Less than a year in, health plans and health care providers continue to work urgently to mitigate risk and improve outcomes in the COVID-19 pandemic. As health organizations seek better solutions, more of them are turning to artificial intelligence. While skepticism may persist, this global crisis is an opportunity to demonstrate the value of AI.
In late 2014, Alignment Healthcare began to develop an AI-first data and technology platform, AVA™ (Alignment Virtual Application), to improve patient experience and outcomes. While many health plans typically rely on outdated claims information and siloed data systems, AVA incorporates near realtime patient admission, discharge, and transfer data from more than 1,500 hospitals nationwide, as well as social determinant data, memberreported information, biometrics, medical claims, clinical notes and data from pharmacies and labs.
Each disparate data source is ingested, cleansed and transformed into our unified data architecture, providing timely, accurate and consistent data across different lines of business. The data is used to power more than 100 AI models, ranging from hospital admission prediction models to various disease risk scores and chronicity levels. AVA’s machine-learning algorithms usually ingest more than 800 data points for each patient, providing explanations to the decisions it makes and giving us a rich set of information for identifying vulnerable populations.
In January 2020, our clinical team recognized the potential problem of COVID-19 – especially for our elderly and chronically ill patient population – and quickly began to respond with changes and new developments to keep our members safe and healthy by leveraging timely, relevant and actionable data and AI. We created a COVID-19 dashboard, providing near real-time COVID-19 insights including confirmed cases, admission risks, outreach efforts, inbound call activities and potentially COVID-19-related hospitalization events, all in one place to enable quick and data-driven decisions.
For example, as we round to check in on our hospitalized members, we have found that many COVID-19 cases are occurring among spouses. By using the admission tracker in our census, which is near real-time within AVA, Alignment’s clinical team can identify a COVID-19-positive patient and immediately reach out to that patient’s spouse to offer guidance. We can ensure that the spouse is staying home, and we can monitor their symptoms and provide them with 24/7 telehealth access to an on-call physician.
By leveraging all these data points, we also built risk stratification models for proactive and personalized member outreach. AI allows us to look at millions of data points in seconds, when it would take years for a human to analyze the same data. This enables us to manage large populations and identify vulnerable groups or individuals for timely interventions. Using AVA, we have been able to quickly identify our health plan members who are diagnosed with, or at high risk of contracting, COVID-19. That analysis helped us reach out rapidly and educate our at-risk populations on ways to mitigate their risk, such as staying home and following proper hand hygiene. At the same time, we have been able to monitor the patients who test positive and provide their caregivers and families with best practices for minimizing and reducing the risk of viral transmission.
By combining AI with telehealth, we also give our members a more robust health care experience. During a telehealth visit, an Alignment physician uses AVA’s capabilities to access comprehensive patient history since they’ve been with us including previous hospitalization events, various risk scores, historic diagnoses, pharmacy, lab, and procedure data. This provides a better picture of the person’s entire health, so the clinician can ask the right questions to provide higher quality care or prevention, even if it’s the patient’s first encounter with him or her. Because Alignment recognized the benefits of telehealth and AI early on, we have been able to scale up our offerings, rather than scramble to implement them.
We reached 16,000 members in a matter of weeks by doing a COVID-19 questionnaire via personto-person telehealth visits. This allows us to learn which members have special concerns, such as the inability to get their medications or food. Alignment responded to these needs with a crisis meal delivery program that provides two weeks of meals to members who cannot otherwise get access to groceries.
The COVID-19 outbreak has revealed the importance of preparing for a public health crisis, as well as some critical gaps in today’s health care technology infrastructure. Many health care information technology servers do not provide the quick data needed to respond to a major public health issue. Datasets are heavily siloed, preventing health plans and providers from having actionable data on different populations. The use of AI to collect, analyze, and share data could help surmount some of these barriers.
Traditional health plans may be able to learn who is at risk and act on some claims data. Without a method to make the data actionable, however, the health plan is simply watching for likely risk, with no way of mitigating it. One of the big changes that AI can bring to health care is the utilization of this data in the right manner at the right time. This outbreak has shown that health providers and plans need not only the right information, but a delivery arm to act on it and mitigate risks and improve outcomes for their members and patients. This is what has made Alignment so successful in reaching our population.
Although many healthcare stakeholders have been slow to adopt AI, the COVID-19 outbreak has created a new opportunity to demonstrate its potential. Being able to identify the right people and the right types of interventions at the appropriate times will be one of AI’s biggest selling points going forward.
Dr. Arta Bakshandeh, DO, MA, is chief medical information officer at Alignment Healthcare, a technology-enabled Medicare Advantage company