At the 2023 ACOG Annual Clinical & Scientific Meeting, a look at how AI and immunofluorescence scanning microscopy can help in detecting vaginitis and increase the health of the vaginal microbiome.
According to Sarah Johnson, BS, a medical student at Edward Via College of Osteopathic Medicine-Carolinas, Spartanburg, South Carolina, “approximately 50% of vaginitis cases are misdiagnosed because of human error with the current diagnostic method of light microscopy wet mount.” If left untreated, vaginitis can lead to a number of complications, including an increased risk of sexually transmitted diseases, as well as pelvic inflammatory disease.
Johnson set out to assess the efficacy of artificial intelligence (AI) and immunofluorescence scanning microscopy, which can be used to look at tissue sections, cultured cell lines, or individual cells, and may be used to analyze the distribution of proteins, glycans, and small biological and nonbiological molecules to help in assessing the health of the vaginal microbiome.
An institutional review board approval was obtained for the study, with patients fitting the inclusion and exclusion criteria. Three swabs from each patient was collected; one for a traditional wet-mount diagnosis (also known as a vaginal smear), and 2 for DayZ Vaginal Health Assessment Assay (a sample collection kit, which includes reagents for staining the vaginal microbiota, and an automated fluorescent imaging system that uses AI to insure better acccuracy.
Targets of interest were clue cells, yeast pseudohyphae, and trichomonads. The assessment compared the wet-mount findings to the AI algorithm results.
In the preliminary data from 58 patient samples, AI had an accuracy of 95% in finding trichomonads; 91% for finding candida, and 90% for clue cells. The AI also detected trichomonads on one sample that had been missed by the expert reader, and detected pseudohyphae on 3 samples that were undetected by the expert reader.
Results clearly showed that the DayZ Vaginal Health Assessment Assay proved superior in its ability to discern wet-prep findings compared to the traditional wet-mount evaluation.
As AI assays become more and more integrated into the medical world, proving elevated levels of accuracy compared to more traditional screenings, no doubt practitioners will make greater use of these algorithms in testing for any number of vaginal disorders.
Johnson SB. Evaluation of vaginal microbiome to estimate vaginal health in gynecology patients presenting with vaginitis. 2023 ACOG Annual Clinical & Scientific Meeting. May 19-21. Baltimore, Maryland.