Artificial intelligence can help patients interpret home tests for COVID-19

Researchers led by Professors Farrokh Alemi and Janusz Wojtusiak found that computerized symptom screenings can complement at-home COVID-19 testing to better confirm the diagnosis for patients and doctors.

New machine learning research led by Professor Farrokh Alemi and Professor Janusz Wojtusiak offers patients and doctors a way to better predict whether symptoms are due to COVID-19, influenza or RSV. A more accurate diagnosis leads to better decisions about the course of treatment to cure patients and prevent the spread of the disease. Together with fellow researchers from George Mason University and Vibrent Health, Alemi and Wojtusiak recently published a series of articles in a special issue of the Journal of Quality Management in Health Care discuss how artificial intelligence (AI) can help diagnose COVID from a combination of symptoms and home testing.

With their research, Alemi and Wojtusiak are now working on a website to provide an AI-based resource to help individuals identify recommended actions as a result of their clinical profile and at-home COVID test results.

“We see AI working to radically improve clinical triage and test-to-treat decisions,” Wojtusiak said. Alemi added: “AI will enable individuals to feel safer when it comes to staying at home, seeking care or socially isolating. Many people test at the end of their symptoms and surprisingly find that they are still positive. What to do when symptoms and home test results don’t match? Our AI will help these people understand what to do.”

The study in Paper 1 (as listed below) found that the timing of symptoms matters in a COVID diagnosis. For example, a runny nose as an early symptom increased the likelihood of testing positive for COVID, and a runny nose as a later symptom reduced the likelihood. Similarly, fever is almost always a late symptom, so early absence of fever should not be used to rule out COVID.

The results in Paper 2 found that COVID cannot be diagnosed based on individual symptoms; However, a cluster of three or more symptoms can help with the diagnosis. The results from Paper 4 found that the accuracy of diagnosing COVID symptoms was highest when symptoms of various physical symptoms were present. For example, a combination of neurological and common respiratory symptoms was more diagnostic than either group of symptoms individually. In addition, COVID has different presentations depending on age, disease severity and viral mutations.

Paper 3 discusses how AI symptom screening could improve and replace home antigen testing for vaccinated individuals. Home tests are not always accurate and require clinical verification, but these tests are performed at home when such verification is not available. AI symptom screening can help make these tests more accurate. The study reports that AI symptom screening is more accurate than a second home test.

The four articles published in the special supplement are:

  1. Order of appearance of COVID-19 symptoms
  2. The role of symptom clusters in triage of COVID-19 patients
  3. Combined symptom screening and home testing for COVID-19
  4. Guidelines for triage of COVID-19 patients with multisystemic symptoms

A fifth essay entitled Modeling the likelihood of COVID-19 based on symptom screening and prevalence of influenza and influenza-like illnesses, from the same group of researchers was also published in Quality management journal in the healthcare system in April/June 2022.

Alemi was Mason’s chief investigator. Mason was a subcontractor from Vibrent Health, where Praduman Jain was the project’s lead researcher. (Jain serves on the Advisory Board of Mason’s College of Public Health.) Other Mason-affiliated researchers on these projects include Associate Professor Amira Roess, Affiliate Faculty Member Jee Vang, graduate student Elina Guralnik, and alumni and Associate Professor Wejdan Bagais. Rachele Peterson and Josh Schilling from Vibrant Health and F. Gerard Moeller from Virginia Commonwealth University were also part of the research team.

The research was funded by the Digital Health Solutions for COVID-19 program established by the National Cancer Institute (NCI) and the National Institute of Biomedical Imaging and Bioengineering (NIBIB).

The methods used in these five papers vary. In Article 4, researchers performed a meta-analysis of the literature using data from published articles. In the other papers, the researchers interviewed patients who took a PCR test and examined the relationship between the patients’ symptoms and the PCR test results. Most of the research was done using data collected between October 2020 and January 2021, i.e. before the current variants such as BA.5 or BQ.1.

Previous related publications by these researchers include a study examining how computers can distinguish between COVID-19 and the flu Analysis of symptomatic university students and social distancing.

Leave a Comment