MJMarko V. Jankovic

Marko V. Jankovic

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Marko V. Jankovic has a total of 42 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • A Dual Mode Adaptive Basal-Bolus Advisor Based on Reinforcement Learning

    by Qingnan Sun, Marko V. Jankovic, Joao Budzinski, Brett Moore, Peter Diem, Christoph Stettler, Stavroula G. Mougiakakou

    Self-monitoring of blood glucose (SMBG) and continuous glucose monitoring (CGM) are commonly used by type 1 diabetes (T1D) patients to measure glucose concentrations. The proposed adaptive basal-bolus algorithm (ABBA) supports inputs from either SMBG or CGM devices to provide personalised suggestions for the daily basal rate and prandial insulin doses on the basis of the patients' glucose level on the previous day. The ABBA is based on reinforcement learning (RL), a type of artificial intelligence, and was validated in silico with an FDA-accepted population of 100 adults under different realistic scenarios lasting three simulated months. The scenarios involve three main meals and one bedtime snack per day, along with different variabilities and uncertainties for insulin sensitivity, mealtime, carbohydrate amount, and glucose measurement time. The results indicate that the proposed approach achieves comparable performance with CGM or SMBG as input signals, without influencing the total daily insulin dose. The results are a promising indication that AI algorithmic approaches can provide personalised adaptive insulin optimisation and achieve glucose control - independently of the type of glucose monitoring technology.


  • Predicting Blood Glucose with an LSTM and Bi-LSTM Based Deep Neural Network

    by Qingnan Sun, Marko V. Jankovic, Lia Bally, Stavroula G. Mougiakakou

    A deep learning network was used to predict future blood glucose levels, as this can permit diabetes patients to take action before imminent hyperglycaemia and hypoglycaemia. A sequential model with one long-short-term memory (LSTM) layer, one bidirectional LSTM layer and several fully connected layers was used to predict blood glucose levels for different prediction horizons. The method was trained and tested on 26 retrospectively analysed datasets from 20 real patients. The proposed network outperforms the baseline methods in terms of all evaluation criteria.


  • Impact of Errors in Carbohydrate Estimation on Control of Blood Glucose in Type 1 Diabetes

    by Qingnan Sun, Marko V. Jankovic, Stavroula G. Mougiakakou

    This article investigates the impact of carbohydrate (CHO) estimation error on three different algorithms for insulin treatment optimisation. The experiments were conducted using the educational version of the UVa/Padova simulator on 11 virtual adult subjects. Under different CHO estimation error levels, two ways of CHO amount announcements were investigated: numerical value and categorical value (Small, Medium, and Large). Results of experiments suggest that by low CHO estimation error, the way of CHO level announcement has low impact on algorithm quality. As the error increases more intelligent algorithmic approaches need to be investigated.


  • Deep prediction model: The case of online adaptive prediction of subcutaneous glucose

    by Marko V. Jankovic, Samuel Mosimann, Lia Bally, Christoph Stettler, Stavroula Mougiakakou

    In this paper, we propose the concept of the deep prediction model for subcutaneous glucose concentration. The concept is based on several layers of prediction models. One aim of this approach is to eliminate time lag, which is more severe in longer prediction horizons. Thus, the prediction accuracy of the algorithm might be increased, even for longer prediction horizons. The second goal is to create new, potentially good predictors that could be obtained by combining existing predictors. The effectiveness of the proposed model is illustrated in several examples of two-layer networks. In the first layer, a specific linear/non-linear prediction model is used. In the second (correction) layer, an extreme learning machine is used, due to its rapid learning capabilities. In almost all experiments, the proposed method has reduced the time lag and improved the accuracy of the method.


  • The Detection of Series Arc Fault in Photovoltaic Systems Based on the Arc Current Entropy

    by Nikola L. Georgijevic, Marko V. Jankovic, Srdjan Srdic, Zoran Radakovic

    This paper presents a series arc-fault detection algorithm for photovoltaic (PV) systems, which relies on the quantum probability model theory. The algorithm determines the presence of the arc by calculating the modified Tsallis entropy of the PV panel current. Based on the calculated entropy, the algorithm is able to differentiate an arc state (when the current variations are chaotic) from a no-arc state (when the current variations are ordered). The proposed algorithm enables arc-fault detection on a plug-and-play principle, requiring no prior information about the PV system in which it operates. The operation of the algorithm was first simulated on the prerecorded data from the system with the PV panel simulator and the commercial PV inverter. A laboratory prototype of the detector was then built and tested in a real 1.6-kW grid-connected PV system with different PV inverter, without any additional parameter adjustments. Both the sensitivity and the robustness of the detector were confirmed. The tests have shown that both the sustained series arcs and the small sparking series arcs were successfully detected, and there was no false detection due to the MPP tracker operation, PV current step change, or inverter turn on transient.


  • Applications of probabilistic model based on main quantum mechanics concepts

    by Marko V. Jankovic, Tomislav Gajic, Branimir D. Reljin

    Recently, the several applications of the probabilistic model based on two of the main concepts in quantum physics - a density matrix and the Born rule, have been introduced. It was shown that the model can be suitable for the modeling of learning algorithms in biologically plausible artificial neural networks framework, like it is the case of on-line learning algorithms for Independent /Principal/Minor Component Analysis, which could be realized on parallel hardware based on very simple computational units. Also, it has been shown that the quantum entropy of the system, related to that model, can be successfully used in the problems like change point detection, with some examples of applications in the area of power electronics and general classification problems. Here, we present a robust on-line Principal Component Algorithm based on the proposed model, which extracts several principal components simultaneously. Also, we will show usefulness of the proposed method in a simple example of image segmentation.

  • Applications of probabilistic model based on joystick probability selector

    by Marko V. Jankovic, Nikola Georgijevic

    Recently, it has been shown that a probabilistic model based on two of the main concepts in quantum physics - a density matrix and the Born rule, can be suitable for the modeling of learning algorithms in biologically plausible artificial neural networks framework. It has been shown that the proposed probabilistic interpretation is suitable for modeling on-line learning algorithms for Independent /Principal/Minor Component Analysis, which could be realized on parallel hardware based on very simple computational units. Also, it has been shown that the quantum entropy of the system, related to that model, can be successfully used in the problems like change point detection. Here, it will be shown that the proposed model can be successfully used in other areas of applied signal processing, with some examples of applications in the area of power electronics and general classification problems.


  • Unsupervised neural networks for principal component analysis

    by Marko Jankovic

    Janković Marko V, Doktorska disertacija, Univerzitet u Beogradu, Elektrotehnički fakultet, Beograd 2006, Odbranjena 23.03.2006., Jezik srpski - latinica

  • Open Access

    Analysis of possibilities of early diagnostics criteria for Parkinson's disease based on analysis of the input-output curve

    by Jankovic Marko, Kacar Aleksandra

    In this paper, we analyze the possibilities of the diagnosis of Parkinson's disease at an early stage, based on characteristics of the input-output curve. The input-output (IO) curve was analyzed in two ways: we analyzed the gain of the curve for low-level transcranial stimulation and we analyzed the overall 'quality' of the IO curve. The 'quality' of the curve calculation is based on basic concepts from quantum mechanics and calculation of Tsallis entropy.

  • Probabilistic model for minor component analysis based on born rule

    by Marko V. Jankovic, Milos Manic, Branimir D. Relijn

    Minor component analysis (MCA) is commonly applied technique for data analysis and processing, e.g. compression or clustering. In this paper we propose a probabilistic MCA model based on the Born rule. In off-line realization it can be seen as a successive optimization problem. In the on-line realization it will be solved by introduction of two different time scales. It will be shown that recently proposed time oriented hierarchical method, can be used as a concept for on-line realization of the proposed algorithms. The proposed model gives general framework for creating different MCA realizations/algorithms. A particular realization can optimize locality of calculation, convergence speed, preciseness or some other parameter of interest.