3. Injury metrics and criteria for occupant protection

Crash response dashboards are only as trustworthy as the injury metrics that feed them. A dashboard that plots "HIC" or "chest deflection" without encoding the underlying criteria, units, and pass/fail thresholds is a visualization, not an analytical instrument. This section establishes the four metrics that anchor most occupant-protection analytics — HIC, chest deflection, Nij, and AIS — and shows how each one is computed, interpreted, and surfaced in a dashboard context.

HIC: Head Injury Criterion

HIC quantifies the likelihood of head injury from translational acceleration. It is defined over a time window and takes the maximum value across the event:

HIC = max over (t1, t2) of [ (t2 - t1) * ( (1/(t2-t1)) * ∫ a(t) dt )^2.5 ]

where:
  a(t)  = resultant translational acceleration at the head CG, in g
  t1,t2 = window boundaries, with t2 - t1 ≤ 36 ms (HIC36) or ≤ 15 ms (HIC15)
  the integral is evaluated between t1 and t2

Two practical points dominate dashboard implementation. First, the acceleration must be the resultant of the three translational components, not a single axis; feeding an axial channel into a HIC tile silently understates injury risk. Second, the window length matters: HIC36 and HIC15 are not interchangeable, and a dashboard must label which variant it reports. A common regulatory reference is HIC36 ≤ 1000 for the driver in frontal impact; HIC15 is frequently used in pedestrian and some occupant protocols with lower thresholds. The dashboard should store the window length as metadata alongside the value so that cross-test comparisons remain valid.

Computing HIC in code is a sliding-window search. For each candidate start sample, extend the end sample up to the window limit, accumulate the integral, and evaluate the criterion; retain the maximum. The result is sensitive to sampling rate and to filtering — acceleration channels are typically filtered per SAE J211 (channel class 1000 for head CG), and the dashboard should record the filter class used, because an unfiltered trace can inflate HIC substantially.

Chest Deflection

Chest deflection measures how far the sternum moves relative to the spine under belt and airbag loading, and it correlates with rib fracture and thoracic organ injury. It is reported in millimetres and is direction-sensitive: the sign convention must be fixed before any aggregation. A workable convention is positive deflection toward the spine (compression) and negative for rebound; the dashboard must state this explicitly, because mixing conventions across tests produces meaningless averages.

Deflection is usually captured by a chest band or string potentiometer and is evaluated as a peak value, sometimes with a rate term (deflection rate, in m/s) that flags loading severity. Typical frontal occupant criteria sit in the range of 40–50 mm for the 50th-percentile male, with lower limits for smaller occupants; these are protocol-dependent and should be stored as configurable thresholds rather than hard-coded. In the dashboard, chest deflection is best shown as a time-series overlay with the criterion line drawn in, so an analyst can see both the peak and how close the trace came to the limit before it.

Nij: Normalized Neck Injury Criterion

Nij combines axial force and bending moment at the upper neck into a single normalized value. The four load cases — tension-extension, tension-flexion, compression-extension, compression-flexion — each have their own critical values, and Nij is the maximum of the four normalized ratios:

Nij = max( Fz/Fzc + My/Myc )   evaluated per load case

  Fz  = upper-neck axial force (N), tension positive
  My  = upper-neck bending moment (N·m)
  Fzc, Myc = critical values for the active load case

Example critical values (illustrative, 50th-percentile male):
  tension-extension:      Fzc = 6806 N,  Myc = 310 N·m
  compression-flexion:    Fzc = 6160 N,  Myc = 310 N·m

The critical values above are illustrative and must be replaced with the values from the governing protocol. The key dashboard requirement is that Nij is reported with its load case: a bare Nij of 0.9 tells the analyst nothing about whether the neck was in tension or compression, and the mitigation strategy differs completely between the two. Nij ≤ 1.0 is the usual pass threshold, but the margin to 1.0 and the dominant term (force versus moment) are what drive design decisions.

AIS: The Injury Severity Scale

AIS is not a sensor-derived metric; it is an anatomically based coding of injury severity from 1 (minor) to 6 (maximum, currently untreatable). It is assigned by trained coders from medical records or autopsy, and it is the bridge between physical measurements and real-world outcome. In a dashboard, AIS appears in two roles: as the ground-truth label for validating metric thresholds, and as the basis for composite scores such as MAIS (maximum AIS) and the Injury Severity Score (ISS).

The analytical value comes from pairing AIS with the continuous metrics. A scatter of peak chest deflection against AIS for the thorax, for example, lets an analyst see where the criterion line falls relative to actual coded injuries — and whether the chosen threshold is conservative or permissive for the population in the dataset. This is the core of dashboard injury analytics: metrics predict, AIS validates.

Raw Channels SAE J211 filtered Head CG accel ax, ay, az (g) Chest band deflection (mm) Upper neck load cell Fz (N), My (N·m) Medical records coded injuries Metric Computation HIC36 / HIC15 sliding window, resultant Peak deflection + deflection rate Nij (4 load cases) max normalized ratio AIS coding MAIS, ISS Dashboard Tiles HIC vs 1000 window length shown Deflection vs limit sign convention stated Nij vs 1.0 load case + dominant term AIS overlay metric vs coded injury Validation threshold calibration metric vs AIS scatter review threshold feedback: recalibrate criteria from coded outcomes

Bringing the Metrics into the Dashboard

The SVG above shows the analytical chain the dashboard must preserve: raw channels feed metric computation, metrics feed tiles, and AIS closes the loop by validating whether the thresholds are right. Three implementation rules keep this chain honest.

Carry provenance with every value. Each tile should expose the filter class, window length (for HIC), sign convention (for deflection), and load case (for Nij). A value without its provenance cannot be compared across tests or protocols.

Show margin, not just pass/fail. A HIC of 980 against a limit of 1000 is a pass, but it is a fragile one. Dashboards that color-code only pass/fail hide the margin that engineers need for design decisions. Plot the value against the criterion line and show the ratio.

Validate against AIS, not against intuition. The most valuable dashboard view is a scatter of a continuous metric against coded AIS for the same body region. If the criterion line sits well above the cluster of AIS 2+ injuries, the threshold is permissive for that population; if it cuts through the cluster, it is conservative. This is how metric thresholds earn their place in a protection program.

When a dashboard shows a metric near its limit, check the provenance fields before escalating. A HIC that jumps after a filter change, or a Nij that flips load case between runs, usually indicates a channel or convention error rather than a real change in occupant loading.
Do not aggregate HIC36 and HIC15 values into a single average, and do not average Nij across load cases. These are maxima over different definitions; combining them produces a number that corresponds to no physical criterion and will mislead any downstream threshold review.

With these four metrics and their provenance rules in place, the dashboard becomes a defensible analytical tool: every tile traces back to a defined computation, every threshold traces back to a protocol or a validated AIS relationship, and every comparison across tests is made on like-for-like definitions.

← Previous 📖 Contents Next →