CV19Index Frequently Asked Questions
What is the CV19 Vulnerability Index (CV19_Index)?
The CV19 Vulnerability Index (CV19 Index) is an open source, AI-based predictive model that identifies people who are likely to have a heightened vulnerability to severe complications from COVID-19 (commonly referred to as “The CoronaVirus”). The CV19 Index is intended to help hospitals, federal / state / local public health agencies and other healthcare organizations in their work to identify, plan for, respond to, and reduce the impact of COVID-19 in their communities.
To be absolutely clear – the CV19 Index does NOT predict who will become infected with COVID-19 or geographic locations where the virus might spread. It is meant to identify people with a heightened risk of severe complications should they become infected.
Why is a Vulnerability Index Important?
The risk of death has been difficult to calculate, but a small study of people who contracted covid-19 in Wuhan suggests that the risk of death increases with age, and is also higher for those who have diabetes, disease, blood clotting problems, or have shown signs of sepsis. With an average death rate of 1%, the death rate rose to 6% for people with cancer, high blood pressure and chronic respiratory disease, 7% for people with diabetes, and 10% for people with heart disease. There was also a steep age gradient; the death rate among people aged 80+ was 15%.
What conditions are included in the proxy endpoint?
The proxy endpoint includes diagnosis codes for Pneumonia, Influenza, Acute Bronchitis, and Other Respiratory Infections. They are specified using ICD-10-CM codes from International Classification of Diseases, 10th Revision, Clinical Modification and CCS categories from AHRQ’s Clinical Classifications Software CCS.
How is this different than risk factors the CDC has already posted on their website?
The risk of severe complications from COVID-19 is higher for certain vulnerable populations, particularly people who are elderly, frail, or have multiple chronic conditions.
Identifying who is most vulnerable is not necessarily straightforward. More than 55% of Medicare beneficiaries meet at least one of the risk criteria listed by the CDC. People with the same chronic condition don’t have the same risk, and simple rules can fail to capture complex factors like frailty which can make people more vulnerable to severe infections.
The CV19 Index addresses this complexity at-scale with data that is readily available for more than 40 million Medicare beneficiaries through Blue Button and other transport mechanisms.
Where did you get your training data?
Since real world data on COVID-19 cases is not readily available, the CV19 Index was developed using close proxy events. A person’s CV19 Index is measured in terms of their near-term risk of severe complications from respiratory infections (e.g. pneumonia, influenza). The model was built and tested using administrative claims data for more than two million elderly and disabled individuals.
What risk factors, or “features" did you use in the model?
The CV19 Index uses medical claims data that reflects an individual’s medical history, chronic conditions, comorbidities, prior adverse events, recent medical treatments and procedures, functional limitations, and degree of frailty. It also integrates publicly-available data for several social determinants of health. More details on risk factors, known as “features” used in the model, such as Continuity of Care, Hospital Acquired Infections, and Fall-Related Injuries, along with other supporting technical details are available in the white paper here.
What can I do with the results of the CV19 Vulnerability Index?
The CV19 Index shares some similarities with the CDC Social Vulnerability Index which helps officials respond to environmental or natural disasters. In the case of COVID-19, identifying a community’s more vulnerable individuals can help organizations in the following ways:
- Identify areas with higher concentrations of vulnerable people
- Facilitate local preparedness (e.g. hospitals, first responders, community health workers)
- Plan for additional healthcare use
- Plan for and coordinate efforts with other organizations
The CV19 Index is distinct from other tools in that it also supports individual and targeted outreach, which is critical as communities shift from containment to aggressive mitigation. Efforts here are only beginning to take shape, but will likely be used for targeted outreach to include:
- Provide information about locally-available resources (e.g. food delivery)
- Recommend alternative activities in their area (e.g. virtual worship)
- Identify individuals who need resource support (e.g. food, medicine, etc.)
Conduct regular check-ins (since social distancing is being promoted as a way to slow the spread of infections)
How can I help?
We are on a mission to mobilize health care systems to more effectively target their efforts to protect the vulnerable members of their community. Creating the CV19 Index is just the start. We need your help to truly make a difference. We need your help if it’s going to actually make a difference. Let us know how you would like to get involved. Specific areas where help is needed will be updated soon. In the meantime, please email email@example.com if you have questions or would like to help support these efforts.
Is the CV Index applicable to everyone?
The CV19 Index has been trained and tested on a data set of approximately 2 million anonymized Medicare records. It is applicable for Medicare and Medicare Advantage beneficiaries in the US. It should not be used for healthy, younger adults. If you would like to work with us on expanding the CV19 Index to a wider population, please contact us at firstname.lastname@example.org
Could This Be Used Internationally?
Clinicians and data scientists in other countries have access to the open source models. Many countries already use International Classification of Diseases, 10th Revision, Clinical Modification (ICD-10-CM) codes. Experts could assess the appropriateness of specific features and make adjustments necessary for specific considerations of their country’s healthcare delivery system and its patients.
What is ClosedLoop doing to make sure the COVID-19 Index software is secure?
ClosedLoop has partnered with Ion Channel to ensure that COVID-19 Index code is continuously analyzed. The code and all its dependencies are checked for viruses, vulnerabilities and other risks every time the code is updated and no less frequently than once a day. New code, either from ClosedLoop developers or community contributors, all meet security criteria to be approved for addition to the code base.
Ion Channel is a software supply chain logistics and assurance platform that continuously monitors software used in critical infrastructure. Continuous monitoring gives customers immediate notice of vulnerabilities in vendor products and open source software and measures suppliers’ time-to-response when remediation is required.
Is this HIPAA Compliant?
Use of the CV19 Index in the ClosedLoop cloud hosted environment is HIPAA compliant. This requires a signed Business Associate Agreement, or BAA, available through the download and deploy process.
Do I need to sign a Business Associate Agreement (“BAA”)?
If the CV19 Vulnerability Index is deployed through direct download to your infrastructure or through AWS SageMaker, a BAA is not required. The use of the CV19 Vulnerability Index in the ClosedLoop cloud hosted environment requires a signed BAA, which is available through the download and deploy process.
How do I deploy the CV19 Vulnerability Index?
We are offering three deployment models
- Download the CV19 Index code at https://github.com/closedloop-ai/cv19index and run the code on your infrastructure to deploy the model.
- Subscribe to the “CV19 Vulnerability Index” Mode Package on AWS Sagemaker to launch an instance of the model within your own AWS environment..
- Use the CV19 Vulnerability Index on our multi-tenant HIPAA compliant cloud hosted environment.
No data supports correlating “proxy events” (e.g. pneumonia, influenza) & “near-term risk of severe complications from general [respiratory] infections” specifically to COVID-19 related ARDS. The only known preinfection granularities are comorbidities & age. How do you support your choice of “proxy events” (e.g. pneumonia, influenza)?
We agree no such studies have yet been performed. Unfortunately, data is not available yet that allows us to identify individuals who are at risk for severe complications of respiratory infections due to COVID-19. In conversations with clinical collaborators the decision was made to start by using proxies that reflect people who are generally more susceptible to severe respiratory infections. Research supports that the risk factors for the proxy events are consistent with those raised by the CDC. The models bring the added benefit of more granular prioritization.
If you would like to be involved in iterating on this model, please jump into the discussion on the forums and feel free to contribute to the open source project.
Do you have information on how this compares to some established score, such as Charlson Comorbidity?
The following figure is a direct comparison between our CV19 model and Charlson Comorbidity. I’ve depicted their relative performance using our “percent outcome capture” graph, which shows the tradeoff between the number of adverse events captured (i.e. Sensitivity) as a function of the alert rate. At lower alert rates (5%-10%), we have significant lift over CCI. Particularly, at a 5% alert rate, we have a 34% sensitivity, compared to 17% for CCI.