Artificial Intelligence (AI)-assisted readout method for the evaluation of skin prick automated test results
The result's identifiers
Result code in IS VaVaI
<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F00216208%3A11110%2F25%3A10504131" target="_blank" >RIV/00216208:11110/25:10504131 - isvavai.cz</a>
Result on the web
<a href="https://verso.is.cuni.cz/pub/verso.fpl?fname=obd_publikace_handle&handle=th47zDtXpb" target="_blank" >https://verso.is.cuni.cz/pub/verso.fpl?fname=obd_publikace_handle&handle=th47zDtXpb</a>
DOI - Digital Object Identifier
<a href="http://dx.doi.org/10.1038/s41467-025-64334-w" target="_blank" >10.1038/s41467-025-64334-w</a>
Alternative languages
Result language
angličtina
Original language name
Artificial Intelligence (AI)-assisted readout method for the evaluation of skin prick automated test results
Original language description
The skin prick test (SPT) is the gold standard for diagnosing allergic sensitization to aeroallergies. The Skin Prick Automated Test (SPAT) device has previously demonstrated reduced variability and more consistent test results compared to manual SPT. The current study aims to develop and validate an artificial intelligence (AI) assisted readout method to support physicians in interpreting skin reactions following SPAT. To train the AI algorithm, 7812 wheals (651 patients) are manually labeled. To validate the AI measurement, the longest wheal diameter of 2604 wheals (217 patients) is measured by the treating physician and compared to the AI measurement. In addition, AI-assisted readout is validated on a separate test cohort of 95 patients (1140 wheals). We demonstrate that the AI measurements of the longest wheal diameter exhibit a strong correlation with the physician's measurements. The AI algorithm shows a specificity of 98<middle dot>4% and sensitivity of 85<middle dot>0% in determining positive or negative test results in the validation cohort. In the test cohort, physicians adjust 5<middle dot>8% of AI measurements, leading to a change in the test interpretation for only 0<middle dot>5% of cases. AI-assisted readout significantly reduces inter- and intra-observer variability and readout time compared to manual physician measurements. Altogether, the AI-assisted readout method demonstrates high accuracy, with minimal misclassification of test results. Adding AI to SPAT further improves standardization across the SPT process, significantly reducing observer variability and time to readout.
Czech name
—
Czech description
—
Classification
Type
J<sub>imp</sub> - Article in a specialist periodical, which is included in the Web of Science database
CEP classification
—
OECD FORD branch
30102 - Immunology
Result continuities
Project
—
Continuities
V - Vyzkumna aktivita podporovana z jinych verejnych zdroju
Others
Publication year
2025
Confidentiality
S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů
Data specific for result type
Name of the periodical
Nature Communications
ISSN
2041-1723
e-ISSN
2041-1723
Volume of the periodical
16
Issue of the periodical within the volume
1
Country of publishing house
GB - UNITED KINGDOM
Number of pages
8
Pages from-to
8637
UT code for WoS article
001586620700003
EID of the result in the Scopus database
2-s2.0-105017806289