AI solutions for forecasting, supply-chain, quality control, and safety

Introduction

In any industry, artificial intelligence technologies can give employees “superpowers.” They can help workers accomplish tasks more quickly and with greater accuracy than any human being could ever achieve alone.

Healthcare, however, is a unique case. As far as we’re concerned, medical professionals are already superheroes — and, at best, AI can aspire to be their sidekick, like Batman’s Robin. But that doesn’t make its contribution any less significant. In times of crisis – especially now, with COVID-19 sweeping the nation – everyone could use a little help, even the healthcare industry. And AI can provide that needed support.

Artificial Intelligence for the Front Lines in the Battle Against COVID-19

The courage of healthcare workers in the face of an unknown, microscopic killer is a debt the rest of us will never be able to fully repay. Their empathy, bravery, and creativity are incredible – and certainly could never be replicated by machines (AI or otherwise.)

However, what AI can do is solve complex problems that involve massive amounts of ever-changing data. Problems like, for example, vaccine rollout.

The good news is that we have a vaccine. Many vaccines, in fact, with varying levels of efficacy, logistical complexity (e.g. refrigeration, number of doses), and with new virus variants to handle. These facts constitute a modern-day medical miracle, which someday we will all celebrate (after being vaccinated!), but there’s a big distinction between having a vaccine and getting people vaccinated.  

In this article, we will review many of the variables and problems to consider with U.S. vaccine rollout, and propose several AI solutions that could make an immediate and necessary impact.

The Biggest Data Puzzle of our Lifetime

Distribution of the COVID-19 vaccine requires solving for at least four interrelated problems, including:

  1. Demand forecasting: determining how much vaccine to ship where and when.
  2. Supply chain management: monitoring shipments for bottlenecks.
  3. Quality assurance: validating vaccines and assuring storage temperatures are met and maintained.
  4. Safety surveillance: monitoring vaccine recipients for adverse reactions.

The minority populations hit hardest by COVID-19 infection rates are also the most distrustful of medical treatments due to a long, shameful history of medical experimentation sponsored by the U.S. government. These individuals also are much less likely to have access to adequate health education, a family physician, computer to book a vaccination appointment, and to transportation options. 

As we develop solutions for vaccine distribution in the United States, we must directly confront the barriers stemming from our country’s history of systemic racism, generational poverty, and education inequality. This step is crucial; it will  ensure that those who need the vaccine the most receive it as quickly as possible.

How Artificial Intelligence Can Help

Data Strategy: The current vaccines have a very short shelf life and must be stored at ultra-low temperatures. It is critical to ensure that they are distributed with as little waste as possible. Complicating the issue, more affluent and educated people are statistically more likely to want the vaccine, but they are also less likely to be exposed to the virus.  

How, then, can we ensure equitable access to vaccine distribution?

One AI-enabled strategy to ensure efficient and equitable distribution of the vaccine could be to analyze data collected from satellite imagery, cell phone locations, zip codes, demographics, health, and transportation. We recently completed a data strategy project which integrated 150 years of records into a cloud data lake so it could be instantly retrievable, thereby enabling powerful business insights.  

With the current vaccine rollout problem, data from all of these sources could help determine where people want the vaccine, when they want to get it, and who needs it the most. Pharmacies could then be stocked with vaccines in communities where the population was most eager to be vaccinated, and also in underserved communities where COVID-19 has hit the hardest. In the latter communities, robust educational and communication campaigns could be implemented with targeted messaging in social media and cell phone texts in conjunction with the planned arrivals of  mobile vaccination units.  

Object-tracking: An AI sub-discipline that uses cameras and computer programs to recognize objects and track their movement over time, “object tracking” can help keep tabs on vaccine distribution as cases move through the supply chain. Any anomalies can be picked up via a 24/7 observation system, shining a spotlight on potential bottlenecks – sometimes even before they occur. 

In a recent project with the University of Colorado at Boulder, we built a machine vision model to count pedestrian bridge-crossings in rural Rwanda from surveillance video, which then generated a report with time, count of people, and travel direction. 

A similar object tracking system with report generation could be especially useful tracking vaccines, especially since many existing supply-chain software solutions were built for a single supplier. Significantly, the COVID-19 vaccines must move through many different entities, including several drug manufacturers, shipping companies, hospitals, pharmacies, and mobile vaccination units – all while under strict temperature control. An AI-powered solution that works within all of these systems and is easily adoptable could be a key solution. 

Intelligent document processing. Every manufacturer developing a COVID-19 vaccine reported a small percentage of adverse reactions during the drug trials. As the FDA only approved these vaccines under “emergency use,” these side effects will likely continue. They will need to be monitored as millions more receive the vaccines to determine whether certain  groups with pre-existing conditions should not receive the vaccine. For example, at publication of this article, people with certain allergies are advised against taking the vaccine, but much is still unknown for pregnant women and children. Surfacing this data as quickly as possible and sharing it globally is imperative – and this is exactly what AI could help with.

President Biden intends to reach a distribution of 100 million vaccines in the first 100 days of his new administration. To meet this goal, there has been an aggressive push to establish new vaccine distribution centers across the country, repurposing infrastructure from National Guard hangars to NFL football stadiums. It is critical that any medical data that arises from these new distribution centers is synthesized to flag adverse reactions or unusual patterns. 

Documents containing data from hospitals, hangars, stadiums, and facilities all over the country will need to be scanned and recorded. This is an immense administrative undertaking; fortunately, it can be streamlined with machine vision “intelligent” document processing. For example, Synaptiq built a similar solution for a global immigration law firm to intake, identify, and process diverse legal documents from various sources. 

Conclusion & Next Steps

While our superhero healthcare professionals continue the fight against COVID-19, AI-assisted solutions could help reduce their enormous administrative burdens. Medical professionals could then focus on patient care and specialized, dynamic work – what they were trained to do best – while AI finds patterns in data to streamline roll-out, identify bottlenecks in the supply-chain, and even surface adverse side-effects.

To achieve the best possible outcome with deploying AI solutions in this effort, we recommend healthcare providers take these first steps:  

  1. Adopt a data collection tool with a friendly user-interface so everyone who should be recording data knows how to do so, and so that it is easy to use and train others.
  2. Address staff and patient privacy concerns related to AI.
  3. Track everything regularly and holistically.
  4. Prioritize data volume over AI sophistication. 
  5. Engage a firm like Synaptiq that can analyze the data quickly so that your staff gains efficiencies and reduces medical errors and side effects.

If you are a healthcare provider or involved in COVID-19 vaccine rollout, Synaptiq would love to help. We work with clients across a variety of industries including healthcare to help teams discover how the data they already have – or the data they could have – can solve their most pressing problems. Please feel free to contact us directly. Until then, we hope you and your loved ones stay safe and well.

Additional Reading

Applications of AI in Healthcare

We are passionate about helping people improve their health and are fortunate to work with amazing healthcare clients that have technology products serving providers, patients, and manufacturers.  A decade ago it would have been unthinkable that any commercial organization outside of an insurance company could unlock the value of their health data, but it’s clear…

Read more

Can AI Turbocharge Your Healthcare Software? Part 1 of 3

Can AI turbocharge your healthcare software? Part 1 of 3 Introduction How can artificial intelligence (AI) turbocharge your healthcare software? The key is “graph data.” In this three-part series, Tom Blue, Ryan Wright, and Tim Oates explore the evolution of healthcare as it relates to AI-powered technology and graph data. The Truth Has Been Right…

Read more