Whatever mess is going on with VA aside, DoD has demonstrated that Cerner can be deployed successfully at large scale.…
AI and ML – Help Change the Course of the Pandemic and Make Money
By Jeremy Harper
Jeremy Harper, MBI is an independent consultant.
Twenty million people in the USA and 75 million people worldwide have tested positive for COVID-19. Public health estimates that six times more people have had the disease and are not aware. A vaccine will slow the rate of growth, but no one is expecting it to eradicate the virus as it mutates and adapts.
Even as we hit a year since the infections began, we know very little about the long-term consequences and impact for those who have had COVID-19. We are looking at how those impacts are different for people who have asymptomatic, symptomatic, or hospitalization issues. The problem is that research study after study has been released on small populations that are seen at local health systems. That information has been better than not sharing at all, but the small populations of patients at local sites have led us down directions that wasted time, effort, and energy.
The National COVID Cohort Collaborative (N3C) is the largest central data repository in the history of the NIH targeted to a single disease. It has over 400,000 records of patients who have been positively identified as being afflicted with COVID-19 out of 2.5 million patients total. Each week as they onboard additional academic medical centers, the patient population grows. This large initiative is supported by academic centers around the nation, bringing together some of the best healthcare minds to identify solutions.
A large national dataset of people who have had COVID-19, which we call a disease cohort, is required because it gives us the opportunity to pool data to create groups of people to reveal patterns and help people cope with long-term consequences of having the disease. This dataset, however, isn’t only useful for NIH-funded research. This dataset will also be transformational for health systems. Models can be quickly built and deployed to predict the business needs we are experiencing, and will experience, at health systems over the next years. Models that may not have captured intellectual property with this freely available resource, but cannot be implemented within the standard health system without experts to explain and deliver specific actions to take from the information and models that are built.
There is so much that we don’t know as we move forward in the healthcare domain with COVID-19, but we have opportunities to make a difference. We are moving beyond the local environments that only leverage standard Structured Query Language (SQL) to a future with large data lakes. Without such pooled data, we may take a decade to understand the extent of the problem and be able to ask questions across health systems to understand the issues. This centralization will allow us to research and implement within months instead of years after the initial data collection.
Even with the multiple vaccines, this disease state isn’t finished. People who have been immunized with the vaccine can still get the disease, though at lower rates. The natural evolution of the disease has been impacted by the changes we have made to our societies and interventions in which we have engaged. We may never fully understand or be able to model with accuracy where we would have been without what has been needed to control the impact, but we do have fantastic natural experiments to compare variables. We know so little today and we must test and implement interventions that have been held back.
Let’s take some examples from the problems that people face after having COVID-19, the three most commonly known long-term impact areas in the lungs, brain, and heart. These are problems that are waiting for ambitious business solutions.
LUNG: You may know at a local health system how many people have experienced lung scarring, but you won’t have a large enough population to predict the interventions that will be required over the upcoming years. This population is large and diverse enough to have concrete predictions for what will be required.
BRAIN: We have no idea how the widely reported COVID brain impacts will play out, but it’s certainly going to require new interventions. Working with health systems in conjunction with the N3C may help us tease apart genetic, environmental, or disease specific areas that are vital for patient intervention. By being on the forefront for identification of those afflicted, we will be able to package and deliver opportunities to help individuals. Influenza and pneumonia vaccinations have been tied to lower risk of Alzheimer’s Dementia. There is a very real risk that we will see higher prevalence in the future, and monitoring and helping health systems will impact lives
HEART: We have entire hospitals dedicated to this vital organ. We know that myocarditis, which is an inflammation of the heart muscle, is occurring frequently in COVID-19 patients. This has the danger of leading to heart failure in the future. Patients will need to be flagged to be monitored for this going forward. Health systems will need to potentially reach out and notify patients that they need to be vetted for early symptoms of heart failure. As this grows and progresses, health systems will need to pivot to be able to handle the underlying disease states in their patient populations.
The healthcare industry is experiencing disruption as a result of these external forces that is unprecedented. Any time an industry experiences this level of disruption, it provides opportunities for improvement and adoption of third-party solutions. We have the ability to create many metrics, create many perspectives, and work through many issues. The N3C gives us many opportunities to connect and collaborate across organizations. While the N3C will not be appropriate to answer every question, it can answer many urgent scientific and operational questions through its different data access levels.
Examples of the types of questions that can be difficult to tackle include those that look for discrete answers, such as whether someone is asymptomatic or not. There are swaths of people who have been positive without any symptoms. While we can identify the primary cause of some hospital stays, we don’t have a consistent answer over whether someone came into the hospital because of COVID-19 or if they came because of another reason and happened to test positive. There is currently no universal standard to track the new vaccinations and which brand of vaccination may have been administered.
Than N3C has a higher potential for business to partner with research in an agile rapid manner than do most research infrastructures. The N3C team is a team of distributed participants, allowing for communication with the team in real time, while at the same time retaining full opportunity to query the data. The N3C team is also able to work dynamically upon normalizing and rationalizing what is being found within the database. Data can be created and archived in a single location for future analysis, and analysis within a team could provide a new way of communication for your business.
A combination of cloud computing, open data, and hosting ensures that your business can utilize the N3C Data Enclave. This cloud-based platform has taken research from an expensive system that we each need to implement into an inexpensive solution that we can all access. This is the new technology that has replaced outdated and slow research & development (R&D) methods.
It is the time to make it available for your business and your team. The solution ensures the business will eliminate the traditional costs and time associated with large, expensive research facilities. It allows business to do what it does best: rapidly innovate and leverage data to deploy solutions at facilities around the nation.
If you are interested in learning more, onboard to N3C or email me at email@example.com.