Alarm Reduction Research Articles

Optimized Deep Feature Analysis for Enhanced Botnet Attack Prediction in IoT Networks

The Optimized Deep Feature Analysis (ODFA) framework targets botnet attacks in IoT networks by combining deep learning with machine learning techniques. It uses a hybrid CNN-LSTM model to extract spatial and temporal features from network traffic, applying Recursive Feature Elimination (RFE) with Cross-Validation (CV) for refining feature selection. A Random Forest classifier is then employed to classify different types of botnet attacks. Tests conducted on the UNSW-NB15 dataset demonstrate its ability to achieve high precision, sensitivity, specificity, and accuracy. Recurrent feature optimization allows the framework to adapt to changes in attack patterns, ensuring consistent detection performance. The reduction in false alarms and improved classification of attack types highlights its contribution to securing IoT networks.

RESILIENCE OF UKRAINIAN BUSINESS: HOW TO SUPPORT PRODUCTION AND INNOVATIONS IN THE CONDITIONS OF WAR

The article evaluates the main performance indicators of business entities of Ukraine for 2012-2023. Previous studies, based on statistical information until 2021 inclusive, made it possible to draw a conclusion about the ability of business entities to expand reproduction. In addition, it was determined that despite the low specific weight of enterprises employed in industry, their potential is significant. An assessment of the last two years has shown an alarming reduction in jobs. The rate of decline in the number of employed workers is significantly higher than the rate of decrease in the number of employed workers. There is a decrease in the ability to recover - the destruction of the margin of safety of natural persons - entrepreneurs and a significant drop in the physical production volumes of enterprises. The conducted research made it possible to draw conclusions about the highest degree of resilience of the labor potential of our country - among self-employed workers, who at the same time create jobs for employees. The creation of jobs directly and indirectly affects the increase in the solvent demand of the population, which in turn will stimulate the development of the country's economy. The possibility of the exit of Ukrainian enterprises from the state of crisis, recovery and further economic growth is the introduction of technological innovations and innovative approaches to the organization of business processes, which will allow creating conditions for the formation of new diversified sources to increase the country's competitiveness, the basis of which should be human capital, knowledge and results of scientific research, including the creation and active implementation of artificial intelligence and digitization processes. Their effectiveness is evidenced by modern military developments, and the possibility of technology transfer, their implementation with the possibility of entering world markets and attracting investors for the post-war reconstruction of the country will contribute to increasing the resilience of domestic participants in economic relations. Solving issues of finding ways to ensure financing of operational and investment activities of business entities, achieving their stability and competitiveness – determines the state of protection of national interests, which forms resilience - the ability to "hold a blow", the ability to self-reproduce, the formation of a stress-resistant recovery strategy, innovative development and economic growth.

Time-Series Deep Learning and Conformal Prediction for Improved Sepsis Diagnosis in Non-ICU Hospitalized Patients.

Sepsis, a life-threatening condition from an uncontrolled immune response to infection, is a leading cause of in-hospital mortality. Early detection is crucial, yet traditional diagnostic methods, like SIRS and SOFA, often fail to identify sepsis in non-ICU settings where monitoring is less frequent. Recent machine learning (ML) models offer new possibilities but lack generalizability and suffer from high false alarm rates. We developed a deep learning (DL) model tailored for non-ICU environments, using MIMIC-IV data with a conformal prediction framework to handle uncertainty. The model was trained on 83,813 patients and validated with the eICU-CRD dataset to test performance across hospital settings. Our model predicted sepsis at 24, 12, and 6 h before onset, achieving AUROCs of 0.96, 0.98, and 0.99, respectively. The conformal approach reduced false positives and improved specificity. External validation confirmed similar performance, with a 57% reduction in false alarms at the 6 h window, supporting practical use in low-monitoring environments. This DL-based model enables accurate, early sepsis prediction with minimal data, addressing key clinical challenges and potentially improving resource allocation in hospital settings by reducing unnecessary ICU admissions and enhancing timely interventions.

Bio-Cultural Diversity for Food Security: Traditional Wild Food Plants and Their Folk Cuisine in Lakki Marwat, Northwestern Pakistan

Ethnobotanical studies on foraging are essential for documenting neglected or previously unknown wild food plants, which may be crucial for promoting the diversification of food sources and contributing to food security and sovereignty. The Pashtuns of the Marwat tribe in NW Pakistan are renowned for their traditional customs and food systems. Studying the wild food plants (WFPs) and their associated bio-cultural diversity is quintessential for fostering food security and sovereignty in the region. The research presented here investigated the area’s wild food plants traditionally gathered and consumed. The field survey was conducted in 2023 with 87 study participants. A total of 41 plant species belonging to 24 botanical families was documented. The findings include food uses for Atriplex tatarica, Amaranthus graecizans, and Beta vulgaris subsp. maritima that have rarely been recorded in Pakistan. Moreover, the use of Citrulus colocynthus fruits in jam and Zygophyllum indicum leaves and stems in beverages are novel contributions to the gastronomy of NW Pakistan. The comparison with other food ethnobotanical studies conducted in North Pakistan suggests some similarities between the Lakki Marwat traditional WFPs and those from other semi-arid areas in North Pakistan, both Pashtun and non-Pashtun. While the findings underline the significant role of WFPs in local cuisine, we observed that this local knowledge is also threatened: the rapid spread of fast and industrialized food, modernization, and cultural dilution has led to an alarming reduction in these practices among the younger generations. Therefore, suitable measures to safeguard traditional plants, food knowledge, practices, and the associated culture are urgently needed. The urgency of this situation cannot be overstated, and it is crucial that we act now. Furthermore, preserving wild food plant-related cultural heritage may be fundamental to promoting food security and public health.

Hybrid Classifier for Detecting Zero-Day Attacks on IoT Networks

Recently, Internet of Things (IoT) networks have been exposed to many electronic attacks, giving rise to concerns about the security of these networks, where their weaknesses and gaps can be exploited to access or steal data. These networks are threatened by several cyberattacks, one of which is the zero-day distributed denial-of-service (DDoS) attack, which is considered one of the dangerous attacks targeting network security. As such, it is necessary to find smart solutions to address such attacks swiftly. To address these attacks, this research proposed a hybrid IDS to detect cyber-attacks on IoT networks via machine learning (ML) algorithms, namely, XGBoost, K-nearest neighbors, and stochastic gradient descent (SGD), while classifiers are combined via an ML ensemble. Grid search CV was used to find the best hyperparameters for each classifier at each classification stage. Random projection was used to select the relevant features for training the model. In the evaluation and performance testing phase of the model, two cybersecurity datasets (CIC-IDS2017 and CIC-DDoS2019) were used to test the efficiency of the model in detecting zero-day threats. The best results were obtained for the CIC-DDoS2019 dataset, where 20 features out of the total selection were used. The model was able to achieve an accuracy of 99.91% and an intrusion detection time of 0.22 seconds. The confusion matrix results also revealed a reduction in false alarms. The results and their comparison with those of recent relevant studies demonstrated the effectiveness of the hybrid model in securing IoT networks from zero-day attacks as well as its superiority in terms of accuracy and intrusion detection time. This study is an important step in enhancing security in the IoT environment by presenting a new hybrid model that is capable of dealing with zero-day attacks that are difficult to detect with traditional models.

Evading Cyber-Attacks on Hadoop Ecosystem: A Novel Machine Learning-Based Security-Centric Approach towards Big Data Cloud

The growing industry and its complex and large information sets require Big Data (BD) technology and its open-source frameworks (Apache Hadoop) to (1) collect, (2) analyze, and (3) process the information. This information usually ranges in size from gigabytes to petabytes of data. However, processing this data involves web consoles and communication channels which are prone to intrusion from hackers. To resolve this issue, a novel machine learning (ML)-based security-centric approach has been proposed to evade cyber-attacks on the Hadoop ecosystem while considering the complexity of Big Data in Cloud (BDC). An Apache Hadoop-based management interface “Ambari” was implemented to address the variation and distinguish between attacks and activities. The analyzed experimental results show that the proposed scheme effectively (1) blocked the interface communication and retrieved the performance measured data from (2) the Ambari-based virtual machine (VM) and (3) BDC hypervisor. Moreover, the proposed architecture was able to provide a reduction in false alarms as well as cyber-attack detection.

Variations in Alarm Burden, Source, and Cause Across Inpatient Units at a Children's Hospital.

Alarms at hospitals are frequent and can lead to alarm fatigue posing patient safety risks. We aimed to describe alarm burden over a 1-year period and explored variations in alarm rates stratified by unit type, alarm source, and cause. A retrospective study of inpatient alarm and patient census data at 1 children's hospital from January 1, 2019, to December 31, 2019, including 8 inpatient units: 6 medical/surgical unit (MSU), 1 PICU, and 1 NICU. Rates of alarms per patient day (appd) were calculated overall and by unit type, alarm source, and cause. Poisson regression was used for comparisons. There were 7 934 997 alarms over 84 077 patient days (94.4 appd). Significant differences in alarm rates existed across inpatient unit types (MSU 81.3 appd, PICU 90.7, NICU 117.5). Pulse oximetry (POx) probes were the alarm source with highest rate, followed by cardiorespiratory leads (54.4 appd versus 31). PICU had lowest rate of POx alarms (33.3 appd, MSU 37.6, NICU 92.6), whereas NICU had lowest rate of cardiorespiratory lead alarms (16.2 appd, MSU 40.1, PICU 31.4). Alarms stratified by cause displayed variation across unit types where "low oxygen saturation" had the highest overall rate, followed by "technical" alarms (43.4 appp versus 16.3). ICUs had higher rates of low oxygenation saturation alarms, but lower rates of technical alarms than MSUs. Clinical alarms are frequent and vary across unit types, sources, and causes. Unit-level alarm rates and frequent alarm sources (eg, POx) should be considered when implementing alarm reduction strategies.

Machine learning model with output correction: Towards reliable bradycardia detection in neonates

Bradycardia is a commonly occurring condition in premature infants, often causing serious consequences and cardiovascular complications. Reliable and accurate detection of bradycardia events is pivotal for timely intervention and effective treatment. Excessive false alarms pose a critical problem in bradycardia event detection, eroding trust in machine learning (ML)-based clinical decision support tools designed for such detection. This could result in disregarding the algorithm's accurate recommendations and disrupting workflows, potentially compromising the quality of patient care. This article introduces an ML-based approach incorporating an output correction element, designed to minimise false alarms. The approach has been applied to bradycardia detection in preterm infants. We applied five ML-based autoencoder techniques, using recurrent neural network (RNN), long-short-term memory (LSTM), gated recurrent unit (GRU), 1D convolutional neural network (1D CNN), and a combination of 1D CNN and LSTM. The analysis is performed on ∼440 hours of real-time preterm infant data. The proposed approach achieved 0.978, 0.73, 0.992, 0.671 and 0.007 in AUC-ROC, AUC-PRC, recall, F1 score, and false positive rate (FPR) respectively and a false alarms reduction of 36% when compared with methods without the correction approach. This study underscores the imperative of cultivating solutions that alleviate alarm fatigue and encourage active engagement among healthcare professionals.

DIAmante TESS AutoRegressive Planet Search (DTARPS). II. Hundreds of New TESS Candidate Exoplanets

The DIAmante TESS AutoRegressive Planet Search for the southern ecliptic hemisphere (DTARPS-S) project seeks to identify photometric transiting planets from 976,814 southern hemisphere stars observed in Year 1 of the TESS mission. This paper follows the methodology developed by Melton et al. (Paper I) using light curves extracted and preprocessed by the DIAmante project. Paper I emerged with a list of 7377 light curves with statistical properties characteristic of transiting planets but dominated by false alarms and false positives. Here a multistage vetting procedure is applied including: centroid motion and crowding metrics, false alarm and false positive reduction, photometric binary elimination, and ephemeris match removal. The vetting produces a catalog of 462 DTARPS-S candidates across the southern ecliptic hemisphere and 310 objects in a spatially incomplete Galactic plane list. 58% were not previously identified as transiting systems. Candidates are flagged for possible blending from nearby stars based on Zwicky Transient Facility data and for possible radial velocity variations based on Gaia satellite data. Orbital periods and planetary radii are refined using astrophysical modeling; the resulting parameters closely match published values for confirmed planets. The DTARPS-S population and astrophysical properties are discussed in Paper III.

AI-Based Energy Transportation Safety: Pipeline Radial Threat Estimation Using Intelligent Sensing System

The application of artificial intelligence technology has greatly enhanced and fortified the safety of energy pipelines, particularly in safeguarding against external threats. The predominant methods involve the integration of intelligent sensors to detect external vibration, enabling the identification of event types and locations, thereby replacing manual detection methods. However, practical implementation has exposed a limitation in current methods - their constrained ability to accurately discern the spatial dimensions of external signals, which complicates the authentication of threat events. Our research endeavors to overcome the above issues by harnessing deep learning techniques to achieve a more fine-grained recognition and localization process. This refinement is crucial in effectively identifying genuine threats to pipelines, thus enhancing the safety of energy transportation. This paper proposes a radial threat estimation method for energy pipelines based on distributed optical fiber sensing technology. Specifically, we introduce a continuous multi-view and multi-domain feature fusion methodology to extract comprehensive signal features and construct a threat estimation and recognition network. The utilization of collected acoustic signal data is optimized, and the underlying principle is elucidated. Moreover, we incorporate the concept of transfer learning through a pre-trained model, enhancing both recognition accuracy and training efficiency. Empirical evidence gathered from real-world scenarios underscores the efficacy of our method, notably in its substantial reduction of false alarms and remarkable gains in recognition accuracy. More generally, our method exhibits versatility and can be extrapolated to a broader spectrum of recognition tasks and scenarios.

Fatigue and vigilance in medical experts detecting breast cancer

An abundance of laboratory-based experiments has described a vigilance decrement of reducing accuracy to detect targets with time on task, but there are few real-world studies, none of which have previously controlled the environment to control for bias. We describe accuracy in clinical practice for 360 experts who examined >1 million women's mammograms for signs of cancer, whilst controlling for potential biases. The vigilance decrement pattern was not observed. Instead, test accuracy improved over time, through a reduction in false alarms and an increase in speed, with no significant change in sensitivity. The multiple-decision model explains why experts miss targets in low prevalence settings through a change in decision threshold and search quit threshold and propose it should be adapted to explain these observed patterns of accuracy with time on task. What is typically thought of as standard and robust research findings in controlled laboratory settings may not directly apply to real-world environments and instead large, controlled studies in relevant environments are needed.

Improving Safety Culture by Establishing Alarm Fatigue Performance in The Adult Intensive Care Unit, Dammam Hospital

Alarm fatigue is becoming a more significant global public health issue that compromises patient safety. Alarm fatigue is especially common among critical care nurses, who are the frontline healthcare providers that provide patients with the most direct care and round-the-clock patient monitoring. An excessive number of alerts, together with erroneous and ineffective alerts, can cause alarm fatigue and negatively impact the psychological well-being of nurses and patients during their recovery process. To find a way to lower the number of false alarms and assess the effectiveness of the measures put in place, a quantitative, descriptive cross-sectional approach was employed. In the adult ICU at NGHA Dammam, this pilot project was carried out from June 2019 to November 2020. In order to identify true alarms and false alarms, the Central Monitor Nurse Daily Checklist was utilized as a data collection tool in conjunction with education training and departmental policies and procedures. There are about 820 alerts per patient each day, including 104 telemetry and 716 bedside alarms. Three months into the project, the project's top alarm measurement was 1503 alarms per patient per day. The project has shown a constant trend in alarm declinations after this peak, largely as a result of project improvement activities. The lowest recorded alarm count in November 2020 was 226 alerts, a 72% decrease from the initial 820 alarms. Nonetheless, the declining trend over the previous three months showed a 22%, 42%, and 48% drop in the overall number of alarms. This study found that alarm fatigue, which is influenced by a number of variables, including safety culture practices, was more common among critical care nurses working in the Intensive Care Unit at NGHA Dammam. Overall, during the first two months of its deployment, this study effectively met its aim and had a considerable impact on the reduction of "false alarms". The results of this study, taken together with improved psychological health for nurses and the provision of a therapeutic environment for patients, offer crucial direction for future intervention programs aimed at reducing alarm fatigue among critical care nurses. Future studies should concentrate on the traits of adherence and management, as well as the obstacles that prevent ICU staff from continuing to implement interventions and false alarm reduction strategies.

Real-time disruption prediction in multi-dimensional spaces leveraging diagnostic information not available at execution time

This article describes the use of privileged information to train supervised classifiers, applied for the first time to the prediction of disruptions in tokamaks. The objective consists of making predictions with real-time signals during the discharges (as usual) but after training the predictor also with any kind of data at training time that is not available during discharge execution. The latter kind of data is known as privileged information. Taking into account the limited number of foreseen real time signals for disruption prediction at the beginning of operation in JT-60SA, a predictor with a line integrated density signal and the mode lock signal as privileged information has been developed and tested with 1437 JET discharges. The success rate with positive warning time has been improved from 45.24% to 90.48% and the tardy detection rate has diminished from 50% to 8.33%. The use of privileged information in an adaptive way also provides a remarkable reduction of false alarms from 11.53% to 1.15%. The potential of the methodology, exemplified with data relevant to the beginning of JT-60SA operation, is absolutely general and can be applied to any combination of diagnostic signals.

Improving Telemetry use in the Acute Assessment Unit.

Despite published guidelines, telemetry use is inappropriate in 25-43% of cases. This impacts patient safety and telemetry effectiveness. QI methodology was used to review telemetry in a hospital acute medical unit with the aim of reducing inappropriate use and addressing alarm fatigue. A 'Telemetry Indication Form' was created. Eight weeks of baseline data was collated before introducing the 'Indication Form'. Four plan-do-study-act cycles were conducted. At each cycle, data was analysed using statistical process control charts. Inappropriate telemetry use significantly reduced from 32% to 4%. Total telemetry use also fell. Unfortunately, interventions to address alarm rates did not result in significant reduction in false alarms. A 'Telemetry Indication Form' has significant potential to improve patient safety through reducing inappropriate use.

Machine learning in concept drift detection using statistical measures

ABSTRACT In the data stream, the data has non-stationary quality because of continual and inconsistent change. This change is represented as the concept drift in the classifying process of the streaming data. Representing this data drift concept in data stream mining requires pre-labeled samples. However, labeling samples in real-time streaming (online) is not feasible due to resource utilization and time constraints. Therefore, this paper proposes the concept of Probabilistic Concept Drift Detection (PCDD) in the group classifier. PCDD relies on the data stream classification process and provides concept drift without labeled samples. The PCDD model is evaluated through an empirical study on a dataset called Poker-Hand. The study results show a high concept drift detection rate and a significant reduction in false alarms and missed detections compared to contemporary models. Hence, the results of the experimental study prove the accuracy and scalability of the PCDD model.

Multiple views in single‐path synthetic aperture sonar for mine counter‐measures classification and pre‐identification

AbstractThe authors present exploitation of Multiple Views in Single‐Path (MVSP) Synthetic Aperture Sonar (SAS). The authors demonstrate the interest of MVSP SAS for underwater Mine Counter‐Measures (MCM) with unmanned vehicles, to comply with requirements for performances of detection and classification and time constraints. The authors here lay the emphasis on exploitation in reduction of false alarms of classification by alleviating the ambiguity between mines and natural bottom objects. Finally, the authors show that MVSP SAS is offering a pre‐identification tool with high impact on MCM process duration.

Optimization of rainfall thresholds for landslide early warning through false alarm reduction and a multi-source validation

AbstractThis study proposes an innovative approach to develop a regional-scale landslide forecasting model based on rainfall thresholds optimized for operational early warning. In particular, it addresses two main issues that usually hinder the operational implementation of this kind of models: (i) the excessive number of false alarms, resulting in civil protection system activation without any real need, and (ii) the validation procedure, usually performed over periods too short to guarantee model reliability. To overcome these limitations, several techniques for reducing the number of false alarms were applied in this study, and a multiple validation phase was conducted using data from different sources. An intensity-duration threshold system for each of the five alert zones composing the Liguria region (Italy) was identified using a semiautomatic procedure called MaCumBA, considering three levels of criticality: low, moderate, and high. The thresholds were developed using a landslide inventory collected from online newspapers by a data mining technique called SECaGN. This method was chosen to account for only those events that echo on the Internet and therefore impact society, ignoring landslides occurred in remote areas, not of interest for civil protection intervention, which would adversely affect the model performance because they would result in false alarms. A calibration phase was performed to minimize the impact of false alarms, allowing at least one false alarm per year over the moderate criticality level. In addition, an innovative approach to include antecedent rainfall as the third dimension of the intensity-duration thresholds was applied, generating a consistent reduction in false alarms. The results were validated through an independent landslide inventory and were compared with (i) the alert issued by the regional civil protection agency to observe the improvements achieved with the proposed model and to evaluate to what extent the proposed model is consistent with the assessments of the civil protection and (ii) a dataset of the national states of emergency to verify the suitability of the developed thresholds for alerting citizens. The thresholds obtained showed high predictive capabilities, confirming their suitability for implementation in an operational landslide early warning system.

Robust small infrared target detection using weighted adaptive ring top-hat transformation

Robust small infrared target detection using weighted adaptive ring top-hat transformation

Detection of application-layer DDoS attacks using machine learning and genetic algorithms

Detection of application-layer DDoS attacks using machine learning and genetic algorithms

Response Characteristics of Smoke Detection for Reduction of Unwanted Fire Alarms in Studio-Type Apartments

Photoelectric smoke detectors (SDs) often emit false alarms in studio-type apartments, where fire prevention is crucial. This study investigates the response characteristics of conventional and analog smoke detection factors to reduce false positives in studio-type apartments. A mock-up was tested based on relevant domestic laws, standards, statistical data, and experimental cases. A simulation of a cooking scenario involving burned food items was conducted, and optical density, particulate matter (PM), and carbon monoxide levels were measured and compared with actual smoke detection at six different locations. The measured values of conventional smoke detectors (CSDs) and analog smoke detectors (ASDs) at these locations were used to derive the activation time of CSDs and ASDs for the entire mock-up space. The results showed that the CSD activated at 7.42 min, while the ASD activated at 11.57 min. PM10.0, CO, and CO2 showed similar activation time trends. The PM10.0, CO, and CO2 concentrations at the time of SD activation were estimated. The findings suggest that a sensor with a consistent coefficient of variation, such as PM10.0 and CO, should be recommended.