University of Southern California


RFA-OD-20-017: Emergency Awards RADx-RAD: Screening for COVID-19 by Electronic-Nose Technology (SCENT) (U18 Clinical Trial Not Allowed)

Slots:                                                     1

Internal Deadline:                             First-come, first-served.

LOI:                                                        August 18, 2020

External Deadline:                            September 18, 2020

Award Information:                         Type: Cooperative Agreement

Funds Available and Anticipated Number of Awards: NIH intends to commit $10 million in total costs over a two-year period. NIH anticipates funding 5 awards.

Link to Award:

Submission Process:                       PIs must submit their application as a Limited Submission through the Office of Research Application Portal:

Materials to submit include:

(1) Single Page Proposal Summary (0.5” margins; single-spaced; font type: Arial, Helvetica, or Georgia typeface; font size: 11 pt). Page limit includes references and illustrations. Pages that exceed the 1-page limit will be excluded from review.

(2) CV – (5 pages maximum)

Who May Serve as PI:                   Standard NIH eligibility requirements.

Purpose:                                              This initiative seeks to advance the development of novel, safe and effective biosensing and detection technologies for volatile organic compound (VOC) signatures of COVID-19 from human skin or the oral cavity. To this end, leveraging dedicated engineering and artificial intelligence systems is required. This initiative anticipates the implementation of such technologies in everyday settings and routines for detection, diagnosis, prediction, and monitoring of COVID-19 in clinical, community or applied settings. It is envisioned that these technologies will complement traditional virus and antibody detection to monitor the onset, progression, and resolution of COVID-19.

Specific Research Objectives:       Scanning for COVID-19 with Electronic Nose Technology (SCENT). A systems approach must be applied to product development and preclinical performance testing of the proposed device to establish a robust proof-of-concept feasibility.

  1. Assembly and integration of the prototype SCENT platform: The SCENT platform must comprise (1) a Volatile Organic Compound (VOC) sampler; (2A) an electronic nose or (2B) a gas chromatographic (GC) column; (3) an appropriate detector, (4) AI/Machine Learning capabilities to distinguish VOC skin signatures across many levels of COVID-19 infection for accurate diagnosis, and (5) an intuitive, user-friendly interface. Assembly of the hardware and software for either the E-nose or GC prototypes from off-the-shelf components is encouraged to expedite development.

The VOC skin sampler can be a simple cup where the VOCs emanating from skin are delivered to the e-Nose or GC column via nitrogen gas or other nonreactive gas, or it can be slightly more sophisticated as in solid-phase microextraction (SPME) where the gases are adsorbed on a material and desorbed into the e-Nose or GC column. The skin VOC sampling system will be developed through testing and optimization of potential designs preferably by Design of Experiments (DoE). The sampling system(s) will have to be validated against known mixtures of VOCs with and without in vitro skin models. Other novel skin sampling designs are encouraged.

For the oral cavity sampler, the oral environment presents specific design and performance requirements for VOC sampling that need to be addressed. Therefore, a system’s engineering approach must address major challenges imposed by the oral environment including, varying pH levels and temperatures, oral flora, adhesion to wet intraoral tissues, and material biocompatibility/biofouling.

The sampler for VOCs in the oral cavity can be a blow tube for breath and/or particulates/droplets. However, prevention of infection/contamination of the healthcare worker must be built-in into the design. The sampler can also be an oral probe (analogous to an oral thermometer) placed in contact with oral tissues (e.g., under the tongue) to collect VOCs emanating to the tissue surface. Alternatively, VOCs can be collected from the head space of saliva collected in a closed container. This is a standard procedure that has been applied to VOC analysis of urine. As in the skin sampler, other novel sampling designs for the oral cavity are encouraged.

The E-Nose is generally an array of a number of materials including conducting polymers, quartz crystal microbalances, fluorescence sensors, semi-conducting metal oxides, etc. The collective signals are processed in an artificial neural network and pattern recognition software. For GC the detector can be a mass spectrometer (MS) or flame ionization detector (FID). Identification of the VOC components by comparison to an MS data base is desirable; however, the pattern recognition is more important. Deposition of data/learning sets into the DCC is required.

The portable E-nose or GC instrumentation and detector/s will be integrated with the VOC sampler. For off-the-shelf components, compatibility of software must be considered even before assembly of the SCENT platform. The prototype has to be validated against the current standard, laboratory-grade, commercially-available GC or E-nose instrumentation (e.g., MEMS, etc.) and tested for sensitivity and accuracy against known mixtures of VOCs.

Quality by Design (QbD) and incorporation of process analytical technologies (PATs) are required and must be prominently described in the application. QbD will allow for ease and precision of future manufacturability and ensure that device to device differences are at a minimum, while adverse events that are of device origin are limited. Process analytical technologies are required for accuracy and precision of measurements between devices and between patients. PATs for devices may include standard VOC mixtures, pressure sensors, etc. PATs for patient to patient standardization must include skin temperature sensors, skin permeability sensors (e.g., Transepidermal Water Loss [TEWL], skin impedance, etc.) or skin stiffness among others, that can affect the quantity (and perhaps quality) of VOCs sensed by the detectors.

Adoption of human-factors-engineering and usability-engineering principles must be considered during the development process. This includes usability criteria such as comfort to ensure user acceptance and compliance.

Additionally, VOC sensing approaches for COVID-19 must incorporate design specifications and performance criteria for risk mitigation of potential measurement interferents including, but not limited to: metabolites produced in pathological conditions other than COVID-19; compounds introduced during patient treatment such as drugs, plasma expanders, and anticoagulants; substances ingested by the patient such as alcohol or nutritional supplements. Lastly, the proposed approaches must address other potential causes of examination (analytical) interference, such as: chemical, physical and detection artifacts; non-selectivity and non-specificity of detection; and other sources of error that might affect COVID-19 diagnostics. Risk mitigation and alternative methods are expected.

Most of the technological components required to build these sensing platforms are already developed and used for other purposes. The innovation in this initiative will be the bringing together of expertise in these technologies and coupling with clinical and infectious disease expertise to develop an integrative noninvasive device that will be used specifically for the diagnosis of COVID-19. Future applications of SCENT have potential for monitoring of overall health and detection of other diseases and will be a consideration but not required. The metrics/requirements for successful SCENT are accuracy, sensitivity and selectivity comparable to or exceed current standard, FDA-approved COVID-19 diagnostics. Portability, accessibility, and affordability are also key considerations for SCENT.

Since SCENT can potentially have global applicability and use, it must follow the principles of ASSURED (affordable, sensitive, specific, user-friendly, rapid and robust, equipment-free and deliverable to end users) criteria, outlined by the World Health Organization (WHO), which provides a good framework for evaluating point of care devices specially for resource-limited environments.

  1. Software Development. Code for the testing protocol from VOC sampling to analysis will be written and tested for smooth, error-free operation of the device. Off-the-shelf components from different manufacturers must have compatible software, which must be checked before assembly.

Use of commercially available pattern-recognition and machine learning software is allowed and even encouraged in the interest of time. The sensitivity and accuracy of the software must be tested on surrogate samples as in (1) above. In addition, reference/training and validation sets from actual clinical samples in (3) below. will be used to show proof of principle of the machine learning algorithm.

  1. Testing SCENT Prototype on Patients. Data training sets must be collected on known (1) COVID-19 positive, symptomatic, (2) COVID-19 positive, asymptomatic and (3) COVID-19 negative subjects as determined by the current standard FDA approved method.

NCATS Clinical and Translational Science Awards (CTSA) hubs can be used in the clinical validation for recruitment and trial implementation.

A statistically significant number of patients with known condition(s) are expected to be tested for confident delineation of (1), (2) and (3) based on training sets and comparable accuracy to FDA-approved COVID-19 diagnostics.

  1. Regulatory Approval Plan. A plan for the regulatory approval of technologies, tests and approaches must be developed based on the data generated from the research objectives. The plan should describe the expected regulatory pathway for the technology and describe foreseeable regulatory risks that could impact the technology development. The plan must also describe how the technology would fit with current standard of care.