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

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.

Neural network-based real-time prediction of glucose in patients with insulin-dependent diabetes.

BACKGROUND: Continuous glucose monitoring (CGM) technologies report measurements of interstitial glucose concentration every 5 min. CGM technologies have the potential to be utilized for prediction of prospective glucose concentrations with subsequent optimization of glycemic control. This article outlines a feed-forward neural network model (NNM) utilized for real-time prediction of glucose. METHODS: A feed-forward NNM was designed for real-time prediction of glucose in patients with diabetes implementing a prediction horizon of 75 min. Inputs to the NNM included CGM values, insulin dosages, metered glucose values, nutritional intake, lifestyle, and emotional factors. Performance of the NNM was assessed in 10 patients not included in the model training set. RESULTS: The NNM had a root mean squared error of 43.9 mg/dL and a mean absolute difference percentage of 22.1. The NNM routinely overestimates hypoglycemic extremes, which can be attributed to the limited number of hypoglycemic reactions in the model training set. The model predicts 88.6% of normal glucose concentrations (> 70 and < 180 mg/dL), 72.6% of hyperglycemia (≥ 180 mg/dL), and 2.1% of hypoglycemia (≤ 70 mg/dL). Clarke Error Grid Analysis of model predictions indicated that 92.3% of predictions could be regarded as clinically acceptable and not leading to adverse therapeutic direction. Of these predicted values, 62.3% and 30.0% were located within Zones A and B, respectively, of the error grid. CONCLUSIONS: Real-time prediction of glucose via the proposed NNM may provide a means of intelligent therapeutic guidance and direction.

Development of a neural network model for predicting glucose levels in a surgical critical care setting.

Development of neural network models for the prediction of glucose levels in critically ill patients through the application of continuous glucose monitoring may provide enhanced patient outcomes. Here we demonstrate the utilization of a predictive model in real-time bedside monitoring. Such modeling may provide intelligent/directed therapy recommendations, guidance, and ultimately automation, in the near future as a means of providing optimal patient safety and care in the provision of insulin drips to prevent hyperglycemia and hypoglycemia.

Intensivist supervision of resident-placed central venous catheters decreases the incidence of catheter-related blood stream infections.

Catheter-related blood stream infections (CRBSI) cause significant morbidity and mortality. A retrospective study of a performance improvement project in our teaching hospital's surgical intensive care unit (SICU) showed that intensivist supervision was important in reinforcing maximal sterile barriers (MSB) use during the placement of a central venous catheter (CVC) in the prevention of CRBSI. A historical control period, 1 January 2001-31 December 2003, was established for comparison. From 1 January 2003-31 December 2007, MSB use for central venous line placement was mandated for all operators. However, in 2003 there was no intensivist supervision of CVC placements in the SICU. The use of MSB alone did not cause a significant change in the CRBSI rate in the first year of the project, but close supervision by an intensivist in years 2004-2007, in conjunction with MSB use, demonstrated a significant drop in the CRBSI rate when compared to the years before intensivist supervision (2001-2003), p < .0001. A time series analysis comparing monthly rates of CRBSI (2001-2007) also revealed a significant downward trend, p = .028. Additionally, in the first year of the mandated MSB use (2003), 85 independently observed resident-placed CVCs demonstrated that breaks in sterile technique (34/85), as compared those placements that had no breaks in technique (51/85), had more CRBSI, 6/34 (17.6%) vs. 1/51 (1.9%), p < .01. Interventions to reduce CRBSI in our SICU needed emphasis on adequate supervision of trainees in CVC placement, in addition to use of MSB, to effect lower CRBSI rates.

Implementation of the "FASTHUG" concept decreases the incidence of ventilator-associated pneumonia in a surgical intensive care unit.

BACKGROUND: Ventilator-associated pneumonia (VAP) is a leading cause of morbidity and mortality in critically ill patients. The Institute for Healthcare Improvement 100,000 Lives Campaign made VAP a target of prevention and performance improvement. Additionally, the Joint Commission on Accreditation of Health Organizations' 2007 Disease Specific National Patient Safety Goals included the reduction of healthcare-associated infections. We report implementation of a performance improvement project that dramatically reduced our VAP rate that had exceeded the 90th percentile nationally. METHODS: From 1 January 2004 to 31 December 2005 a performance improvement project was undertaken to decrease our critical care unit VAP rate. In year one (2004) procedural interventions were highlighted: aggressive oral care, early extubation, management of soiled or malfunctioning respiratory equipment, hand washing surveillance, and maximal sterile barrier precautions. In year two (2005) an evaluative concept called FASTHUG (daily evaluation of patients' feeding, analgesia, sedation, thromboembolic prophylaxis, elevation of the head of the bed, ulcer prophylaxis, and glucose control) was implemented. To determine the long-term effectiveness of such an intervention a historical control period (2003) and the procedural intervention period of 2004, i.e., the pre-FASTHUG period (months 1-24) were compared with an extended post-FASTHUG period (months 25-54). RESULTS: The 2003 surgical intensive care VAP rate of 19.3/1000 ventilator-days served as a historical control. Procedural interventions in 2004 were not effective in reducing VAP, p = 0.62. However, implementation of FASTHUG in 2005, directed by a critical care team, resulted in a rate of 7.3/1000 ventilator-days, p