Noninvasive Acute Compartment Syndrome Diagnosis Using Random Forest Machine Learning

TL;DR

The proposed objective and noninvasive ACS diagnostics excelled in key performance metrics, including sensitivity and specificity, with a statistically insignificant performance difference in motion present cases.

Abstract

Acute compartment syndrome (ACS) is an orthopedic emergency, caused by elevated pressure within a muscle compartment, that leads to permanent tissue damage and eventually death. Diagnosis of ACS relies heavily on patient-reported symptoms, a method that is clinically unreliable and often supplemented with invasive intracompartmental pressure measurements that can malfunction in motion settings. This study proposes an objective and noninvasive diagnostic for ACS. The device detects ACS through a random forest machine learning model that uses surrogate pressure readings from force-sensitive resistors (FSRs) placed on the skin. To validate the diagnostic, a data set containing FSR measurements and the corresponding simulated intracompartmental pressure was created for motion and motionless scenarios. The diagnostic achieved up to 98% accuracy. The device excelled in key performance metrics, including sensitivity and specificity, with a statistically insignificant performance difference in motion present cases. Manufactured for 73 USD, our device may be a cost-effective solution. These results demonstrate the potential of noninvasive ACS diagnostics to meet clinical accuracy standards in real world settings.

Figures (7)

• Figure 1: Visual schematic of fundamental electronic components. The force sensitive resistor (FSR) serves as the main way pressure data is collected while the Arduino:micro analyzes the data. The 10k Ohm resistor is used to reduce the amount of power fed into the FSR.
• Figure 2: Cross-section of testing set-up. Acute compartment syndrome (ACS) was simulated artificially using an intravenous (IV) bag, foam and silicone skin. Ground truth pressure is the inflated pressure of the IV bag.
• Figure 3: Single decision tree visualization. Darker shading indicates higher probability for the corresponding class, where blue is a positive detection and red is a negative detection.
• Figure 4: RFC algorithm final classification using decision trees. The model averages predictions across 100 decision trees.
• Figure 5: Comparison of model performance metrics in motion and no motion present scenarios. Higher scores indicate greater performance. Accuracy, precision, sensitivity, and specificity tended to be of stronger values from the range of 0 to 1 in both test cases. Performance during motionless diagnosis was found to be statistically insignificant compared to the motion present case. Ten percent error bars were used to determine significance.