Abstract
To evaluate the acoustic emissions (AE) and kinematic instability (KI) of the osteoarthritic (OA) knee joints, and to compare these signals to radiographic findings. Sixty-six female and 43 male participants aged 44-67 were recruited. On radiography, joint-space narrowing, osteophytes and Kellgren-Lawrence (KL) grade were evaluated. Based on radiography, 54 subjects (the study group) were diagnosed with radiographic OA (KL-grade ≥ 2) while the remaining 55 subjects (KL-grade < 2) formed the control group. AE and KI were recorded with a custom-made prototype and compared with radiographic findings using area-under-curve (AUC) and independent T-test. Predictive logistic regression models were constructed using leave-one-out cross validation. In females, the parameters reflecting consistency of the AE patterns during specific tasks, KI, BMI and age had a significant statistical difference between the OA and control groups (p = 0.001-0.036). The selected AE signals, KI, age and BMI were used to construct a predictive model for radiographic OA with AUC of 90.3% (95% CI 83.5-97.2%) which showed a statistical improvement of the reference model based on age and BMI, with AUC of 84.2% (95% CI 74.8-93.6%). In males, the predictive model failed to improve the reference model. AE and KI provide complementary information to detect radiographic knee OA in females.
Introduction
Knee Osteoarthritis (OA) is a leading cause of disability worldwide, yet its diagnosis remains a significant clinical bottleneck. The current "gold standard," conventional radiography (X-rays), presents several challenges: it involves radiation exposure, requires expensive equipment, and often correlates poorly with a patient's actual clinical symptoms. Furthermore, X-rays typically detect OA only after significant structural damage has already occurred.
There is a pressing need for low-cost, non-invasive, and objective "digital biomarkers" that can detect OA earlier or monitor its progression in primary care or home settings. This study explores whether Acoustic Emissions (AE)—the sounds generated by joint friction (crepitus)—and Kinematic Instability (KI)—objective measures of joint wobbling—can provide functional insights that complement traditional radiographic findings.
Methods
The researchers developed a custom, 3D-printed wearable apparatus resembling a knee brace to capture joint health data during movement.
The device was equipped with:
- Non-contact air microphones: Positioned on the medial and lateral sides of the knee to capture high-frequency (5–20 kHz) acoustic signals.
- Inertial Measurement Units (IMUs): Accelerometers and gyroscopes placed on the thigh and shin to track movement and stability.
The study included 109 participants (66 female, 43 male) aged 44–67. Participants performed simple tasks, including ten flexion-extension cycles, ten sit-to-stand tests, and one-leg stands. Standard weight-bearing X-rays were used to assign Kellgren–Lawrence (KL) grades (0–4) as the ground truth, with 54 subjects classified as having radiographic OA (KL ≥ 2).
Instead of simply counting joint "clicks," the authors used signal processing to measure the consistency of acoustic patterns. They hypothesized that damaged cartilage produces more erratic, inconsistent sounds. Logistic regression with Leave-One-Out (LOO) cross-validation was used to build predictive models, comparing a "reference model" (Age + BMI) against a "predictive model" (Age + BMI + AE + KI).
Results
The study revealed a significant gender-based difference in the effectiveness of the digital biomarkers.
In female participants, the predictive model achieved an AUC of 90.3%, a significant improvement over the reference model's AUC of 84.2%. Furthermore, the model's predictions in females showed a strong Spearman correlation (0.724) with actual radiographic severity (KL grades). Specific biomarkers, such as the "medial extension ratio" (an AE parameter) and kinematic instability, were identified as the most potent predictors for detecting osteoarthritic changes like joint space narrowing and osteophytes.
Conversely, the model performed poorly in male subjects, with the predictive model (AUC 52.6%) actually performing worse than the reference model (AUC 61.1%). This suggests that the current parameters or sample size were insufficient for the male subgroup.
Discussion and Conclusion
The findings suggest that AE and KI provide information that X-rays cannot capture. While X-rays show structural changes like bone spurs, AE captures the functional state of the cartilage surface, and KI captures joint stability during load-bearing.
However, several limitations remain. The failure of the model in male subjects is a significant hurdle; future research needs larger male cohorts to determine if different features are required. Additionally, the "one-size-fits-all" hardware prototype caused data artifacts in participants with extreme BMIs, indicating a need for adjustable components. Future work should also compare these signals against MRI, which visualizes cartilage directly, and conduct longitudinal studies to determine if these wearable signals can predict future OA progression.
In conclusion, combining acoustic emissions and kinematic instability with traditional risk factors creates a highly accurate predictive model for radiographic knee OA in females, offering a promising path toward non-invasive, functional joint assessment.
Summary updated