Hesham W. Gomma

A Recurrent Neural Network Approach for Predicting Glucose Concentration in Type-1 Diabetic Patients

Estimation of future glucose concentration is important for diabetes management. To develop a model predictive control (MPC) system that measures the glucose concentration and automatically inject the amount of insulin needed to keep the glucose level within its normal range, the accuracy of the predicted glucose level and the longer prediction time are major factors affecting the performance of the control system. The predicted glucose values can be used for early hypoglycemic/hyperglycemic alarms for adjustment of insulin injections or insulin infusion rates of manual or automated pumps. Recent developments in continuous glucose monitoring (CGM) devices open new opportunities for glycemia management of diabetic patients. In this article a new technique, which uses a recurrent neural network (RNN) and data obtained from CGM device, is proposed to predict the future values of the glucose concentration for prediction horizons (PH) of 15, 30, 45, 60 minutes. The results of the proposed technique is evaluated and compared relative to that obtained from a feed forward neural network prediction model (NNM). Our results indicate that, the RNN is better in prediction than the NNM for the relatively long prediction horizons.

Developing a treatment for neurogenic bladder dysfunction using Model Predictive Control (MPC)

Abstract Neurogenic Lower Urinary Tract Dysfunction (NLUTD) which is usually called neurogenic bladder, is a dysfunction of the urinary bladder due to malfunction in the central nervous system or peripheral nerves which involved in the control of micturition (urination). Current treatments vary between medications, surgery and open loop Electrical-stimulatory therapy. This paper presents new a treatment approach for this disease using totally a new approach namely the control engineering. A nonlinear model for the LUTD is developed and controller using the Model Predictive Control (MPC) is designed to compensate for the faulty, weak, or absent control signals while considering the constraints in the nerve control signals. The MPC controller shows a significant ability in controlling the micturition process and bringing the bladder behavior to its normal function pattern while satisfying the nerve signal constraints. Meanwhile, the MPC shows significant robustness and excellent ability to deal with wide range of model uncertainties.

Reducing the bullwhip effect in supply chains using genetic algorithm and control engineering

Supply chains suffer from bullwhip effect which is the amplified variance of the demand information at the order end of the chain. Differently from the majority of research efforts in reducing the bullwhip efforts that consider business modeling approaches and use information technology enhancements, this paper introduces alternative techniques that mainly focus on using the strength of Genetic Algorithm (GA), PI and PID controllers as tools for the bullwhip reduction. The results show the ability of the proposed techniques in providing substantial reduction specially when compared with the current conventional approaches.

Blood glucose regulation using a neural network predictor with a fuzzy logic controller

Current insulin therapy for patients with type 1 diabetes often results in high variability in blood glucose concentration and may cause hyper-and hypoglycemic episodes. Closing the glucose control loop with a fully automated control system improves the quality of life for insulin-dependent patients. This paper presents a nonlinear model predictive control technique for glucose regulation in type 1 diabetic patients. The proposed technique uses a neural network as a nonlinear model for prediction of future glucose values and a fuzzy logic controller FLC to determine the insulin dose required to regulate the blood glucose level, especially after unmeasured meals. In the proposed technique, to avoid errors of meal estimation, the patient is not required to enter any data such as the meal time and size which was, in previous systems, necessary to determine the insulin bolus. The use of neural networks in predicting future glucose levels helps the proposed control strategy to handle delays associated with insulin absorption and time-lag between subcutaneous glucose readings and the plasma glucose level. The FLC uses the predicted glucose values to determine the required insulin bolus. A feed forward neural network FFNN and a recurrent neural network RNN are tested and evaluated as nonlinear glucose prediction models. Simulation results for three meal challenges are demonstrated. our results indicate that, the use of a neural network as a predictor along with a FL controller can decrease the postprandial glucose concentration, avoids hyper glycemia, and dynamically responds to glycemic challenges. The simulation results also indicate that, the use of a RNN in glucose prediction gives better results than the use of a FFNN. The RNN provides much better prediction performance than the FFNN especially at longer prediction horizons.

Development of mathematical model for lower urinary tract dysfunctions

The main objective of this paper is to simulate different types of diseases that cause deformities which occur in the lower urinary tract system together with abnormal micturition , which results from these diseases. To achieve this a modified version of the normal micturition process model proposed by William Fletcher (1) was developed and simulated using MATLAB. The new model is assuming abnormal nerve signals and bladder, ureteral muscles disorders The newly developed model allows to simulate six types of lower urinary tract system (LUTS) disorders caused by dysfunctions in bladder, urethral muscles and nerves control system and gives mathematical representation for abnormal micturition process. Moreover, the simulation developed has got a wide number of advantages in the medical sector such as design intelligent control scheme to correct any deformations that occur in lower urinary tract system, and can be allowed to test the system safely before trial with humans.

Spread spectrum and noise rejection with application to ILC controlled systems

This paper deals with noise effect on control systems. It uses the principle of spread spectrum to reduce the noise effect on deteriorating system performance. The proposed technique can be applied to networked control systems and conventional control systems which is the scope of this paper. Numerical example is introduced to ILC controlled system due to its well known sensitivity to noise measurement to confirm the efficiency of the proposed technique.

Stability analysis for generalized predictive control (GPC) with different classes of uncertain systems

This paper presents stability analysis for the well known generalised predictive control (GPC) when dealing with different classes of uncertain systems namely systems with unmodelled poles, uncertain gains and uncertain poles. The proposed analysis is based on the explicit relation between the system transfer function and the step response coefficients. It reveals part of the mystery behind the stability strength of the GPC.

Stability analysis for generalized predictive control (GPC) and time varying weighting generalized predictive control (TGPC)

Since predictive control algorithms have been introduced, most of them have been known as effective tools for the control of many practical systems. Despite this success, most of them do not have a general stability theory. This paper proposed new stability analysis to two of those algorithms, namely generalized predictive control (GPC) and time varying weighting generalized predictive control (TGPC).

Adaptive Neuro-Fuzzy Inference System Controller Technique for Lower Urinary Tract System Disorders

This article presents an intelligent controller system using hybrid approach of adaptive neuro-fuzzy inference system (LUTS) for control detrusor pressure and urethral pressure by correction disorders that occurs in the LUTS. Six types of LUTS disorders are used to test the adaptive network-based fuzzy inference system controller, which are an absence of bladder contraction caused by the nervous signal disorder, absence of bladder contraction caused by bladder muscle disorder, absence of urethra contraction caused by nervous signal disorder, obstructed urine flow disorder, irregular urine flow disorder, and intermittent urine flow disorder. Simulations were run in MATLAB 7, and the results of our work have demonstrated a very low transient response and a nonoscillating steady-state response with excellent stabilization.

Parasympathetic Nervous Signal Damping Using the Adaptive Neuro-Fuzzy Inference System Method to Control Overactive Bladder

Overactive bladder is a sudden bladder contraction without the patient’s control despite the bladder may contain only a small amount of urine. The overactive bladder is a chronic dysfunction caused primarily by a problem in the nerves that causes an increasing discomfort and burden on the patient. Therefore, the management of overactive bladder is crucial for the health. This study investigated a new controller approach based on Adaptive Neuro-Fuzzy Inference System to damp the parasympathetic nervous signal and furthermore to stimulate the stimulation locations such as the tibial nerve and dorsal penile nerve to enhance the bladder stability of the control continence. The complete simulations are performed in the MATLAB Simulink to provide comprehensive understanding of the issue. Simulation results demonstrated that the developed Adaptive Neuro-Fuzzy Inference System–based controller would be more effective in damping signal oscillations.