Graph Neural Network based Root Cause Analysis Using Multivariate Time-series KPIs for Wireless Networks

This RCA case study has been initiated before the Gas Turbo generator failure took place by timely intervention when the Gas Turbine incident alarm triggered. Root cause analysis (RCA) process was introduced in ADGAS back in 2004, the process was designed to ensure the elimination of defects and incidents with an integrity exposure or causing production losses from repeating by identifying the root cause and putting in place robust actions to ensure incidents don't reoccur. The RCA process enabled ADGAS to put in place a practical methodology for the initiation and execution of an investigation to resolve an incident that could cause a significant loss in production, Loss of functions and / or process contamination, an integrity risk or a repetitive problem that had to be eliminated. The RCA findings would have to undergo a review before the findings and actions are approved and issued. LG6 Gas Turbine is generating 1/3 of Das Island electrical power and very critical for ADMA / ADGAS production sustainability. In May2011 LG6 encountered a severe failure of its Gear Box, entire mechanical package misalignment, load coupling compression, severe foundation tilting and load gear box tilting. The inspection findings revealed a high risk of coupling destruction and production losses. Risk assessment highlighted that if the coupling damage while machine running, it could cause a flying objects to surrounding by 30.5 meters away and 4 meter width. This RCA case study has been initiated before the LG6 power generation Gas Turbine failure took place by timely intervention when the Gas Turbine incident alarm triggered. Risk has been assessed and decision made to shut down the machine immediately for site investigation and remedial actions. RCA team members faced a big challenge to get the machine back in service safely and efficiently as soon as possible to save the plant production and avoid bad reputation of partial electrical power off- black out. This also saved a company a rental fees of pack up power generation unit (estimated 24 MM US$ for two years) and assure providing of the required power to keep the operation of LNG, OAG trains and commission of IGD projects on time with no power shortage. This project resulted in significant potential enhancements for all of the ADGAS Turbo machinery units through the innovative application of a proven root cause analysis and remedial actions to identify the truly underlying "root causes" of incidents and developing sustainable repairs. Using a fault tree analysis process this project successfully identified 85 critical possible causes for the rare Frame 5 Gas turbine fault, eliminate by verification one by one pinpointed the cause of failure and developed a repair actions that prevent failure reoccur.

Graph neural networks (GNNs) have recently achieved remarkable success on a variety of graph-related tasks, while such success relies heavily on a given graph structure that may not always be available in real-world applications. To address this problem, graph structure learning (GSL) is emerging as a promising research topic where task-specific graph structure and GNN parameters are jointly learned in an end-to-end unified framework. Despite their great progress, existing approaches mostly focus on the design of similarity metrics or graph construction, but directly default to adopting downstream objectives as supervision, which lacks deep insight into the power of supervision signals. More importantly, these approaches struggle to explain how GSL helps GNNs, and when and why this help fails. In this article, we conduct a systematic experimental evaluation to reveal that GSL and GNNs enjoy consistent optimization goals in terms of improving the graph homophily. Furthermore, we demonstrate theoretically and experimentally that task-specific downstream supervision may be insufficient to support the learning of both graph structure and GNN parameters, especially when the labeled data are extremely limited. Therefore, as a complement to downstream supervision, we propose homophily-enhanced self-supervision for GSL (HES-GSL), a method that provides more supervision for learning an underlying graph structure. A comprehensive experimental study demonstrates that HES-GSL scales well to various datasets and outperforms other leading methods. Our code will be available in https://github.com/LirongWu/Homophily-Enhanced-Self-supervision.

ABSTRACTThis study evaluates whether auditor use of root cause analysis (RCA) for an identified client misstatement affects auditors' assessments of underlying control issues and materiality in an integrated audit setting. We also test whether auditor cognitive style moderates these effects given prior findings that a misfit between task structure and cognitive style undermines performance. This research is motivated by concerns about integrated audit quality and auditors failing to consider control deficiencies indicated by client misstatements. We randomly assigned 147 auditors to four RCA treatments (No RCA, Unstructured RCA, Structured “5 Whys” RCA, Structured “Fishbone” RCA). As predicted, the results suggest that auditors using (not using) structured RCA are more (less) likely to identify control‐related root causes of a financial misstatement and judge the misstatement to be more (less) material. Also consistent with our prediction, we find that the assessed severity of identified control deficiencies mediates RCA's effect on materiality judgments. Finally, the materiality judgment results reveal the expected significant interaction between structured RCA method and auditor cognitive style, suggesting the importance of allowing flexibility in the specific RCA method applied in practice. Overall, the study's results provide evidence that auditor use of RCA as a judgment framework prompts deeper evaluation of audit findings and stronger consideration of critical links between financial reporting and related internal controls in integrated audit settings.

Wells Process Safety (WPS) is a major aspect of delivering a well/ workover safely. An increasing trend in WPS Level 3 incidents was observed in the first four months of 2021. WPS Level 3 incidents are well control events where the first well control barrier was lost leaving one single tested barrier in place. Data shows that in March and April, the number of incidents increased by 60% compared to the first two months. Based on data, it was noticed that 85% of the incidents happened in hoist workover operations. The Practical Problem-Solving (PPS) technique was selected out of the Lean tools to solve this problem. PDO has imported and re-structured the PPS technique to be fit for the Oil and Gas industry and its processes. The problems which are normally tackled by a PPS can be: - ▪ Sudden change/deviation ▪ Unsafe event ▪ Defective product Data was analyzed to break down the problem and figure out its dimensions. Bar charts were used with the aim of getting a logical trend that would enable a better understanding of the problem. Four main trends were observed from the graphical analysis. It was noticed that the WPS incidents were happening on a certain well type, workover unit, activity and geographical locations. Root cause analysis started by using the fishbone tool to brainstorm possible causes of the problem. With the advice of the experts and key stakeholders, a Root Cause Analysis (RCA) was conducted using the "5 Whys" method. After the RCA, a mitigation plan was developed to work out the best countermeasures to tackle the root causes. As a result of the PPS root cause analysis, some of the operating procedures were updated for the relevant operations. Similarly, controls were put in place to tackle all root causes. After 3 months of piloting the countermeasures, no WPS L3 incidents occurred in our workover operation. With the improvement in the data quality, five incidents were reclassified or removed from the data base. Overall, we achieved a 40% reduction in WPS L3 incidents in our Workover Department, mainly driven by the implementation of this PPS approach. This paper will give a brief about the application of Lean methodology in PDO's operations and how the PPS tool can help identify and solve day to day problems. With this project, the Practical Problem Solving (PPS) technique was used for the first time to address WPS related issues. Tools including, stakeholder engagements, root cause analysis and process mapping will be explained in this paper in detail.

Developing a Tool for Assessing Competency in Root Cause Analysis

Background and purpose: Drug dispensing errors have a major impact on the incidence of medication errors, and can be very dangerous to patients and affect the quality of care. Root cause analysis (RCA) is an approach to prevent dispensing errors by identifying the root of the problem, make efforts to improve comprehensively across departments so that incidence of dispensing errors are not repeated. The purpose of this study is to determine the differences in the incidence of dispensing errors before and after the RCA. Methods: This is a descriptive study by calculating the incidence of dispensing errors before and after the RCA, conducted in four inpatient wards at Sanglah Hospital Denpasar. Dispensing errors were categorized into ten types namely errors in drug’s identity, name, dosage, number of drugs, absence of high alert labels, incorrect drug preparations, incorrect strength, expired drugs, incorrect place of delivery and incorrect instructions. In February and April 2018 there were 80 and 40 dispensing errors which met the study criteria, respectively. Results: Crude incidence of dispensing errors before and after RCA were 2.54 per 1000 (95%CI: 1.98-3.10) and 1.26 per 1000 (95%CI: 0.87-1.65), respectively. The decrease in dispensing error after RCA was 50.39%. The ratio of dispensing error incidence rates after and before the RCA is 0.49. Significant decrease in the incidence of dispensing errors was only found in drug identity errors of 1.27 per 1000 (95%CI: 0.88-1.66) before an RCA and 0.06 per 1000 (95%CI: -0.03-0.15) after RCA, or decreased by 95.27% and there was no significant decrease in other dispensing errors. Conclusion: The decrease of dispensing errors after the RCA was only found in the medication identity error and not found in the other dispensing errors. Direct observation and intensive supervision are needed in order to accurately calculate and reduce the incidence of dispensing errors.

Root cause analysis (RCA) is widely used to investigate adverse events in health care and is mandated by many organizations and governments. RCA employs a combination of techniques to establish the factors contributing to a harmful outcome. Once the factors are identified, then interventions are usually designed in order to prevent further harms from occurring. Prior to deciding which intervention(s) to implement, we must make judgements of causal importance in the context of multiple, interacting conditions. Clearly, we must take action to prevent adverse events, but we need not take action against every contributing cause. I aim to show that many causal factors can be identified by RCA, but current approaches do not adequately distinguish among these causes, leaving numerous potential targets of intervention. I argue that this is because the literature on RCA (with few exceptions) has largely neglected the literature on philosophy of causation and explanation. In this paper, I focus on demonstrating how Strevens' kairetic approach to explanation and Tinbergen's four questions to explain behaviour have the potential to work synergistically with the present RCA methods. There are important lessons we can bring to RCA from the literature on causation and explanation, particularly from the philosophy of science and biology. This work could enhance the effectiveness of RCA by identifying and understanding the causes that really make a difference to adverse events. This ought to reduce the number of targets and focus intervention following RCA.

While root cause analysis (RCA) is used to analyze medical errors with a systems approach, evidence demonstrating its effectiveness in reducing patient harm remains sparse. The heterogeneity of the RCA methodology at different health care organizations has posed challenges to studying its value. The Department of Veterans Affairs (VA) has an established and standardized RCA approach, making it an ideal context to study RCA's impact. This review assessed whether implemented interventions recommended by RCAs were effective in mitigating preventable adverse events at the VA. PubMed, Web of Science, CINAHL and Business Source were searched for studies on RCAs performed at the VA that evaluated effectiveness of interventions and were published between 2010 and 2020. The Appraisal Tool for Cross-sectional Studies (AXIS) was used to assess bias of bias. The majority of studies eliminated during our eligibility process reported on RCAs without attention to their specific impact on patient safety. Ten retrospective studies met inclusion criteria and were part of the final review. Studies were grouped into adverse events related to incorrect surgical/invasive procedures, suicides, falls with injury, and all-cause adverse events. Six studies reported on effectiveness by demonstrating quantitative changes in adverse events over time or by location following a specific intervention. Four studies reported on the effectiveness of implemented interventions using a facility-based rating of "much better" or "better." Of the studies included in this review, all reported improvements following interventions implemented after RCAs, but with variability in study definitions and methodology to assess effectiveness. Increased reporting of outcomes following RCAs, with an emphasis on quantitative patient-related outcome measures, is needed to demonstrate the impact and value of the RCA.

<h3>Background</h3> Following the introduction of a new iodine-containing contrast medium (IC) (iomeprol) to our hospital formulary, an unexpected increase in adverse drug reactions (ADRs) to the IC was observed. Consequently, a root cause analysis (RCA) was made to identify the causes of adverse events, and to suggest improvements to prevent recurrences. <h3>Purpose</h3> To determine the corrective measures resulting from an RCA to investigate the unexpected increase in ADRs associated with the introduction of a new IC (iomeprol) in the hospital, and to evaluate the effectiveness of measures implemented. <h3>Material and methods</h3> The study was made in a 460-bed specialty hospital. The RCA multidisciplinary team was composed of medical and nursing directors and supervisors of radiology and pharmacy units. Steps followed in RCA involve: 1-problem description and data collection, 2-organisational system analysis, 3-human factor analysis, 4-team and material management analysis, 5-patient-related factors analysis, 6-RCA team proposal. The monthly number of ADRs notified before and after RCA was used to assess the effectiveness of corrective measures. Notifications of IC ADRs were extracted from the records of yellow card reports, as well as patient data, drug data and ADR data. <h3>Results</h3> Proposals generated by RCA were: 1-to retrain radiology technicians and doctors on the use of IC pumps for administration, 2-to inspect pump calibration, 3-to verify the premedication needed. In the period 2009–2012 average notification of IC ADRs was 0.87/month. Since iomeprol was introduced in 2013/03 until RCA was performed (2013/05/08) 9 notifications were recorded (6/month); type and severity were similar to the previous period. Retraining and pump calibration inspection were finalised in June. The average IC ADR notification rate in the period July–December 2013 was 1.67/month. In the period January–October 2014, the notification rate was 0.89/month, similar to that before introduction of the new IC. <h3>Conclusion</h3> 1. Technical training, machine calibration and patient evaluation were identified by RCA as aspects to improve. 2. After corrective measures had been taken, IC RCA notification has returned to the usual value. <h3>References and/or acknowledgements</h3> No conflict of interest.

Adverse events reportedly occur in 5% to 10% of health care episodes. Not all adverse events are the result of error; they may arise from systemic faults in the delivery of health care. Catastrophic events are not only physically devastating to patients, but they also attract medical liability and increase health care costs. Root cause analysis (RCA) has become a key tool for health care services to understand those adverse events. This study is a review of all the RCA case reports involving neurosurgical patients in New South Wales between 2008 and 2013. The case reports and data were obtained from the Clinical Excellence Commission database. The data was then categorized by the root causes identified and the recommendations suggested by the RCA committees. Thirty-two case reports were identified in the RCA database. Breaches in policy account for the majority of root causes identified, for example, delays in transfer of patients or wrong-site surgery, which always involved poor adherence to correct patient and site identification procedures. The RCA committees' recommendations included education for staff, and improvements in rostering and procedural guidelines. RCAs have improved the patient safety profile; however, the RCA committees have no power to enforce any recommendation or ensure compliance. A single RCA may provide little learning beyond the unit and staff involved. However, through aggregation of RCA data and dissemination strategies, health care workers can learn from adverse events and prevent future events from occurring.

BackgroundConducting root cause analysis (RCA) is complex and challenging. The aim of this study was to better understand the experiences of RCA team members and how they value their involvement in the RCA to inform future recruitment, conduct and implementation of RCA findings into clinical practice.MethodsThe study was set in a health network in Adelaide, South Australia. A qualitative exploratory descriptive approach was undertaken to provide an in-depth understanding of team member’s experience in participating in an RCA. Eight of 27 RCA team members who conducted RCAs in the preceding 3-year period were included in one of three semi-structured focus groups. Thematic analysis was used to synthesise the transcribed data into themes.ResultsWe derived four major themes: Experiences and perceptions of the RCA team, Limitations of RCA recommendations, Facilitators and barriers to conducting an RCA, and Supporting colleagues involved in the adverse event. Participants’ mixed experience of RCAs ranged from enjoyment and the perception of worth and value to concerns about workload and lack of impact. Legislative privilege protecting RCAs from disclosure was both a facilitator and a barrier. Concern and a desire to better support their colleagues was widely reported.ConclusionsClinicians perceived value in reviewing significant adverse events. Improvements can be made in sharing learnings to make effective improvements in health care. We have proposed a process to better support interviewees and strengthen post interview follow up.

BackgroundDespite over a decade of efforts to reduce the adverse event rate in healthcare, the rate has remained relatively unchanged. Root cause analysis (RCA) is a process used by hospitals...

The identification of defect causes plays a key role in smart manufacturing as it can reduce production risks, minimize the effects of unexpected downtimes, and optimize the production process. This paper implements a literature review protocol and reports the latest advances in Root Cause Analysis (RCA) toward Zero-Defect Manufacturing (ZDM). The most recent works are reported to demonstrate the use of machine learning methodologies for root cause analysis in the manufacturing domain. The popularity of these technologies is then summarized and presented in the form of visualizing graphs. This enables us to identify the most popular and prominent methods used in modern industry. Although artificial intelligence gains more and more attraction in smart manufacturing, machine learning methods for root cause analysis seem to be under-explored. The literature survey revealed that only limited reviews are available in the field of RCA towards zero-defect manufacturing using AI and machine learning; thus, it attempts to fill this gap. This work also presents a set of open challenges to determine future developments.

Objective To explore the effect of root cause analysis(RCA) on reducing the turnover rate of nurse staff from a tertiary stomatology hospital. Methods It was collected that the nurses in Stomatological of Chongqing Medical University from January 2012 and December 2016. 2015-2016, using the method of RCA scientific management, establishment of RCA team, the reason of high turnover rate of nurses were analyzed using RCA method, the improvement plan was worked out. It was analyzed that the nurses turnover rate before and after improvement plan. Results The turnover rate of nurses was 9.35%, 12.30%, 13.70%, 4.23% and 5.05% respectively from 2012 to 2016. From 2012 to 2014, the total number of nurses was 971, with a total turnover of 113 and an average turnover rate of 11.80%. From 2015 to 2016, the total number of nurses was 831, with a total turnover of 37, an average turnover rate of 4.66%, Comparing the turnover rate from 2012 to 2014, the average turnover rate was reduced significantly after improvement, the difference was statistically significant (χ2=29.292, P<0.005). Conclusion RCA is a scientific method in the management of nursing human resources. RCA method can solve the problem of high turnover rate of nurse staff to reduce the loss of the oral specialized nurse. Key words: Root cause analysis; Stomatology hospital; Specialized nurse; Turnover rate

Root Cause Analysis (RCA) has been recommended as a specific problem-solving approach for addressing disproportionate discipline outcomes. RCA differs from other problem-solving models through its in-depth analysis that draws from the perspectives of individuals who are impacted by the issue being addressed. When applied to disproportionate discipline, this means that RCA would include perspectives from all school staff, as well as students and families who are over-represented in discipline outcomes. This paper describes an approach to conducting RCA and presents two case studies that illustrate what stakeholders may gain by incorporating RCA into a school’s efforts to increase disciplinary equity. Suggestions for first steps school teams can take to conduct a root cause analysis will be provided.