While most of the public’s attention is focused on the horse race for an approved COVID-19 vaccine, another major hurdle lies just around the corner: the distribution of hundreds of millions of vaccine doses. In today’s highly complex and disconnected health data landscape, technologies like AI, Machine Learning, and robotic process automation (RPA) will be essential to making sure that the highest-risk patients receive the vaccine first.
Why identifying at-risk patients is incredibly difficult
Once a vaccine is approved, it will take months or years to produce and distribute enough doses for the U.S.’ 330 million residents. Hospital systems, primary care physicians (PCPs), and provider networks will inevitably need to prioritize administration to at-risk patients, potentially focusing on those with underlying conditions and comorbidities. That will require an unimaginable amount of work by healthcare employees to identify patient cohorts, understand each patient’s individual priority level, and communicate pre- and post-visit instructions. The volume of coordination required between healthcare systems and the pressing need to get the vaccine to high risks groups makes the situation uniquely different than other nationally distributed vaccinations, like the flu.
One key challenge is that there’s no existing infrastructure to facilitate this process – all of the data necessary to do so is locked away in disparate information silos. Many states have legacy information systems or rely on fax for information sharing, which will substantially hamper efforts to identify at-risk patients. Consider, in contrast, the data available in the U.S. regarding earthquake risk– you can simply open up a federal geological map and see whether you’re in a seismic hazard zone. All the information is in one place and can be sorted through quickly, but that’s just not the case with our healthcare system due to its fragmentation as well as HIPAA and patient privacy laws.
There are several multidimensional barriers that make it nearly impossible for healthcare workers employed by providers and state healthcare organizations to compile patient cohorts manually:
– Providers will need to follow CDC guidelines on prioritization factors, which based on current guidelines for those with increased risk could potentially include specific conditions, ethnicities, age groups, pregnancy, geographies, living situations (such as multigenerational homes), and disabilities. Identifying patients with these factors will require intelligent analysis of patient profiles from existing electronic health record data (EHR) used by a multitude of providers.
– Some hospital networks use multiple EHR and care management systems that have a limited ability to share and correlate data. These information silos will prevent providers from viewing all information about patient population health data.
– Data on out-of-network care that could require prioritization, like an emergency room visit, is often locked away in payer data systems and is difficult to access by hospital systems and PCPs. That means payer data systems must be analyzed as well to effectively prioritize patients.
– All information must be shared and analyzed in accordance with HIPAA laws, and the mountain of scheduling communications and pre- and post-visit guidance shared with patients must also follow federal guidelines.
– Patients with certain conditions, like heart disease, may need additional procedures or tests (such as a blood pressure reading) before the vaccine can be administered safely. Guidelines for each patient must be identified and clearly communicated to their care team.
– Providers may not have the capacity to distribute vaccines to all of their priority patients, so providers will need to coordinate care and potentially send patients to third-party sites like Walgreens, Costco, etc.
All of these factors create a situation in which it’s extremely difficult – and time-consuming – for healthcare workers to roll out the vaccine to at-risk patients at scale. If the entire process to analyze, identify, and administer the vaccine takes only two hours per patient in the U.S., that’s 660 million hours of healthcare workers’ time. A combination of analytics, AI, and machine learning could be a solution that’s leveraged by healthcare workers and chief medical officers in identifying the priority of patients supplemented with CDC norms.
How RPA can automate administration to high-risk patients
Technology is uniquely poised to enable health workers to get vaccines into the hands of those who need them most far faster than would be possible using humans alone. Robotic process automation (RPA) in the form of artificial intelligence-powered digital health workers can substantially reduce the time spent prioritizing and communicating with at-risk patients. These digital health workers can intelligently analyze patient records and send communications 24 hours a day, reducing the time needed per patient from hours to minutes.
Consider, a hypothetical situation in which the CDC prioritizes certain risk profiles, which would put patients with diabetes among those likely to receive the vaccine first. In this scenario, RPA offers significant benefits in the form of its ability to:
Analyze EHR and population health data:
Thousands of intelligent digital health workers could prepare patient data for analysis and then separate patients into different cohorts based on hemoglobin levels. These digital health workers could then intelligently review documents to cross-reference hemoglobin levels with other CDC prioritization factors (like recent emergency room admittance or additional pre-existing or chronic conditions ), COVID-19 testing and antibody tests data to identify those most at risk, then identify a local provider with appointment availability.
Automate patient engagement, communications and scheduling:
After patients with diabetes are identified and prioritized, communications will be essential to quickly schedule those at most risk and prepare them for their appointments, including making them feel comfortable and informed. For example, digital health workers could communicate with diabetes patients about the protocol they should follow before and after their appointment – should they eat before the visit, what they should expect during their visit, and is it safe for them to return to work after. It’s also highly likely that widespread vaccine administration will require a far greater amount of information than with other health communications, given that one in three Americans say they would be unwilling to be vaccinated if a vaccine were available today. At scale, communications and scheduling will take potentially millions of hours in total, and all of that time takes healthcare employees away from actually providing care.
While the timeline for approval of a COVID-19 vaccine is unclear, now is the time for hospitals to prepare their technology and operations for the rollout. By adopting RPA, state healthcare organizations and providers can set themselves up for success and ensure that the patients most critically in need of a vaccine receive it first.
About Ram Sathia
Ram Sathia is Vice President of Intelligent Automation at PK. Ram has nearly 20 years of experience helping clients condense time-to-market, improve quality, and drive efficiency through transformative RPA, AI, machine learning, DevOps, and automation.