Evaluation of a model for glycemic prediction in critically ill surgical patients

Abstract

We evaluated a neural network model for prediction of glucose in critically ill trauma and post-operative cardiothoracic surgical patients. A prospective, feasibility trial evaluating a continuous glucose-monitoring device was performed. After institutional review board approval, clinical data from all consenting surgical intensive care unit patients were converted to an electronic format using novel software. This data was utilized to develop and train a neural network model for real-time prediction of serum glucose concentration implementing a prediction horizon of 75 minutes. Glycemic data from 19 patients were used to "train" the neural network model. Subsequent real-time simulated testing was performed in 5 patients to whom the neural network model was naive. Performance of the model was evaluated by calculating the mean absolute difference percent (MAD%), Clarke Error Grid Analysis, and calculation of the percent of hypoglycemic (≤70 mg/dL), normoglycemic (>70 and <150 mg/dL), and hyperglycemic (≥150 mg/dL) values accurately predicted by the model; 9,405 data points were analyzed. The models successfully predicted trends in glucose in the 5 test patients. Clark Error Grid Analysis indicated that 100.0% of predictions were clinically acceptable with 87.3% and 12.7% of predicted values falling within regions A and B of the error grid respectively. Overall model error (MAD%) was 9.0% with respect to actual continuous glucose modeling data. Our model successfully predicted 96.7% and 53.6% of the normo- and hyperglycemic values respectively. No hypoglycemic events occurred in these patients. Use of neural network models for real-time prediction of glucose in the surgical intensive care unit setting offers healthcare providers potentially useful information which could facilitate optimization of glycemic control, patient safety, and improved care. Similar models can be implemented across a wider scale of biomedical variables to offer real-time optimization, training, and adaptation that increase predictive accuracy and performance of therapies.

Figure 1. The main menu of the developed Electronic Clinical Intensive Data-Logger (eCIDL).

This main menu contains buttons that link the user to various interfaces which contain text fields and drop-down menus to log all medical records present in the comprehensive intensive care unit medical record. This software application was utilized to convert paper-based medical records into electronic records suitable for direct neural network model utilization.

Figure 2. Neural network model design.

The feed-forward neural network design implemented for real-time prediction of glucose. Error (mean squared error) is calculated between neural network output and desired response (actual continuous glucose monitoring values). This error is back propagated to each layer in the neural network architecture and a gradient descent with momentum algorithm is implemented to determine optimal weight values to minimize model error.

Figure 3. Glycemic predictions generated by neural network model.

Figure 4. Clarke Error Grid Predictions demonstrating 97.5% clinically acceptable predictive values.