dbGaP Study Accession: phs002588

NIH Institute/Center: NCATS

RADx Data Program: RADx-rad

DOI: 10.60773/cfem-qh07

Release Date: 11/11/2022

Updated Date: 01/18/2024

Study Description: COVID-19 presented a public health emergency: There was a critical need for rapid, not reagent-intensive, non-invasive testing technologies. This program led to the production of a prototype system to diagnose COVID-19 infection using the body odor signature of the disease. The goal of this program was to maximize societal impact by creating a validated prototype that can be used in a community or workplace setting by minimally trained personnel for low-cost, on-the-spot diagnosis within minutes. The system was developed in a manner that puts it on a pathway for rapid FDA approval. The Research Aims were: Aim 1. Optimization, assembly, and integration of a prototype system with the ability to examine odor signature of COVID-19 in samples of body odor. The system was simple to use, posed essentially zero risk to the operator and the test subject, and reported a result within minutes. The production cost at scale was approximately $9,000 for the complete measurement system, with a per test cost of approximately $0.50. The design and construction of the prototype was conducted by Novo Engineering, a leading firm with extensive experience in medical device development. Aim 2. Software development. Software for the system from VOC sampling to final diagnostic result was developed to ensure error-free operation of the device. Preliminary results suggested that simple linear discriminant analysis (LDA) does an excellent job of classifying VOCs from human body odor as COVID-19 positive or negative (92% sensitivity and 87% specificity). Optimization of the sensor array (Aim 1) and use of richer feature sets in the classifier models led to further performance improvements in the prototype system. Aim 3. System Benchmarking and Validation. The full prototype system was benchmarked against a number of VOC mixtures, with and without in vitro skin models. The system went through extensive testing against body odor samples from individuals with pathological conditions other than COVID-19 and other sources of potentially confounding VOCs. The prototype was validated against 400 samples drawn from the COVID-SAFE program at Penn. The screening included all members of the Penn community, and represented incredible racial and ethnic diversity as well as a wide variance in age, sex, and gender. Aim 4. Regulatory Approval Plan. The plan was developed under the direction of Sr/Key personnel John Fuson, JD, an attorney at Crowell & Moring LLP and a former Associate Chief Counsel at FDA. Novo Engineering had extensive experience in guiding prototype design in alignment with the requirements for FDA approval. The proposed COVID-19 VOC-based testing device was regulated by the FDA, likely as a Class I or II medical device. Because there is no clear predicate device to reference in this case, a direct de novo petition was submitted to FDA asking the agency to categorize and clear the proposed COVID-19 testing device as Class I or Class II without reference to any predicate.

Principal Investigator: Johnson, Alan T

Has Data Files: Yes

Study Domain: Novel Biosensing and VOC; Medical Device/Tool Development; Virological Testing; Rapid Diagnostic Test (RDT)

Data Collection Method: Unspecified COVID Testing Device

Keywords: Networking and Information Technology Research and Development (NITRD); Individuals with Pathological Conditions Other than COVID-19; Body Odor; Infectious Diseases; Bioengineering

Study Design: Longitudinal Cohort

Multi-Center Study: TRUE

Study Sites: University of Pennsylvania; Florida International University

Data Types: Individual Phenotype; Electronic Medical Records; Other

Data Types, Other: VOC signature data and VOC species ID data from chromatography

Study Start Date: 12/21/2020

Study End Date: 05/31/2022

Species: Human Data

Estimated Cohort Size: 400

Study Population Focus: N/A

Acknowledgement Statement: This study was supported through funding, 4U18TR003775-02, for the National Center for Advancing Translational Sciences (NCATS) as part of the RADx-rad program. Sample collection done in partnership with VOC Health (CEO Richard Postrell). Approved users should acknowledge the provision of data access by dbGaP for accession phs002588.v1.p1, and the NIH RADx Data Hub. Approved users should also acknowledge the specific version(s) of the dataset(s) obtained from the NIH RADx Data Hub.

Funding Opportunity Announcement (FOA) Number: RFA-OD-20-017

NIH Grant or Contract Number(s): 4U18TR003775-02

Consent/Data Use Limitations: General Research Use