AIS 3+: Traumatic Injury Severity Metrics

• AIS 3+ refers to injuries rated AIS 3 or higher, indicating serious to life-threatening trauma based on anatomical injury scaling.
• Computational methods such as Peak Virtual Power (PVP) mapping and entropy-based metrics provide deterministic and aggregated injury severity predictions.
• Current models incorporate age-related degradation and pathological sequelae to enhance the precision of AIS 3+ assessments in clinical and biomechanical research.

The Abbreviated Injury Scale (AIS) is a widely used, anatomically based coding system designed to classify and scale the severity of traumatic injuries. While the complete AIS covers six levels (1 = minor, 6 = currently untreatable), the term "AIS 3+" designates injuries of severity 3 or greater, which for many applications represent the threshold for serious, life-threatening trauma. AIS 3+ is critical in trauma biomechanics, automotive safety, clinical research, and forensic pathology, serving as a benchmark for injury prediction models, trauma scoring systems, and outcome metrics.

1. Definition and Clinical Meaning of AIS 3+

AIS assigns severity codes to specific injuries by body region. The "AIS 3+" threshold encompasses the injury levels:

• AIS 3: Serious injury (threat to life but survival probable)
• AIS 4: Severe injury (survival uncertain)
• AIS 5: Critical injury (survival unlikely)
• AIS 6: Maximal injury (virtually unsurvivable)

Empirical fatality data correlate these ratings with the probability of death, e.g., 4.7% (AIS 3), 18% (AIS 4), 50% (AIS 5), 84% (AIS 6) (Neal-Sturgess, 2010). Injuries below AIS 3, although potentially disabling, are generally excluded from population-level risk-to-life calculations, including the Injury Severity Score (ISS) and entropy-based scales.

2. Computational Frameworks for AIS 3+ Prediction

Historically, maximum principal strain (MPS) and von Mises stress derived from finite element (FE) human models have been used to estimate injury risk; however, these metrics are not directly mappable to AIS levels, especially across the critical 3+ threshold (Bastien et al., 2020, Bastien et al., 2019). Two distinct computational approaches have emerged for explicit AIS 3+ determination:

2.1 Organ Trauma Model (OTM) and Peak Virtual Power (PVP)

OTM replaces MPS with a thermodynamic power metric, Peak Virtual Power (PVP), calculated as

$$\mathrm{PVP} = \max_{t} \left[ \sigma_{\rm vM}(t) \cdot \dot\varepsilon_{\rm vM}(t) \right]$$

where $\sigma_{\rm vM}$ is the von Mises stress and $\dot\varepsilon_{\rm vM}$ is the von Mises strain-rate at time $t$ (Bastien et al., 2020, Bastien et al., 2019).

For a given element with volume $V_p$ and impact velocity $V_0$, PVP can be derived algebraically as: $$\mathrm{PVP} = \frac{A_p}{2 V_p} \frac{\rho_p E_p m_c + \rho_c E_c m_p}{\rho_p E_p m_c + \rho_c E_c m_p} V_0^3$$ with $A_p$ = contact area, $E_p$ = organ modulus, $\rho_p$ = tissue density, $m_p$ = organ mass, $E_c$, $\rho_c$, $m_c$ = contact body parameters (Bastien et al., 2020).

2.2 Mapping PVP to AIS Categories

Calibrated FE impactor tests allow the mapping of PVP to discrete AIS thresholds using a cubic power law: $$\mathrm{PVP}_{\rm AIS\,k} = \mathrm{PVP}_{\rm AIS\,4} \left(\frac{k}{4}\right)^3, \quad k=1,2,3,4,5$$ The AIS 3+ region is thus

$$\mathrm{PVP} \geq \mathrm{PVP}_{\rm AIS\,3} = \mathrm{PVP}_{\rm AIS\,4} \left(\frac{3}{4}\right)^3$$

Polynomial fits of the form $\mathrm{PVP}(V) = a V^3 + b V^2 + c V$ are generated for grey and white matter, facilitating direct computation of the minimum impact speed yielding AIS 3+ injury (Bastien et al., 2020, Bastien et al., 2019).

3. Entropy-Based Severity Metrics Utilizing AIS 3+

AIS 3+ injuries also form the computational basis for advanced severity scores derived from statistical mechanics. Neal-Sturgess (2011) (Neal-Sturgess, 2010) demonstrates that trauma entropy, calculated as

$S = -k \sum_i p_i \ln p_i$

summed over all AIS 3+ injuries present, yields a continuous risk metric, with $p_i$ the fatality probability associated with AIS level $i$ (empirically, $p_3 \approx 0.047$, $p_4 \approx 0.18$, $p_5 \approx 0.5$, $p_6 \approx 0.84$).

An additive framework is used:

• Trauma entropy ($S_T$): Summed over all individual AIS 3+ injuries.
• Morbidity entropy ($S_{mb}$): Baseline physiological entropy reflecting age/comorbidity.
• Death threshold ($S_{mt}$): Death occurs when $S_{mb}+S_T = S_{mt}$, with $S_{mt}$ conventionally set to 100 arbitrary entropy units.

This entropy formalism provides a more finely graded scale than ISS, especially for multiple moderate-severity (AIS 3/4) injuries, and explicitly accommodates patient vulnerability due to advanced age or comorbidities.

4. Incorporation of Age and Physiological Degradation

Accurate AIS 3+ assessment must account for age-associated anatomic and tissue changes, which directly affect injury tolerance and thus severity prediction:

• Brain volumetric shrinkage: Modeled as $V_{brain}(t) = -0.0037 t + 1.808$ with age $t$ in years (relative to age 20).
• Material property degradation: Both modulus and density are decreased by up to ~20% from age 20 to 80 years, reflected in reduced impact tolerance (Bastien et al., 2020).
• Morbidity entropy increment: For elderly or multimorbid patients, $S_{mb}$ is increased, producing lower required trauma entropy for fatal outcome (Neal-Sturgess, 2010).

These adjustments yield more accurate population-level and patient-specific predictions at the AIS 3+ threshold.

5. Treatment of Pathological Sequelae: Subdural Haematoma

Lagrangian FE models used for OTM are inherently volume-conserving and cannot directly represent bleeding, such as subdural haematoma (SDH). To address this, a post-processing rule is employed:

• If $\max(\mathrm{MPS}) \geq 0.255$ (i.e., element-level MPS exceeds 25.5%), increment AIS by +1.

Thus, the final computed injury severity is: $$\mathrm{AIS}_{\rm final} = \mathrm{AIS}_{\rm PVP} + 1 \quad \text{if SDH threshold exceeded}$$

This approach aligns model-based severity scores more closely with autopsy outcomes, particularly regarding life-threatening hemorrhagic sequelae (Bastien et al., 2020, Bastien et al., 2019).

6. Empirical Validation of AIS 3+ Prediction Methods

Validation against real-world pedestrian accident data demonstrates that OTM-PVP methodology provides close correspondence (within ±1 AIS level) to post-mortem determined trauma severity, while MPS-based methods exhibit systematic over-prediction, often reporting AIS 4+ where only "serious" (AIS 3) injury was found (Bastien et al., 2020). Entropy-based metrics, as tested with APROSYS In-Depth Pedestrian database (70 cases), correlate linearly with ISS (R² ≈ 0.83), but offer extended scale and improved sensitivity to both moderate injuries and patient reserve (Neal-Sturgess, 2010).

Method	Basis	AIS 3+ Delineation	Validation (examples)
OTM/PVP	$$\max_t [\sigma_{\rm vM}(t) \dot\varepsilon_{\rm vM}(t)]$$	Direct; cubic law maps PVP to AIS	3 real collisions, PM match
Entropy sum	$-k \sum_i p_i \ln p_i$, $i \geq 3$	Summed over all AIS 3+	70 pedestrian cases, APROSYS
MPS threshold	Elementwise max principal strain	≥0.21–0.26 signals severe	Systematic over-prediction

7. Research Implications and Current Limitations

The development of methods capable of directly computing AIS 3+ from FE and simulation data bridges the gap between biomechanics and clinical outcomes. OTM with PVP provides a deterministic, direction- and tissue-specific mapping, while entropy-based scores offer a principled aggregation mechanism for complex multi-injury cases.

Current limitations include:

• Bleeding and evolving secondary injury: Pure mechanical models are limited in predicting outcomes where hemorrhage or delayed tissue response dominates.
• Calibration dependency: Both OTM and entropy scores rely on robust empirical mapping of PVP and $p_i$ to clinical outcomes.
• Population heterogeneity: Variability due to genetics, disease, or atypical anatomy is typically only handled statistically, not on a case-by-case basis.

A plausible implication is that, for complete trauma prediction workflows, continued refinement of FE material models—especially those integrating vascular rupture and evolving soft-tissue degradation—will be necessary to further advance AIS 3+ prediction (Bastien et al., 2020, Neal-Sturgess, 2010, Bastien et al., 2019).