To watch the full discussion in the live event, please visit:

Q. Does [the chatbot] Grace have its own Language Model? Is there a dynamic model update based on user’s input?

A. For the original version of Grace that we developed a couple of years ago, we built our own language models and leveraged several open source NLP libraries. We used a mix of dynamic and static models. We used the static models for managing “alarm phrases.” For these phrases, we manually review and update the models working closely with our clinicians and risk team. This is to ensure the safety of our patients. 

For COVID-19 assessment and FAQ, we partnered with Microsoft to leverage their HealthBot and Azure Cognitive Services including the NLP capability. Our team focused on building the knowledge models and algorithms on top of those services.

Prior to COVID, Grace has helped our consumers to navigate their way to the best care options based on their symptoms, their preference and care availability for lower-acuity conditions. Grace has also answered frequently-asked-questions based on her learnings from our customer service calls. Our vision is to continue to develop a smart health assistant who helps our consumers in every step of their journey, and simplifies caregiving for our providers, while increasing our operational efficiency and care quality.

As part of this vision, Grace needs to scale and solve many hard problems that require AI technologies such as NLP and computer vision. To accelerate and scale innovation and provide the best-of-breed solutions to our consumers, our technology strategy is a hybrid approach that embraces open, flexible architecture. Based on the scenario and the needs, we may build the technology in house, use open source, or leverage services from a great partner such as Microsoft and other technology companies. Our technology choices will always support our mission and do what’s best for our users.

Q. Is there any project on AR or VR to be added to Grace, so users can experience more than text? (adding another layer of interaction)

A. Not currently. Our team has thought about how Grace could potentially leverage AR/VR to help our consumers. There are definitely some opportunities to explore, and without a doubt, these technologies could empower smart assistants to deliver more interactive and better solutions to our users. We prioritize our choice of technology solutions based on many factors, including what our users need most, and how quickly the technology can scale and become available for all of our consumers and across all the scenarios we’re focusing on. While Grace currently doesn’t leverage AR/VR, we continue to consider these technologies when assessing the best solution for the problems we’re solving.

Q. What’s your opinion on TeleDoc and Livingo merger? Are payors more likely to use these off-the-shelf solutions?

A. Payers and providers want to address the needs of their consumers the best way possible. The solutions they select are going to depend on many factors including the specific business problems they are solving, the nature of their clinical operations, and also what their users need for the specific scenarios. At Providence, we keep an open mind when it comes to picking solutions and partners. That means we continuously evaluate what meets the needs of our patients, providers and business best.

The biggest issue is that these payers will create “islands of care” not connected into the health system making it hard to refer patients and transfer medical records. We are working with our provider sponsored health plan, Providence Health Plan, and our payor partners to resolve this via integration. Stay tuned!

Q. How is the burden around care coordination being addressed during times of reduced communication and visitation of outside post-acute providers?

A. We have accelerated digital transformation to maintain or in some cases, increase the care quality for our patients. The investments we have made in the past few years were key in enabling us to quickly address our patients’ needs and even help other health systems, as well as our community. Here are a few examples:

• In partnership with Microsoft, we enhanced Grace, our existing smart health assistant, to ask a series of questions and help navigate patients to the right modality of care.
• We scaled virtual visits using DexCare solution, which is an integrated care platform across different care modalities, including virtual, retail and at-home visits along with an intelligent routing engine for prioritization and load balancing of care demand and supply. We are now deploying that solution with other health systems.
• We leveraged our portfolio solutions such as Xealth (digital prescribing tool) and Twistle for connecting providers and patients, care automation and remote monitoring.
• Grace has a self-assessment tool which enables consumers to assess their exposure and symptoms, and make recommendations based on clinical guidelines.

Q. Sounds like there will be a gap on detecting the virus infections on people with no symptoms. How will Grace or caregivers be able to detect the infection?

A. The assessment bot is not a replacement for getting tested. It’s developed based on the latest CDC guidelines to help route the patients to providers who can further assess and recommend testing. Also, symptoms are not the only factors used in the assessment. 

Q. Does Arpan Wagray’s team’s work tie into the work you all are doing?  He's been a big part of tele-behavioral health initiatives.

A. DIG builds technologies that can be leveraged by various teams at Providence or other health systems. We work closely with Arpan and team in supporting the relevant care scenarios. For instance, we’re working with Arpan’s team to scale CBT across our ambulatory care settings.

Q. Do other health care providers use chatbots like this?

A. Health systems have started to explore the usage of chatbots for a few scenarios. We believe both chat and voice will have a huge impact in the future of healthcare. Our goal is to help accelerate innovation for the relevant use cases and make the learnings and technology solutions available to other health systems and overall our community. For example, we developed our COVID-19 chatbot in March in partnership with Microsoft and then we made our solution and learnings available to many others.

Q. Are there other areas in healthcare that you think a chatbot would be helpful?

A. When we think about chatbots, we consider two key elements: 1) chat as a conversational user interface; and 2) the intelligence behind the bot and its ability and skills to process and complete tasks. That’s why we use the term “smart health assistant.”

There are many scenarios for which conversational user interface (chat or voice) can be helpful. If you think about the consumer health journey, the chatbot assistant can help with many use cases including concierge services, answering questions, navigation to the right care modality, and completing tasks such as appointment booking and prescription refills. Chat/voice bots can also be helpful in simplifying caregiving for providers, nurses, medical assistants and customer service.

While the potential scenarios are abundant in healthcare, we’re still in the early stages of design and development of these chatbots. We’re learning about the users’ (consumers or providers) behavior and the best design practices. We’re improving NLP and other relevant technologies to meet specific healthcare needs. We’re partnering closely with our clinical, operations, legal, risk and IT teams to ensure the best way these solutions can get integrated with other technologies and our clinical workflows.

No matter what the scenario is, we need to always prioritize what’s best for our patients and community, and innovate with that goal in mind.