A new study by Mayo Clinic and Nference Inc. researchers suggests that prior childhood and adult vaccinations for illnesses such as polio, measles and flu may provide protection against COVID-19 infection. The study, which analyzed patient data using Nference artificial intelligence (AI) software, underscores the critical role immunizations play in curbing the spread of diseases and preventing future pandemics.

For the study, the researchers examined immunization records for 137,037 tested in the Mayo Clinic health system for coronavirus infection and then compared matched pairs of vaccinated and unvaccinated patients using Nferx, Nference’s augmented reality software. The results showed that individuals who received childhood vaccinations for diseases like measles and polio, as well adult flu shots and other vaccines, had lower COVID-19 infection rates than people without earlier vaccinations.

‘Compelling evidence’

“What we discovered represents compelling evidence that vaccinations are a critical element in prevention of disease, even diseases one doesn’t anticipate. We know now, more than ever, that greater acceptance of inoculations can have enormous health benefits all over the world,” said Venky Soundararajan, co-founder and chief scientific officer of the Cambridge, Mass.-based company.

Andrew Badley, a viral disease specialist at Mayo Clinic and enterprise chair of the COVID-19 Task Force, underscored that message. “If you’ve received a number of different vaccines previously, your risk of having a positive SARS-CoV-2 diagnosis was about 60 to 80% of the rate of positive tests in unvaccinated patients. If you think of your immune system as a muscle, the more often you exercise that muscle, the stronger it is.”

The paper, “Exploratory analysis of immunization records highlights decreased SARS-CoV-2 rates in individuals with recent non-COVID-19 vaccinations,” was published Feb. 26, 2021, in Scientific Reports.

The Nference software has synthesized millions of patient's completely de-identified health records, together with the historical and ongoing data sets on SARS-CoV-2 PCR diagnostic testing, immunization records with either mRNA vaccine from Moderna Inc. or Pfizer Inc./Biontech SE, and COVID-19 outcomes including hospitalization, intensive care unit admissions and deaths, Soundararajan told BioWorld. “The automatically extracts the context, sentiment and causal linkages between any concept mentioned in the text of the physician notes in the patient EHRs and triangulates those insights across any cohorts defined by the user.”

This process of creating cohorts and comparing them to detect statistically significant features takes seconds to complete vs. weeks or months when done manually or using existing software.

Founded in 2013, Nference has been collaborating with the Mayo Clinic since early 2020. It has raised a total of $145 million to date to advance its augmented intelligence software, the latest in a series C round completed in December that reeled in $60 million. Mayo is one of four investors, along with Matrix Capital Management, Matrix Partners and NTT Venture Capital.

Assessing vaccine safety

Nference has been collaborating with the Mayo Clinic since early last year. The team previously used AI software to scan the medical records of early recipients of the Moderna and Pfizer/Biontech vaccines to see how real-world efficacy compared with that seen in clinical trials. According to the results, the risk of contracting COVID-19 was roughly 89% lower in people who received two doses of vaccine, compared with unvaccinated individuals, beginning 36 days after they received the first shot.

In addition to the vaccine study, Nference is actively implementing its real-time clinical decision support software to identify biomarkers that might correlate with or be indicative of rare instances of vaccine-associated side effects, Soundararajan said. “Building on our other recent demonstrating very high real-world efficacy of the mRNA vaccines, we are also trying to qualify for the first time the real-world durability after one vaccine shot, as well as both the vaccine shots across various subpopulations.”

Implications for diagnostics

Beyond COVID-19, Nference and Mayo are working to construct a universal early disease diagnosis software that speed the detection of diseases – and actional responses or therapies – months or years earlier than by current methods.

Soundararajan sees clear implications for Nference software in diagnostics. “In addition to relying on clinical trial reporting, there is a clear need for sophisticated natural language processing technology, such as the Nference augmented Intelligence platform, that automatically parses the context, sentiment and nature of associations between concepts described in the unstructured text of the EHR clinical notes, towards enabling more real-time diagnostic inferences from real-world data sets,” he said. Such near-real-time implementation of EHR-wide inference software across health systems has the potential to transform how we continue monitoring the efficacy, safety, durability and tolerability of the COVID-19 vaccines against currently circulating and newly emergent strains of the coronavirus.”

Meanwhile, Nference is actively collaborating with a number of top pharma companies using data science across a range of therapeutic areas. In June 2020, the company formed a multiyear alliance with Janssen Research & Development LLC to use AI to find novel targets and disease subsets, as well as to stratify patients and identify optimal clinical trial sties.

The company is also expanding its collaborations with med-tech companies. “The potential to enable AI-powered operational efficiency and efficacy across biopharma and med-tech enterprises is nearly limitless,” Soundararajan said.