Several prior studies focused on the estimation of the probability of Type 2 Diabetes Mellitus (T2DM) complications. None of the references used sophisticated mathematical models to simulate the long-term progression of the disease when estimating the probability of a complication. This study extends an existing T2DM model to predict the probability of Diabetic Retinopathy. The benefit of this approach is that we will need much less information to predict the probability of a complication than in a standalone model, as the progression of diabetes will be estimated with an identifiable model. To expand the model, a detailed qualitative analysis was required first to write the model equation along with relationships that can be supported by the pathogenesis of diabetes and the biological background of the development of complications. Then the collection of data necessary for setting up the model, which in the course of our present work was implemented by quantitative analysis of publicly available data releases. The next step was to write the model and identify it using the available data. Finally, we validated the model using a data set from a data source independent of the previous ones and evaluated the results. In the course of our present work, we focused on the probability of developing Diabetic Retinopathy. The completed model is suitable for supplementing the simulation of the initial model in such a way as to estimate the probability of retinopathy appearing in a patient during the progression of the disease.

Diabetes mellitus is a chronic metabolic disorder requiring meticulous blood glucose regulation to minimize both acute complications and long-term vascular damage. Traditional glucose monitoring approaches—such as finger-prick tests and continuous glucose monitoring (CGM)—primarily support reactive interventions, often falling short in enabling proactive management. This study proposes a deep learning-based predictive framework for blood glucose level estimation using historical CGM data. The model’s performance was evaluated using standard metrics including Root Mean Squared Error (RMSE), Mean Absolute Percentage Error (MAPE), and R-squared (R2) score. Experimental results across multiple patients reveal that the model achieved RMSE values ranging from 19.37 to 28.57, and MAPE values between 7.76 and 13.31. The highest predictive accuracy was observed for Patient 570 (RMSE: 20.38, MAPE: 7.76), while the model struggled with higher variability in Patient 559. These findings demonstrate the model’s potential in delivering personalized, anticipatory glycemic control, thereby supporting more effective diabetes management strategies.

This paper introduces an integrated Hardware-In-the-Loop (HIL) testing framework, combining the UVA/Padova Type 1 Diabetes Simulator with AndroidAPS, an open-source artificial pancreas system. This integration forms a testing environment capable of evaluating insulin regulation algorithms under both virtual and real hardware conditions. The FDA-approved UVA / Padova Simulator models glucose-insulin dynamics and meal digestion. Paired with AndroidAPS, the system can actuate real-world insulin pumps to test insulin delivery control algorithms. The framework is tied together by various REST APIs and uses the Flask framework for efficient data exchange and system connectivity. The HIL approach provides a robust platform for functional and reliability testing of these algorithms. The developed APIs are open-source: https://github.com/OE-Diab/AP-HIL

Automated brain tumor classification is an intensively investigated problem, which recently attracted significant attention. Convolutional neural networks (CNN) and deep learning represent the standard for the foundation of any recent solution. This paper proposes two simplified VGG architectures and investigates their capabilities and limitations, in comparison with state-of-the-art CNN networks deployed via transfer learning. Various parameter settings are involved in the evaluation process, including different kernel sizes, dropout rules, loss functions, etc. Networks are trained and tested on a public brain tumor classification data set consisting of 3064 images and three tumor classes (meningioma, glioma and pituitary tumor). The thorough evaluation process revealed that the proposed CNN models can achieve competitive performances with regard to state-of-the-art methods in several scenarios. The best achieved accuracy benchmarks are 98.2% overall Dice similarity score and correct decision rate, and AUC values over 99.6% for each of the three tumor classes.

Reinforcement learning (RL) has shown promise in controlling blood glucose levels in a personalized way in type 1 diabetic patients. In this study, we investigate the impact of different activation functions and layer numbers on RL performance in blood glucose control. We train RL agents with various combinations of activation functions and layer numbers on a virtual patient model. The RL agents are evaluated based on their ability to maintain blood glucose levels within a target range while minimizing the frequency and magnitude of hypoglycemia and hyperglycemia events. Our results show that the choice of activation function and layer number significantly affects the RL performance. Specifically, the agents with ReLU activation functions and two or three hidden layers outperform the other agents, achieving a higher percentage of time in the target range and fewer hypoglycemia and hyperglycemia events. These findings provide valuable insights for the development of RL-based blood glucose control systems in type 1 diabetic patients.

Despite the increasing availability and reliability of artificial pancreas devices, many people with type 1 diabetes mellitus are still on multiple daily injections therapy consisting of a daily basal insulin injection and mealtime boluses. The use of an insulin bolus advisor may improve glycaemic control as well as reduce the burden of the disease on these patients. This paper investigates the application of an impulsive model predictive controller in a bolus advisor system. The bolus calculator is assessed in closed-loop simulations using different basal insulin scenarios. The simulations show that a model predictive controller can achieve good glycaemic control and may be suitable to give bolus recommendations to people with diabetes. Higher basal insulin levels can result in better times in the target range but also carry a higher risk of hypoglycemia.

Like many physiological systems, the various mathematical models describing the glucose-insulin system can only have non-negative inputs. Thus, the control method — often model predictive control in artificial pancreas systems—must provide a non-negative control signal. Existing solutions include saturation and constrained optimization. In this paper, we propose a dynamic extension of the patient model as a way of ensuring the positivity of the control signal, a method previously applied in tumor growth control. We evaluate the controller in a closed-loop simulation and compare the results with a controller using saturation. Our simulations show that control performance with the extended model can reach or exceed the performance achieved with saturation.

Background:Human workers are indispensable in the human–cyber-physical system in the forthcoming Industry 5.0. As inappropriate work content induces stress and harmful effects on human performance, engineering applications search for a physiological indicator for monitoring the well-being state of workers during work; thus, the work content can be modified accordingly. The primary aim of this study is to assess whether heart rate variability (HRV) can be a valid and reliable indicator of acute work-content-related stress (AWCRS) in real time during industrial work. Second, we aim to provide a broader scope of HRV usage as a stress indicator in this context.Methods:A search was conducted in Scopus, IEEE Xplore, PubMed, and Web of Science between 1 January 2000 and 1 June 2022. Eligible articles are analyzed regarding study design, population, assessment of AWCRS, and its association with HRV.Results:A total of 14 studies met the inclusion criteria. No randomized control trial (RCT) was conducted to assess the association between AWCRS and HRV. Five observational studies were performed. Both AWCRS and HRV were measured in nine further studies, but their associations were not analyzed. Results suggest that HRV does not fully reflect the AWCRS during work, and it is problematic to measure the effect of AWCRS on HRV in the real manufacturing environment. The evidence is insufficient for a reliable conclusion about the HRV diagnostic role as an indicator of human worker status.Conclusion:This review is valuable in the Operator 4.0 paradigm, calling for more trials to validate the use of HRV to measure AWCRS on human workers.

One of the most challenging area of diabetes research is to provide such automated insulin delivery systems – so called artificial pancreas systems – that have robust and adaptive capabilities in a highly sophysticated way. I.e. they are able to provide robust insulin delivery actions at the beginning of the therapy to satisfy the requirements of the patients without knowing the users daily lifestyle and preferences however adaptive on the short-term to learn these patient specifics to increase the quality of the therapy. One possible solution is the closed-loop systems that have self-learning features. In the present study, we have examined a glucose regulatory problem using direct reinforcement learning based controller. The approach represents the fully automatic insulin administration as the timepoint and the carbohydrate content of the meals were unknown and randomized. We constructed a virtual environment of the patient with type 1 diabetes by applying a mathematical model. Proximal policy optimization learning model with continuous action space was used. Furthermore, we evaluated the effect of different training lengths on the test scenario.

Electromyography-based control applications became popular in recent days since they can be implemented using cheap brain-computer interface devices. In the past, for classifying electromyogram data, human experts would extract features from raw signals manually. Unfortunately, this is a time-consuming process and requires extensive domain knowledge. With the advancement of artificial intelligence, such procedures can be done automatically using deep learning architectures. In this study, we designed a deep convolutional network to classify electromyogram signals. Furthermore, a support vector machine-based classifier was also implemented as a reference system. The goal of this study was to develop a control architecture, in which the user interface is equipped with a cheap commercial brain-computer interface system for mobile robot control. We collected a considerable amount of electromyogram signals during different facial expressions that can be used as intervention signals. Finally, we implemented an online real-time control method based on the deep convolutional network. The system enables the user to drive a mobile robot platform solely with electromyogram signals. Manipulating the designed system, after a short initial practice period, the users could navigate through a predefined path with a low error rate.

The effects of individual physiological phenomena play an important role considering the accuracy of artificial pancreas systems. An example of these phenomena is the heart rate which is easy to measure. There is a connection between heart rate and endogenous glucose production that significantly influences the blood glucose level. The proper implementation of heart rate could lead to defining physical activity in TIDM models. The aim of the current study is to examine how the change in endogenous glucose production influences the fitting accuracies in model versions with different complexity. Our extensions include a heart rate dependent endogenous glucose production equation, modeling the effect of physical activity, and a part defining the effect of insulin on endogenous glucose production. The joint effect of the mentioned extensions was also considered.

Linear regression is one of the fundamental tasks of mathematical statistics and machine learning related disciplines. Several techniques have been elaborated, however the most common one is still the application of the least squares technique. In this paper we present an alternative, ”robustified” approach for linear regression that instead of relying on the minimization of the L2-norm, minimizes the P-norm that was introduced by Steiner et. al. in connection with earth sciences and Most Frequent Value (MFV) calculations. Although the proposed alternative serves with a rather robust and outlier-resistant regression, a nonlinear equation system has to be solved in an iterative way. The present study serves with a comparison between the Newton’s and Broyden’s method solving the nonlinear system in the iterative procedure in case of a test example and in case of real life economic data regarding unconditional economic $\beta$ convergence of the EU countries and regions.

Most multi-layer neural networks used in deep learning utilize rectified linear neurons. In our previous papers, we showed that if we want to use the exact same activation function for all the neurons, then the rectified linear function is indeed a reasonable choice. However, preliminary analysis shows that for some applications, it is more advantageous to use different activation functions for different neurons - i.e., select a family of activation functions instead, and select the parameters of activation functions of different neurons during training. Specifically, this was shown for a special family of squashing functions that contain rectified linear neurons as a particular case. In this paper, we explain the empirical success of squashing functions by showing that the formulas describing this family follow from natural symmetry requirements.

In case of physiological systems state and parameter estimation is a crucial question. It is key to describe given patient population with appropriate accuracy. Furthermore, state feedback kind of applications also require some sort of estimation procedure in order to get internal information about the controlled system. Linear parameter varying (LPV) framework is beneficial for controller design as well. However, to realize the necessary scheduling parameters, estimation of both state variables and model parameters is needed. A possible solution is the application of Dual Extended Kalman Filter (DEKF) which is able to estimate these signals. The developed framework can be used to design LPV based controller in our further work. In this study we introduce our developed DEKF solution by using the widely applied Cambridge Type 1 Diabetes Mellitus (TIDM) model for virtual patient generation. We have found that our solution is able to estimate the state variables with good accuracy. The variation of parameters can also be tracked by using the proposed solution.

In natural sciences, especially in life sciences, controller designers frequently meet the problem that though the controlled system is modeled by a set of nonlinearly coupled Ordinary Differential Equations (ODE) containing various independent variables, only a single control input is available by the use of which the propagation of only one variable has to be controlled. Normally the controlled state variable can be observed by direct measurements, while no sensors are available for obtaining information on the propagation of the other ones. Though in the possession of a reliable system model one has good odds to develop state observers, in the practice just the reliable model used to be missing. While the traditional control design methodologies normally need some state estimation, the Fixed Point Iteration-based (FPI) Adaptive Controller was developed to evade this difficulty. In this design instead modeling the effects of the propagation of the various state variables, these effects are directly observed and compensated on this basis. This approach can be used without structural modification if certain effects appear through some time-delay (Delayed Differential Equations - DDE). In many cases simple and effective models can be developed that contain only pure delay effects. In this paper it is shown that a recent DDE model of Diabetes Mellitus can be used in the FPI-based adaptive blood glucose concentration level control even if the available model is very imprecise. This statement is substantiated by numerical simulations.

The advantages of using advanced control techniques related to physiological applications are unquestionable as it was proven in many cases in the recent times. Although, there are several challenges that practitioners need to face. For example, the lack of precise information about the internal state of the patients, i.e. the inter-and intra-patient variabilities which cause uncertainties that need to be tolerated by the applied controllers. In this study an alternative solution is presented for control of tumor growth. Uncertainties and nonlinearities are handled by the applied Linear Parameter Varying (LPV) methodology completed by Tensor Product (TP) model transformation. Linear Matrix Inequalities (LMI) based optimization are used for controller design. The lack of information about the internal state is solved by using Extended Kalman Filter (EKF) to estimate the non-measurable state variables. The developed control structure is able to enforce the controlled system to behave as a predefined reference system. We show that the control framework operates well and reaches the determined aims of the control.

The utilization of modern and advanced control engineering related methods for the control, estimation and assessment of physiological applications is widespread. It is also well-known that this engineering apparatus is executed on digital computers. The current insufficiency of available and accurate discretized models, especially in case of Diabetes Mellitus (DM), provides incentive for this research. The researchers typically approximate the continuous solutions which may not be the best alternative in many cases, in particular considering numerical stability and cost-effectiveness. In this paper we performed an analysis of the available discretization options in order to develop discrete models with a special focus on the Linear Parameter Varying (LPV) systems. LPV techniques are very useful frameworks which allow the application of linear controller, observer and estimator design. In this study, three LPV discretization and two Jacobian based discretization methods are introduced and analyzed to provide a basis for our further investigations in the topic.

In this paper we examine how can be combined the Linear Parameter Varying (LPV) modeling technique with the Linear Matrix Inequality (LMI) based controller and observer design methodology in order to control the tumor growth via anti-angiogenic inhibition. We introduce the important physiological knowledge regard to the control problem together with the design procedure. We used a recently developed minimal model which describes the tumor growth dynamics beside anti-angiogenic inhibition and we transformed this model into the difference based qLPV model. After, LMI based controller and observer were designed by pole clustering LMIs and we realized the control structure. Our aim was to develop a control environment which is - however - advanced, but it can be easily used and provides good performance from the designing properties and the robustization possibilities points of view, respectively. As our results showed, the framework provides appropriate results for tumor control.

The application of the Soft Computing based methods, especially, the Tensor Product (TP) transformation has several beneficial properties from the biological modeling and control point of view, because complex, nonlinear processes can be handled by them effectively. Another advantage of these tools consist on the Linear Parameter Varying (LPV) and Linear Matrix Inequality (LMI) based techniques can be easily connected to them. The aim of this study is to develop TP models, which can describe the tumor growth beside anti-angiogenic treatment. The role of the anti-angiogenic therapies is to decrease the size of the tumor to operable or maintainable level. From control engineering point of view, the treatment process can be formulated as a control task. In this work, we realized two TP models, which approximates the initial transformed model with high accuracy, regardless the kind of input load and without stability problems. The TP models will be used for TP-based controller design on LMI basis.

In case of physiological related researches the appropriate adjustment of the parameters of the mathematical models describing biological phenomenons is a crucial issue. These models are essential in many research field such as the personalized health care or the control of physiological processes. Despite the available identification techniques there is no general solution in those cases where the mathematical model is given, but highly nonlinear in order to capture the main dynamical attitude of the physiological processes to be described. One of our aims was to develop such a general nonlinear identification framework which is flexible, can be easily used and supports the identification of these kind of models. We defined different metrics to measure the performance of the developed system. From the other hand, our goal was to successfully realize the identification framework in case of tumor growth beside anti-angiogenic treatment which is essential in our future work in order to validate the performance of advanced control algorithms. The results show that the nonlinear identification framework performed well in this case, since the predefined requirements from the applied metrics points of view were satisfied in all cases.

In-silico modeling is an important part of biomedical engineering. Advanced controllers providing high quality control can be validated through it checking if the available mathematical model of the given biomedical process produces the desired output. However, due to high patient variability the advanced linear control methods applied on linearized models could produce several distortions compared to the original nonlinear models; hence, these errors should be reduced. Hierarchical control strategies could be a possibility or from modeling point of view using different control-oriented modeling methodologies. Linear Parameter Varying (LPV) approaches with Linear Matrix Inequality (LMI) based modeling and controller design represent one choice. In this paper, we investigate their generalized extension, the Tensor Product (TP) model transformation demonstrated on diabetes modeling. In concrete, the type 1 diabetes modeling on Intensive Care Units (ICU) is envisaged. The achieved results will be used for TP transformation based controller design in our later work.

Tensor Product (TP) transformation based modeling and control can be useful in biomedical engineering, since complex nonlinear control tasks can be handled easier with it. Moreover, the modeling approach can handle the Linear Parameter Varying (LPV) models and produces a tensor based system description, which can be used during Linear Matrix Inequality (LMI) based controller design. The TP property makes the usability of the method beneficial as LMI connected techniques allows using the Lyapunov theorems. The aim of the current work is to demonstrate the usability of TP models in biomedical applications, i.e. diabetes modeling. The core model, the minimal model is investigated and simulation results are presented under Matlab.

Adaptive, model-free control of Type 1 Diabetes Mellitus (T1DM) is a lack in the field of diabetes control, since, most of the applied control strategies are model-based ones. The main problem is that difficult to formulate exact mathematical models to replicate the physiological processes, not just because of their behavior, rather then these processes are changing patient-by-patient. Furthermore, the developed models so far, are highly non-linear and difficult to manage. A possible adaptive control solution can be the recently developed Robust Fixed Point Transformation (RFPT)-based control design method, which can provide control action, based on the observations about the actual output of a controlled system. In this paper we show a survey, how can be used this novel technique related with a known, highorder glucose-insulin model, to investigate the usability according to diabetes control.

In physiological control there is a need to estimate signals that cannot be measured directly. Burdened by measurement noise and unknown disturbances this proves to be challenging, since the models are usually highly nonlinear. Sigma-point filters could represent an adequate choice to overcome this problem. The paper investigates the applicability of several different versions of sigma-point filters for the Artificial Pancreas problem on the widely used Cambridge (Hovorka)-model.