Browsing by Author "Vidanagamachchi, S. M."

Now showing 1 - 3 of 3

Embedded scoring methodology for the self - Assessment of the privacy and security concerns in telemedicine systems in Sri Lanka
(Faculty of Science, University of Kelaniya Sri Lanka, 2023) Siriwardhana, V. T. N. S.; Mallikarachchi, P. S.; Vidanagamachchi, S. M.
Telemedicine, a subset of telehealth, has garnered increasing attention for its potential to transform healthcare delivery. This analysis was undertaken with the goal of developing an embedded scoring system for a self-assessment questionnaire aimed at gauging patients' familiarity with privacy and security concerns associated with telemedicine adoption in Sri Lanka. To achieve this objective, our methodology commenced with an exhaustive review of published research papers pertaining to privacy and security issues within the telemedicine sector and associated scoring mechanisms. Subsequently, we deployed an online questionnaire to gather comprehensive data encompassing a range of scenarios, including privacy, data storage, consent, encryption, authentication, authorisation, and network security. The selection of these categories was rooted in international policies such as HIPAA, adapted to suit the Sri Lankan telemedicine landscape. Upon the formulation of the questionnaire, we employed the Likert scale to quantify responses, enabling us to assess the significance of various dimensions. Data analysis was executed utilising IBM SPSS (Statistical Package for Social Sciences) software, with qualitative inquiries supplemented by predefined response options. Key findings from our study revealed notable gaps in patient awareness and understanding. Over 45% of telemedicine users admitted to not having reviewed the privacy or security policies associated with their telemedicine applications. Similarly, more than 40% of telemedicine system users lacked knowledge of wellestablished privacy and security regulations, including HIPAA, HL7, and GDPR. Additionally, patients exhibited uncertainty regarding the average size of documents or images shared through telemedicine applications. Alarmingly, approximately 50% of patients were unfamiliar with encryption algorithms such as DES, AES, RSA, Blowfish, and Twofish, despite being well-versed in data-recovery techniques and antivirus software usage. This study, conducted with a sample size of 100 respondents, underscores the pervasive limitations in patient understanding of critical aspects related to telemedicine application use. Moreover, it emerged that telemedicine system users often accessed government-blocked, insecure, or unavailable websites within their regions. In response to these findings, we have developed an informative website aimed at enhancing telemedicine users' knowledge by disseminating the insights gleaned from our analysis. In conclusion, the implementation of our embedded scoring method yielded not only high completion rates but also valuable, thoughtful responses. This study underscores the imperative of bolstering patient education and awareness to ensure the secure and responsible adoption of telemedicine in Sri Lanka's healthcare landscape.
Hardware architecture for superoxide production in CKDu initiation
(4th International Research Symposium on Pure and Applied Sciences, Faculty of Science, University of Kelaniya, Sri Lanka, 2019) Upamalika, S. W. A. M.; Vidanagamachchi, S. M.; Wannige, C. T.
Simulation of molecular mechanisms of a disease is important in drug discovery and treatments of the disease. There are many difficulties in performing extensive biological experiments to investigate of a disease with living organisms or even with cell lines because of high cost, the requirement of expensive instrumentation and expertise knowledge. Further, there are lots of ethical and legal limitations for doing these experiments. Modeling these processes using mathematical modeling and simulating them with latest technologies would accompany to further investigate the disease condition while finding and testing new drugs by bypassing limitations. In recent past, usage of hardware simulations for the biological system modeling has become more frequent due to high speed solving capacity of hardware compared to much time consuming process of simulation using software. Field Programmable Gate Array is a semiconductor device designed to build reconfigurable digital circuits and has been used more frequently for simulation and acceleration purposes these days. In this work, we propose a hardware architecture utilizing Field Programmable Gate Array for superoxide production in initiation of Chronic Kidney Disease of uncertain etiology with heavy metal exposure which is one of the identified etiological factors of the disease. Since oxidative stress is identified as main mediator of heavy metal induced renal injury in the disease, mechanisms for increase of oxidative stress is further explored. To that end, the increase of reactive oxygen species initiated first with superoxide generation is identified as a leading cause for increase of oxidative stress. Therefore, the superoxide increment was mathematically expressed using kinetic laws. Ordinary differential equations based mathematical expression is used to describe the variation of superoxide concentration with time in cells. The ordinary differential equations are then can be converted into the hardware description language code which could run on Field Programmable Gate Array. In this conversion, Register Transfer Level design of the superoxide increment process was created based on the mathematical expression as the initial step before the implementation on Field Programmable Gate Array. In the circuit, input signals are the main variables considered in the mathematical model and adders, sub tractors, multipliers, and dividers are the algebraic representations. The constant parameter values are included in the algebraic representation. The rate of the reaction is the final output of this reconfigurable architecture. In addition to acceleration, Field Programmable Gate Array has advantages as optimization and initiating system-on-chip implementations. Accuracy of simulation can be confirmed by observing the experiment data patterns.
Optimizing selfish mining strategies through deep reinforcement learning
(Faculty of Science, University of Kelaniya Sri Lanka, 2024) Wijewardhana, W. T. R. N. D. K.; Vidanagamachchi, S. M.; Arachchilage, N. A. G.
Selfish mining is a type of mining attack where miners strategically release blocks to create forks in the main branch with the intention of acquiring a large portion of the mining reward. Traditional strategies use a Markov Decision Process (MDP) with a non-linear objective function that requires variable blockchain parameters, which are hard to determine, while model-free approaches like multidimensional Q-learning overcome this by learning optimal policies without prior blockchain information. Despite this, existing algorithms remain largely impractical for real blockchain networks, as they fail to account for realistic blockchain features, exhibit inefficient learning in large state spaces, and suffer from slow convergence rates. In this work, we propose a novel model-free Deep Reinforcement Learning (DRL) algorithm for optimal selfish mining, enabling dynamic learning without requiring prior knowledge of network parameters. The study aims to leverage deep neural networks along with advanced exploration and experience replay mechanisms to achieve faster convergence and improved learning efficiency in large state spaces which are inherent in real-world blockchain instances. The non-linearity of the objective function is addressed by incorporating two Double DQNs (DDQNs), one for adversary and one for honest network, which work together to effectively optimize the non-linear objective function. The proposed model is evaluated by constructing a Bitcoin-like Proof-of-Work blockchain simulator which takes into account various real-world blockchain parameters such as stale block rates, propagation delays, and eclipse attacks. Our simulations indicate that the proposed model achieves optimal gains while enhancing the robustness and convergence of the algorithm in large state spaces and dynamically adjusting the mining policy as the blockchain environment evolves.