HR001123S0011-Amendment-01.pdf

The purpose of Amendment 1 to HR001123S0011, the DARPA Triage Challenge Broad Agency Announcement, is to extend the full proposal deadline and remove restrictions on Human Subjects Research (HSR). Changes bolded/highlighted/underlined below.

Part I Overview Information is updated as follows:

o Full Proposal Due Date and Time: February 27 2023, 4:00 PM ET o BAA Closing Date: February 27, 2023

Section 1.1.3.1 Triage Signatures is updated as follows:

The notional DARPA Triage Challenge primary triage setting is the first few minutes of an MCI, with automated capture and processing of data from stand-off sensors to support medical responders in evaluating an overwhelming number of casualties. DARPA expects to provide the training data and test data necessary for the workshops and competition events, though competitors may use or develop their own training data (see 1.4.2). Human subjects research (HSR) may be proposed (see 4.2.3.3). Sensor-algorithm systems should assess aspects of victim status that medical personnel would evaluate in this initial, rapid stage of triage, such as ability to move, severe hemorrhage, respiratory distress, and alertness. Examples of stand-off strategies could include computer vision, motion tracking, remote photoplethysmography, lidar assessment of respiratory or cardiac activity, and speech recognition, among other potential strategies.

Algorithms should be trained to integrate data from multiple stand-off sensor streams and generate a real-time assessment of victim status, most importantly the need for immediate LSIs that medical personnel are trained and equipped to provide in military and civilian pre-hospital settings, such as external hemorrhage control and airway interventions.

Section 1.2.1 Primary triage competitions, Real World (Tracks A and B) is updated as follows:

Challenge events for Real-World competitors will be physical simulations of casualty scenarios.

Although the setting and complexity of challenge events will vary over the course of the DARPA Triage Challenge, the following features are expected to be maintained across events. Each competitor will have access to the same scenario, and no two (2) teams will operate in the same location simultaneously. Competitors will have only general information on the setting beforehand—for example, that it is a battlefield scenario, or a collapsed building following an earthquake. There will be actors and manikins exhibiting varying injuries and severity of injury (subject to the limitations of what can be simulated).

Competitors will use their stand-off sensors, robotic mobility platforms (e.g., UAVs, UGVs), and algorithms to autonomously process sensor data and provide real-time casualty identification and injury assessment. No part of a competitor’s system may touch a casualty or manipulate the scene (e.g., clear rubble). Each scenario will have a time limit, with no scenario expected to have a duration greater than one hour. Scoring will be based on a combination of speed and accuracy measures as described in Section 1.2.1.1.

At the beginning of each phase, DARPA will provide additional training data on the participant portal based on multimodal recordings of physical simulations of casualty scenarios similar to the end-of-phase challenge event. To assist Primary Triage Real-World competitors in developing their strategies, DARPA will host workshops at month 8 of Phase 1 and month 4 of Phases 2 and

3. These workshops will provide an opportunity for Real-World competitors to collect data from physical simulations of scenes similar to end-of-phase events. Real-world competitors should plan for 4 or 5 days of travel to each workshop and challenge event, each to be held at various locations in the continental United States. DARPA expects to provide the training data and test data necessary for the workshops and competition events, but competitors may use or develop their own training data (see Section 1.4.2; DARPA will not fund competitors to conduct animal research). Research plans to collect data from human subjects must be approved by local Institutional Review Boards (IRBs) and DoD Human Research Protections Office (HRPO) (see 4.2.3.3).

Section 1.2.2 Secondary triage competition is updated as follows:

The objective of the Secondary triage competition is to identify physiological signatures of injury derived from data captured by non-invasive sensors (contact-based or stand-off) to enable anticipatory decisions and prioritization for medical evacuation and care. Performers will develop algorithms that detect signatures in these data streams to provide decision support appropriate for austere and complex pre-hospital settings. Of particular interest are early signatures indicating a need for LSIs against conditions that medics are trained and equipped to treat during secondary triage, such as hemorrhage and airway injuries.

The Secondary triage competition is virtual only, and will use DARPA provided de-identified, multi-modal physiological data from trauma patients in diverse settings and cohorts provided by the DARPA Research Infrastructure for Trauma with Medical Observations (RITMO) program.

DARPA will provide access to a subset of these data for algorithm training, and evaluate competitor algorithms on holdout test data in end-of-phase challenge events. DARPA expects to provide the training data and test data necessary for Secondary triage competition workshops and competition events, but competitors may use or develop their own training data (see Section 1.4.2; DARPA will not fund competitors to conduct animal research). DARPA expects that any HSR conducted for tracks D and E will be restricted to use of pre-existing datasets.

Scoring may include metrics that combine speed (time to make a prediction) and accuracy measures (Section 1.2.2.1). Secondary triage competitors may be DARPA-funded or self-funded.

Challenge events will become progressively more complex and realistic from Phases 1 to 3, as described in Section 1.2.2.2.

Section 1.3.2 Secondary Triage: Predictive Signature Development

Proposers must define the analytical approaches they will use to generate real-time predictions of LSIs based on multimodal non-invasive (contact-based or stand-off) physiological monitoring data. Proposers should refer to the RITMO BAA (HR001122S0043) for DARPA’s solicitation of proposals to plan their approach, as training data will be produced by RITMO performers.

https://sam.gov/opp/c8d42ae0a28a4103a9ceb1d3eae57439/view

Specifications on the DARPA-provided data for training and challenge events will be announced by the DARPA Triage Challenge Kickoff. In short, DARPA expects to provide DARPA Triage Challenge Secondary triage competitors access to de-identified, large-volume, multimodal sensor, intervention, and medical outcome data obtained from trauma patients during the early post-injury period from multiple trauma settings. Additional patient information, such as demographic data and injury mechanism and pattern, may be available as well. DARPA will provide access to an initial dataset at DARPA Triage Challenge kickoff. At the end of Phase 1 and during Phase 2, we anticipate additional releases of datasets collected by RITMO performers that will include additional sensor data that could enhance development of predictive signatures. DARPA expects to provide the training data and test data necessary for Secondary triage competition workshops and competition events, but competitors may use or develop their own training data (see Section 1.4.2; DARPA will not fund competitors to conduct animal research).

Section 4.2.3.3 Humans Subjects Research is added:

Proposers that anticipate involving human subjects or animals in the proposed research must comply with the approval procedures detailed at https://www.darpa.mil/work-with-us/humanresearch, to include providing the information specified therein as required for proposal submission.

https://www.darpa.mil/work-with-us/humanresearch https://www.darpa.mil/work-with-us/humanresearch