Next Article in Journal
State of the Art of Digital Twins in Improving Supply Chain Resilience
Previous Article in Journal
Sustainability of the Collection of Norwegian Household Waste
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Logistics
Volume 9
Issue 1
10.3390/logistics9010023
logistics-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Self-Healing Databases for Emergency Response Logistics in Remote and Infrastructure-Poor Settings

by
James McGarvey
1,
Martha R. Grabowski
2,*,
Buddy Custard
3 and
Steven Gabelein
3
1
Management, Leadership & Information Systems Department, Madden College of Business & Engineering, Le Moyne College, Syracuse, NY 13214, USA
2
Industrial & Systems Engineering Department, Rensselaer Polytechnic Institute, Troy, NY 12180, USA
3
Alaska Chadux Network, Anchorage, AK 99507, USA
*
Author to whom correspondence should be addressed.
Logistics 2025, 9(1), 23; https://doi.org/10.3390/logistics9010023
Submission received: 7 November 2024 / Revised: 17 January 2025 / Accepted: 3 February 2025 / Published: 6 February 2025
Browse Figures
Versions Notes

Abstract

:
Background: Accurate, real-time data about response technologies, capabilities, and availabilities are key to effective emergency response logistics; this is particularly important in remote settings, such as in the Arctic, where limited infrastructure, logistics, and technologies occasion the need for careful planning and immediate response in a fragile, pristine, and rapidly changing ecosystem. Despite persistent calls for improved data quality, processing, and analysis capabilities to support Arctic emergency response logistics, these issues have not been addressed and advanced analytical methods available in other safety-critical and oil and gas settings, such as machine learning, artificial intelligence (AI), or emergent, self-aware, and self-healing databases, have not been widely adopted. Methods: This work explores this research gap by presenting a machine learning algorithm and self-healing database approach, describing its application in Arctic logistics and emergency response. Results: The self-healing algorithm could be applied to other safety-critical databases that could benefit from technology that automatically detects, diagnoses, and repairs data anomalies and inconsistencies, with or without human intervention. Conclusions: The results show significant improvements in data cleaning and analysis, and for emergency response logistics data, planning, and analysis, along with future research and research needs in remote and infrastructure-poor settings.

1. Introduction

Climate change, energy and natural resource developments, and thinning Arctic sea ice have increased the accessibility of Arctic waters for energy exploration and development [1], fishing [2], tourism [3], and maritime shipping [4]. Over the past decade, vessel transits through the Bering Strait have increased by 88%, from 262 in 2009 to 494 in 2019 [5]. At the same time, transits through the Northern Sea Route (NSR), to and from Russia across Norway, increased by 28% between 2019 and 2021 [6]. Ship traffic is expected to increase over the next decade and beyond, as the NSR becomes a more attractive shipping route for vessels traveling between Western Europe and Asia along the Russian coast, and offshore oil development in the Arctic Ocean; onshore oil, gas, and mining operations in Alaska; and cruise vessel traffic are expected to grow [5].
Just-in-time expectations for goods and services have increased global pressure for shorter shipping timelines, for faster turn-rounds, and for increased shipping activity in the Arctic, perhaps with different attention to safety risks. In September 2024, the first large Panamax container ship transited from Europe to Shanghai in 16 days, which saved days and weeks of Suez Canal transit [7]. A month later, in October 2024, a Chinese vessel without an ice class certification traveled across the Northern Sea Route with sanctioned goods for Russia’s Arctic LNG 2 project [8]. As pressures and expectations mount with increased Arctic shipping, so too does the potential for a major Arctic oil spill, which could have enormous impacts on the region’s economy, ecosystems, social and cultural resources, and fragile subsistence communities [9,10]. Nations, states and provinces, industries, and non-profit enterprises have significantly invested in emergency response logistics planning to address these risks, through systemic and localized risk assessments [11], resource allocation modeling [12], studies of oil and gas characteristics and toxicity [13], the development of Arctic transport decision support systems [14], and studies of the impacts of oil on wildlife and marine biota, communities, and ecosystems, particularly in ice and extreme conditions [15].
Key to effective emergency response logistics are accurate, real-time data about technologies, capabilities, and availabilities; this is particularly important in remote settings, such as in the Arctic, where limited infrastructure, logistics, and technologies occasion the need for careful planning and immediate response in a fragile, pristine, and rapidly changing ecosystem [16,17,18]. In the Arctic, the lack of information, communications, and transportation infrastructure, and the enormous distances between sparsely populated villages, coupled with the unpredictability of weather, and the magnitude and power of sea states and environmental conditions, complicate access to information, and can delay deployment and supply of critical resources. Public expectations for four-season response capabilities in the event of an incident in the Arctic also increase the need for timely analysis of reliable equipment, personnel, and resource information to support effective Arctic emergency response logistics [19].
Data issues in Arctic emergencies and oil spill response have long been studied, with many calls for improved data quality, availability, and integration [9,10,20,21,22]. Emergency response logistics data are often disaggregated, incomplete, and captured by multiple stakeholders using different platforms, standards, and data definitions [9,22]. Disaggregated and non-standard data in logistics and emergency response events, which involve federal, state and local agencies, industries, and environmental and cultural stakeholders, can result in delayed response activities, driven by a lack of shared situational awareness, and data integrity and compatibility issues [9,20,21]. Most recently, the U.S. Government Accountability Office called for improvements in Coast Guard operational data in the Arctic, citing incomplete and unavailable data, reports, and performance metrics [23].
Despite persistent calls for improved data quality, processing, and analysis capabilities to support Arctic emergency response logistics, these issues have not been addressed and advanced analytical methods available in other safety-critical and oil and gas settings, such as machine learning, artificial intelligence (AI), or emergent, self-aware, and self-healing databases [24,25], have not been widely adopted. This is a critical research gap in remote and infrastructure-poor logistics settings such as Arctic emergency response, where the need for real-time response is critical and the human capital resources to effect response, and to maintain safety-critical databases, are sparse. After describing previous work, this paper addresses that research gap by presenting a machine learning algorithm and self-healing database approach, and describing its application in Arctic logistics and emergency response. The self-healing algorithm could be applied to other logistics, emergency response, and safety-critical databases that could benefit from technology that automatically detects, diagnoses, and repairs data anomalies and inconsistencies, with or without human intervention. The results suggest improvements for emergency response logistics data, planning, and analysis, along with future research and research needs, particularly in remote and infrastructure-poor settings.

2. Previous Work

Effective emergency response and logistics operations require a confluence of well-orchestrated activities: efficient resource allocation, shared situational awareness of the response effort, and efficacious response [26,27]. Central to the logistics of resource allocation, situation awareness, and effective response are accurate, complete, and available data about personnel, equipment, and funding availability and capabilities, and the ability to develop response strategies in real-time. Optimized response in resource-constrained settings such as the Arctic also require attention to information about events, response timelines, resource transit times, and logistics, as well as environmental and social considerations, such as wildlife and marine mammal impacts [28], cultural concerns [29], oil and fuel characteristics, and behavior in extreme settings, and impacts on local and subsistence communities [22,30].
Emergency response logistics thus require high-quality, accurate, and complete data about transportation, personnel, routes, weather, events, and other contingencies, as well as information about response equipment, resources, and personnel. Information about events includes location, timing, duration, severity, trends, or change data, as well as information about resources deployed, their timing, and effectiveness. Resources have costs, locations, shelf lives, optimal deployment periods, transit times, constraints about their use and maintenance, as well as weight, height, transportation, and storage considerations, knowledge of which is critical for effective response and mission completion. In the Arctic, emergency response logistics also rely on current environmental information such as weather, ice, and storm impact data, as well as information about community, marine, and subsistence economy information, including wildlife, sea mammal and bird migration, and hunting seasons and patterns, all of which can impact an effective response [9].
Data analysis and modeling also rely on common data definitions, consistent data structures, and data capture standards, especially since participants in Arctic logistics and emergency response are geographically scattered, with varying degrees of access to and understanding of the on-the-ground operational picture. Distributed decision-makers in large-scale safety–critical systems often rely on standard data definitions and classifications to ensure that all members are using the same data to develop shared mental models of the system and to make decisions [31]. Standard data definitions and ontologies help ensure that data have been captured, represented, labeled, and stored with consistent data descriptions and nomenclature, that different datasets can be shared and integrated, and that data dictionaries represent and store the same data. Absent standard data definitions and ontologies, and classification schemes, decision-makers, regulators, analysts, and modelers can face difficulties relying on critical data, and significant data cleansing, reconciliation, and management efforts may be required.
Real-time analysis of emergency response logistics data is also important so that data sources, analytical tools, and methods for assessing data and identifying trends reflect current conditions and response needs. However, real-time data analysis for Arctic emergency response logistics is complicated by long-standing data challenges [9,10,12]. Improved analytic processes, including machine learning applications for global supply chain enhancement, have been developed [32], but applications in Arctic emergency response logistics have been needed for some time, stemming from pressure for real-time analysis of emergency response data, and for real-time predictive models for risk analysis, resource allocation and planning. Machine learning applications integrated with AI and the Internet of Things (IoT) are beginning to be introduced, linking data from assets, personnel, and systems, and increasing expectations for real-time analysis [30].
The power of real-time machine learning and other analytical approaches has often been limited; however, in cases where data are inconsistent, missing, or inaccurate [33]. As a result, interest developed in automated and autonomous capabilities to detect system anomalies, diagnose solutions, and, with or without human intervention, correct anomalies [34]. These systems, which may utilize embedded neural nets or other machine learning techniques, have shown improved database accuracy and integrity [35], correcting faults automatically and, optimizing query performance through machine learning, AI, and deep learning models [36], and performing self-maintenance [37].
Self-healing systems are one form of autonomic computing capabilities, which also include self-configuration, self-optimization, and self-protection [38]. Self-healing systems detect and can address system and data anomalies, integrating with machine learning models and learning from historical data, real-time analytics, and predictive models [39]. The integration of self-healing databases with AI, machine learning, and deep learning offers the potential for anomaly detection, self-optimization, and automated recovery in databases without human intervention [36,37]. Despite these advances in other domains, however, self-healing systems have not been utilized extensively in emergency response logistics, although their potential has been recognized [40].
Emergent systems are another form of learning technology that mimics self-organization principles and capabilities in social and organizational systems [41] and offers capabilities for self-configuration, optimization, healing, and protection [42]. Pattern recognition capabilities in emergent systems can be used to identify anomalies in behavior, movement, analysis, or data, often using sequences that match specific patterns and suggesting novel solutions [43]. AI-driven anomaly detection for self-optimizing systems has been used in cloud management systems to analyze patterns, offer predictions, or enhance system performance [44], and applications in emergency response and logistics planning are emerging [45,46].
Technology approaches and solutions to address long-standing and persistent data, processing, and analysis issues in emergency response logistics data, therefore, exist; however, they have not been developed or applied in remote and infrastructure-poor logistics settings such as the Arctic emergency response, where the need for a real-time response is critical and the human capital resources to effect response, and to maintain safety-critical databases, are sparse. This paper describes one such approach and relates experience with a neural network-based approach for anomaly detection, diagnosis, and self-healing in emergency response logistics databases, with attention to the limits and complexity they introduce in real-time analysis in safety-critical systems, particularly in remote locations [44]. The following section describes an Arctic logistics and emergency response case study with challenges in missing, incomplete, and redundant data in an infrastructure-poor setting where real-time analysis and logistics planning are required.

3. Materials and Methods

Timely analysis of oil spill prevention and emergency response logistics information is particularly important in remote settings such as the Arctic, as the lack of Arctic infrastructure and the need for careful planning and immediate response makes operations difficult in this fragile, pristine and rapidly changing ecosystem with extended logistical supply chains, personnel shortages, and long distances to travel [9,10,11]. Despite the need for advanced analytics and machine learning techniques to optimize and improve logistics and emergency response in the Arctic, these approaches have not been applied in this setting. Previous work identified the need for high-quality, accurate, and complete data, common data standards, and real-time logistics and emergency response analysis. However, little work has developed advanced analytical approaches to address these challenges, including applications of machine learning and self-healing capabilities that could improve logistics and emergency response analyses and processes, and surmount human capital challenges in remote and infrastructure-poor settings. This paper presents such an approach and describes its application in Arctic logistics and emergency response. The self-healing technique addresses the need for improved data quality and processing for real-time emergency response and logistics in an environmentally sensitive and safety-critical system with limited physical, technological, and human resources.

3.1. Setting and Background

The setting for this study is the U.S. Arctic and Western Alaska, where vessels transit the Gulf of Alaska, the Bering Sea and Bering Strait, the Aleutian Islands, and along the Western coast of Alaska, through the Chukchi and Beaufort Seas (Figure 1).
As Arctic waters warm, more vessels are engaged in Arctic voyages, particularly as the prospect of an ice-free passage through the Northern Sea Route across Russia or the Northwest Passage across Canada looms [5]. To protect the Arctic and other waters, and to support safe and secure maritime transportation, the U.S. Coast Guard specifies oil spill response requirements for owners or operators of regulated tank and non-tank vessels when in U.S. waters. These requirements are codified in the Oil Pollution Act of 1990 (OPA-90) [9,47], which was passed by the U.S. Congress following the 1989 Exxon Valdez oil spill in Prince William Sound, Alaska [47,48]. For Arctic vessel transits, the state of Alaska also prescribes vessel operations, maintenance, and oil spill response planning and equipment requirements for vessels that operate within three nautical miles of the shore and that call on ports in Alaska [49].
Vessels operating in the U.S. Arctic and Western Alaska must document their required oil spill response resources and plans in Vessel Response Plans (VRPs), which are guided by international standards and rules. In addition to specifying what equipment (containment mechanisms, deployment tools, etc.) needs to be available, VRPs also describe oil spill response logistics best practices, guidelines for vessel outfitting and equipment, requirements for oil spill response drills, fuel sulfur limits, and maritime transit areas to be avoided. Federal laws and regulations also require VRPs to identify contracts for oil spill response services, salvage, and qualified oil spill response individuals. These requirements are linked to a vessel’s worst-case discharge (WCD) metric, the volume of oil that could be discharged from the vessel in adverse weather (U.S. 33 Code of Federal Regulations (CFR) 155.1020).

3.2. Approach

In this study, data were collected by our research partner, an Oil Spill Removal Organization (OSRO), as part of their oil spill response monitoring and data collection activities. The dataset was cleaned using automated scripts to reduce duplicative and conflicting data records. Following the automated scripts data cleaning, the process was repeated on the same dataset with a self-healing process using machine learning. Process improvement metrics were then calculated, and observations and insights were made regarding the use of self-healing and emergent approaches to address data quality issues for Arctic oil spill response and logistics analysis.

3.3. Data

The data comprised 11,568 oil spill response records from 2048 participating vessels making transits through waters off Western Alaska and the Arctic between 1 January and 31 August 2023. The attributes in the dataset are shown in Table 1, and the vessels and vessel data types are shown in Table 2.
Table 2 shows the different types of vessels in the database; there were no records for break bulk, ferry, fishing, heavy lift, oil, oil chemical or miscellaneous vessels in the dataset. Incomplete data about one pleasure craft/yacht and two research/survey vessels were deleted from the dataset. There were no ‘blank’ fields for vessel type.

3.4. Method

Visual Basic (VBA) scripts using the Excel macros shown in Appendix A were developed to organize and clean the original dataset. The baseline scripts were reviewed using a generative AI tool, ChatGPT; the reviewed scripts were then modified to provide a context for further development (red, green, and blue color-coded data for certain data criteria (e.g., missing data)), and script instructions were developed to handle the resulting dataset (e.g., condense the multiple color-coded rows to one row with multiple columns). Once the missing data scripts were finalized, additional scripts to eliminate duplicate data were then created organically using lessons learned from the missing data script development. The scripts were tested with an open-source dataset from an academic setting; no oil spill response data were ingested into a generative AI tool. The tested scripts were then used to clean a copy of the oil spill response dataset, by moving data in color-coded cells to green (cleaned cells), eliminating duplicate information. The resulting database was then reviewed by visual inspection to ensure that all missing and duplicative data had been flagged, captured, and resolved. The Visual Basic scripts are provided in Appendix A.
The automated analysis showed that, of the original 11,568 records, 6490 (56.1%) were duplicative, and 118 were missing transit exit dates or vessel type data (1.02%). Other incomplete records were then identified and deleted from the database, which resulted in an additional reduction of 2891 records (an additional 24.99% reduction). The inspection also showed that three data points skewed the data distribution; these outlier data were reviewed with the research partner, confirmed to be incorrect, and were removed from the dataset. The dataset resulting from the automated cleaning resulted in 2069 records, which was 17.89% of the original dataset (Table 3).

4. Analysis and Results

The initial database analysis using automated anomaly detection scripts highlighted a number of data difficulties. Incompatible data capture methodologies, data standards, and data and metadata definitions have been emblematic of emergency and oil spill response data challenges for many years [9,50]. In this case study, incompatible data capture policies and standards led to the development of a database that required substantial cleaning prior to analysis. Data cleaning has long been highlighted as a bottleneck in data analytics [51]; in emergency response logistics data analysis, the lack of common standards and definitions have resulted in data quality issues with a significant impact on data accuracy and usability [20].
Interest in improving the data analysis process arose following the automated scripts process, which, although automated, still required significant time for anomaly detection and data cleaning. As a result, neural network techniques were applied for data cleaning with the original dataset. A neural network model was created using freely available open-source resources (Appendix B). A preliminary algorithm was developed that used color coding to highlight duplicative and incomplete data. The results were then reviewed by a human subject matter expert. The algorithm was then updated using a second set of training data from the original dataset, and was run again on the large dataset. Redundant data were deleted, and a tracer record was created for audit and compliance testing and analysis. Missing and incomplete data were flagged by the neural net so that the dataset could be updated with the missing information. Machine learning was used to propose potential data to complete the data records based on contextual information and data in the dataset. Process improvements in terms of time differences were recorded, comparing the use of the automated scripts and the neural nets algorithm. The results are shown in Table 4.
Interest in improving the data analysis process arose following the automated scripts process, which, although automated, still required significant time for anomaly detection and data cleaning. As a result, neural network techniques using the Python code in Figure A5 in Appendix B were applied for data cleaning with the original dataset. A neural network model was created using freely available open-source resources. A preliminary algorithm was developed that used color coding to highlight duplicative and incomplete data. The results were then reviewed by a human subject matter expert. The algorithm was then updated using a second set of training data from the original dataset, and was run again on the large dataset. Redundant data were deleted, and a tracer record was created for audit and compliance testing and analysis. Missing and incomplete data were flagged by the neural net so that the dataset could be updated with the missing information. Machine learning was used to propose potential data to complete the data records, based on contextual information and data in the dataset. Process improvements in terms of time differences were recorded, comparing the use of the automated scripts and the neural nets algorithm. The Python code developed is shown in Figure A5 in Appendix B.
The differences in time required for data cleaning using the automated anomaly detection scripts and using the neural nets approaches were significant, with a 78.33% reduction in time with the neural nets (Table 4), which used Python code to accelerate the cleaning process. The Visual Basic script data cleaning required over three times the cleaning time compared to the use of the neural nets, showing significant process improvements.

5. Discussion

Real-time data analysis using current and accurate data is imperative in safety-critical settings, including logistics planning for emergency and oil spill response in the Arctic, processes which are time-sensitive, carried out in high-pressure settings, and are impacted by physical, technological, and human resource limits [52,53]. In this context, this research explored an approach to addressing the need for high-quality data, real-time data analysis, consistent data structures, and improved data analytic processes in an environmentally sensitive and safety-critical setting with limited physical, technological, and human resources. Previous research identified gaps in Arctic emergency and oil spill response data, logistics planning, and analysis [9,22,54], and suggested the role that advanced analytics and AI could play in advancing emergency response logistics planning and analysis [37,55]. As hardware, software, and analytic solution costs decrease over time, AI and machine learning solutions have become accessible and realistic possibilities, even in settings with limited infrastructure. Human resource limitations in terms of technological knowledge and availability compound challenges in Arctic emergency response logistics and planning, but those limits could be addressed with the advent of reliable, affordable, and sustainable machine learning support [14,21].
High-quality, complete, and consistent data are required for effective decision support in emergency response and safety settings, not only to ensure that decisions taken are informed by accurate data, but also in order to support timely and efficient information processing. This research utilized self-healing databases that embedded a machine learning classification algorithm in order to improve data analytics processes. The process improvements focused on data cleaning since those tasks require significant time and effort [51]. The neural net model deployed was able to detect, diagnose, and correct anomalies in the dataset, resulting in several improvements:
  • Improved data quality, as missing, incomplete, and redundant data were identified and resolved using the neural net’s classification algorithm;
  • Improved data analysis processes, particularly with data cleaning tasks, compared to the use of automated anomaly detection scripts;
  • Improved support for real-time data anomaly and inconsistency detection, as inconsistent, conflicting, and missing data were automatically flagged with the use of the classification algorithm;
  • Improved utilization of scarce human resources, who could supervise and review the results of the automated data anomaly detection, diagnosis, and correction processes, rather than engage in time-consuming, cumbersome, and error-prone human data cleaning and review processes.
This research fills a gap in Arctic and remote logistics and emergency response systems, describing a self-healing machine learning algorithm that addresses the need for improved data quality and data processing for real-time emergency response and logistics in an environmentally sensitive and safety-critical system with limited physical, technological, and human resources. Previous work identified the need for high-quality, accurate, and complete data, common data standards, and real-time logistics and emergency response analysis. However, little work has developed advanced analytical approaches to address these challenges, including applications of machine learning and self-healing capabilities that could improve logistics and emergency response analyses and processes, and surmount human capital challenges in remote and infrastructure-poor settings. In addition, little work has developed these analytic techniques for real-time applications in these settings; thus, the novel contributions of this research are the development of machine learning and self-healing database capabilities, and their real-time application in a safety- and time-critical setting, showing improved data quality, data analysis, and data processes, all critical goals in safety and logistics analyses.
Consistent data structures, data capture, and standards are integral to emergency response logistics planning [56,57]. However, Arctic emergency response logistics occur in a setting plagued by technological, data, and network infrastructure challenges [9]. Coast Guard data needs for Arctic missions, including search and rescue, emergency response, and oil spill prevention and response, have long been of concern [23], and maritime databases, such as the one utilized in this research, have similar issues. Central to maritime data quality issues are inconsistent data standards, metadata and data structures, and data capture processes [20]. The machine learning techniques employed in this research could be utilized to address the underlying and systemic issues related to maritime data quality, introducing self-healing databases in this setting. Yet to be resolved, however, are data integration and data management issues associated with linking self-healing databases and traditional databases [58], a challenge that was noted, but not addressed, in this research.
The use of machine learning and AI approaches in emergency logistics planning is still emerging, and interest in these tools and approaches is growing [59,60]. The use of AI-driven anomaly detection using algorithms and predictive models is increasing, including the potential for self-aware and self-healing databases [40]. However, these techniques introduce challenges with respect to integrating or embedding machine learning models with or in traditional emergency response logistics datasets. Such integration can lead to ‘data friction,’ when models and datasets developed using different standards and methodologies are linked, leading to potential disconnects in analysis, data, and data processing [61]. Integrating machine learning capabilities into existing emergency response logistics databases will make analyses more data-driven, with increasing needs for interoperability among data, tools, and models. Metadata—usually viewed simply as ‘data about data’— will play an increasingly important role in interoperability, even if metadata may be a source of friction between differing models, tools, and datasets [62]. Improved data analytic processes, therefore, will likely occasion a need for continued attention to integration, interoperability, and the impacts of disparate datasets [23,62].
There are a number of limitations with this research that should be noted. First, the self-healing algorithms were developed with a single dataset comprising data from 2048 ships and multiple Arctic shippers for an 8-month period in 2023. The time period covered was limited, and future work should clearly extend the data collection longitudinally, and with more shipping companies, over a longer time period. The data were not validated against another dataset, and their representation across a diverse set of Arctic shipping companies, although wide, was not universal.
Data quality issues in emergency response logistics have been noted for many years. Historical criticisms include the lack of complete, accurate, and validated datasets that continue to the present day [10,23]. Developing and training machine learning and self-healing algorithms and databases on incomplete, missing, or conflicting data introduces challenges in algorithm development, creates non-standard data, and raises questions about the validity of the models, and their learning over time [63]. Dataset size also impacts the validity and reliability of machine learning results [64]; missing, incomplete, and redundant data contribute to reduced sample sizes for machine learning algorithms and other statistical analyses. Thus, caution should be exercised with the results of this study, and the generalization of these results outside of the Arctic and Western Alaska should be explored. Those facets not examined in this research were the root causes of the data quality issues, or the lack of systemic organizational response to address the underlying data quality issues [23]. Root cause analyses, perhaps employing AI, might provide important input to learning algorithms, leading to training and learning that could address and rectify the human vulnerabilities and errors that led to the development of weak datasets [65].
In this study, missing and incomplete data were flagged by the self-healing algorithm so that the dataset could be updated with the missing information. Machine learning was used to propose potential data to complete the data records, based on contextual information and data in the dataset. Imputation and data replacement methods could also have been employed [66]. Redundant data were deleted, and a tracer record was created for audit and compliance testing and analysis. In future work, these steps could be automated, with or without appropriate human intervention, to further improve and accelerate the process of developing clean data. The strength of the findings of this research could be improved in the future with longitudinal analysis, using the improved or ‘healed’ data, which were beyond the scope of this initial research; insights into the process and decision support improvements could then be noted. Finally, this research was carried out in a technologically challenging and infrastructure-poor setting. Future work could address the impacts of technology, operational, and infrastructure constraints on advanced analytical work, including the development of self-healing databases.

Author Contributions

Conceptualization, J.M., M.R.G. and B.C.; Methodology, J.M., M.R.G. and S.G.; Validation, J.M., M.R.G., B.C. and S.G.; Formal Analysis, J.M., M.R.G. and S.G.; Investigation, J.M., M.R.G. and S.G.; Resources, M.R.G. and B.C.; Data Curation, J.M. and S.G.; Writing—original draft preparation, J.M. and M.R.G.; Writing—review and editing, J.M., M.R.G., B.C. and S.G.; supervision, M.R.G. and B.C.; Project administration, M.R.G.; Funding acquisition, M.R.G. and B.C. All authors have read and agreed to the published version of the manuscript.

Funding

This work was supported by the Alaska Chadu x ̂ Network and by the McDevitt Foundation at Le Moyne College.

Data Availability Statement

No new data were created in this study. The datasets presented in this article are not readily available because of the competitive sensitivity of the data. Requests to access the datasets should be directed to the Alaska Chadux Network.

Acknowledgments

The authors are grateful for support from Alaska Chadu x ̂ Network, a 501(c)(4) non profit, and from the U.S. Coast Guard, Sector Western Alaska and the U.S. Arctic, particularly CAPT Christopher Culpepper, USCG; CDR Scott Farr, USCG; LCDR Abbie Foster, USCG. The authors gratefully acknowledge the reviews and suggestions of 3 anonymous reviewers and those of the managing editor, all of whose suggestions greatly improved this manuscript.

Conflicts of Interest

The authors declare no conflicts of interest.

Appendix A

Visual Basic (VBA) scripts using the Excel macros shown in Appendix A were developed to organize and clean the original dataset. The baseline scripts were reviewed using a generative AI tool, ChatGPT; the reviewed scripts were then modified to provide a context for further development (red, green, and blue color-coded data for certain data criteria (e.g., missing data)), and script instructions were developed to handle the resulting dataset (e.g., condense the multiple color-coded rows to one row with multiple columns). Once the missing data scripts were finalized, additional scripts to eliminate duplicate data were then created organically using lessons learned from the missing data script development. The scripts were tested with an open-source dataset from an academic setting; no oil spill response data were ingested into a generative AI tool. The tested scripts were then used to clean a copy of the oil spill response dataset, by moving data in color-coded cells to green (cleaned cells), eliminating duplicate information. The resulting database was then reviewed by visual inspection to ensure that all missing and duplicative data had been flagged, captured, and resolved. Figure A1, Figure A2, Figure A3 and Figure A4 show the macros created to clean the data.
Logistics 09 00023 g0a1aLogistics 09 00023 g0a1b
Figure A1. Transfer values in Column V across green cells across V to AA.
Figure A1. Transfer values in Column V across green cells across V to AA.
Logistics 09 00023 g0a1aLogistics 09 00023 g0a1b
Logistics 09 00023 g0a2aLogistics 09 00023 g0a2b
Figure A2. Changes values in Column AK.
Figure A2. Changes values in Column AK.
Logistics 09 00023 g0a2aLogistics 09 00023 g0a2b
Logistics 09 00023 g0a3aLogistics 09 00023 g0a3b
Figure A3. Transfers cells in Column I to green cells across I-K.
Figure A3. Transfers cells in Column I to green cells across I-K.
Logistics 09 00023 g0a3aLogistics 09 00023 g0a3b
Logistics 09 00023 g0a4
Figure A4. Transfers blue colored cells in Column M to green cells above.
Figure A4. Transfers blue colored cells in Column M to green cells above.
Logistics 09 00023 g0a4

Appendix B

Interest in improving the data analysis process arose following the automated scripts process, which, although automated, still required significant time for anomaly detection and data cleaning. As a result, neural network techniques using the Python code in Figure A5 were applied for data cleaning with the original dataset. A neural network model was created using freely available open-source resources. A preliminary algorithm was developed that used color coding to highlight duplicative and incomplete data. The results were then reviewed by a human subject matter expert. The algorithm was then updated using a second set of training data from the original dataset, and was run again on the large dataset. Redundant data were deleted and a tracer record was created for audit and compliance testing and analysis. Missing and incomplete data were flagged by the neural net so that the dataset could updated with the missing information. Machine learning was used to propose potential data to complete the data records, based on contextual information and data in the dataset. Process improvements in terms of time differences were recorded, comparing the use of the automated scripts and the neural nets algorithm. The Python code developed is shown in Figure A5.
Logistics 09 00023 g0a5aLogistics 09 00023 g0a5b
Figure A5. Neural Network model for data cleaning.
Figure A5. Neural Network model for data cleaning.
Logistics 09 00023 g0a5aLogistics 09 00023 g0a5b

References

  1. Dale, R.F.; Gross, L. The Arctic: Last Frontier for Energy and Mineral Exploitation? In Handbook on International Development and the Environment; Edward Elgar Publishing: Cheltenham, UK, 2023; pp. 154–169. Available online: https://www.elgaronline.com/edcollchap/book/9781800883789/book-part-9781800883789-18.xml (accessed on 4 November 2024).
  2. Kuletz, K.J.; Ferguson, S.H.; Frederiksen, M.; Gallagher, C.P.; Hauser, D.D.; Hop, H.; Kovacs, K.M.; Lydersen, C.; Mosbech, A.; Seitz, A.C. A review of climate change impacts on migration patterns of marine vertebrates in Arctic and Subarctic ecosystems. Front. Environ. Sci. 2024, 12, 1434549. [Google Scholar] [CrossRef]
  3. Vogler, R.; Stoll, C. Transforming local Arctic tourism businesses in times of multiple crises: A post-pandemic perspective. Polar Geogr. 2024, 47, 33–48. [Google Scholar] [CrossRef]
  4. Hu, W.; Cervone, G.; Trusel, L.; Yu, M. Arctic accessibility: Recent trend in observed ship tracks and validation of Arctic transport accessibility model. Ann. GIS 2024, 30, 455–4740. [Google Scholar] [CrossRef]
  5. World Wildlife Fund. Safety at the Helm: A Plan for Smart Shipping Through the Bering Strait; World Wildlife Fund-Anchorage: Anchorage, Alaska, 2020; Available online: https://www.worldwildlife.org/pages/safety-at-the-helm-a-plan-for-smart-shipping-through-the-bering-strait (accessed on 8 September 2024).
  6. Northern Sea Route Information Office. NSR Shipping Traffic—Activities in May 2022; Nord University: Bodø, Norway, 2022; Available online: https://arctic-lio.com/nsr-shipping-traffic-activities-in-may-2022/ (accessed on 8 September 2024).
  7. Humpert, First Panamax Containership Sprints Across Arctic Reaching China in Just Three Weeks. gCaptain. 25 September 2024. Available online: https://gcaptain.com/first-panamax-containership-sprints-across-arctic-reaching-china-in-just-three-weeks/ (accessed on 1 November 2024).
  8. Staalesen, A. Chinese Heavy Lift Ship Without Ice-Class Battles on Freezing Arctic Route. Barents Observer. 30 October 2024. Available online: https://www.thebarentsobserver.com/news/chinese-heavy-lift-ship-without-iceclass-battles-on-freezing-arctic-route/419719 (accessed on 2 November 2024).
  9. National Research Council; Transportation Research Board; Marine Board; Division on Earth; Life Studies; Polar Research Board; Ocean Studies Board; Committee on Responding to Oil Spills in the US Arctic Marine Environment. Responding to Oil Spills in the US Arctic Marine Environment; National Academies Press: Cambridge, MA, USA, 2014. [CrossRef]
  10. O’Rourke, R.; Comay, L.B.; Frittelli, J.; Kaboli, E.; Keating-Bitonti, C.; Marshak, A.R.; Ramseur, J.L.; Ryan, L.; Sheikh, P.A. Changes in the Arctic: Background and Issues for Congress; Report R41153; Congressional Research Service: Washington, DC, USA, 2024. Available online: https://crsreports.congress.gov/product/details?prodcode=R41153 (accessed on 8 August 2024).
  11. Johannsdottir, L.; Cook, D. Systemic risk of maritime-related oil spills viewed from an Arctic and insurance perspective. Ocean Coast. Manag. 2019, 179, 104853. [Google Scholar] [CrossRef]
  12. Das, T.; Goerlandt, F.; Pelot, R. A mixed integer programming approach to improve oil spill response resource allocation in the Canadian Arctic. Multimodal Transp. 2024, 3, 100110. [Google Scholar] [CrossRef]
  13. Zhong, Z.; You, F. Oil spill response planning with consideration of physicochemical evolution of the oil slick: A multiobjective optimization approach. Comput. Chem. Eng. 2011, 35, 1614–1630. [Google Scholar] [CrossRef]
  14. Matveev, A.; Bogdanova, E. Functional model of an intelligent decision support system for responding to transport emergencies in the Arctic zone. Transp. Res. Procedia 2021, 57, 363–369. [Google Scholar] [CrossRef]
  15. Fingas, M.F.; Hollebone, B.P. Review of behaviour of oil in freezing environments. Mar. Pollut. Bull. 2003, 47, 333–340. [Google Scholar] [CrossRef] [PubMed]
  16. Paardenkooper, K. The role of data-driven logistics in Arctic shipping. In Arctic Maritime Logistics: The Potentials and Challenges of the Northern Sea Route; Springer International Publishing: Cham, Switzerland, 2022; pp. 173–191. Available online: https://link.springer.com/chapter/10.1007/978-3-030-92291-7_10 (accessed on 1 November 2024).
  17. Andreassen, N.; Borch, O.J. (Eds.) Crisis and Emergency Management in the Arctic: Navigating Complex Environments; Routledge: London, UK, 2020. [Google Scholar]
  18. Camur, M.C.; Sharkey, T.C.; Dorsey, C.; Grabowski, M.R.; Wallace, W.A. Optimizing the Response for Arctic Mass Rescue Events. Transp. Res. Part E Logist. Transp. Rev. 2021, 152, 102368. [Google Scholar] [CrossRef]
  19. Kelman, I.; Loe, J.S.P.; Rowe, E.W.; Wilson, E.; Poussenkova, N.; Nikitina, E.; Fjærtoft, D.B. Local perceptions of corporate social responsibility for Arctic petroleum in the Barents region. Arct. Rev. Law Politics 2016, 2, 152–178. [Google Scholar] [CrossRef]
  20. Grabowski, M.R.; Rizzo, C.; Graig, T. Data challenges in dynamic, large-scale resource allocation in remote regions. Saf. Sci. 2016, 87, 76–86. [Google Scholar] [CrossRef]
  21. Barker, C.H.; Kourafalou, V.H.; Beegle-Krause, C.; Boufadel, M.; Bourassa, M.A.; Buschang, S.G.; Androulidakis, Y.; Chassignet, E.P.; Dagestad, K.-F.; Danmeier, D.G.; et al. Progress in operational modeling in support of oil spill response. J. Mar. Sci. Eng. 2020, 8, 668. [Google Scholar] [CrossRef]
  22. Yang, Z.; Chen, Z.; Lee, K.; Owens, E.; Boufadel, M.C.; An, C.; Taylor, E. Decision support tools for oil spill response (OSR-DSTs): Approaches, challenges, and future research perspectives. Mar. Pollut. Bull. 2021, 167, 112313. [Google Scholar] [PubMed]
  23. U.S. Government Accountability Office. Coast Guard: Complete Performance and Operational Data Would Clarify Arctic Resource Needs. GAO-24-106491; U.S. Government Accountability Office: Washington, DC, USA, 2024. Available online: https://www.gao.gov/products/gao-24-106491 (accessed on 5 September 2024).
  24. Hanga, K.M.; Kovalchuk, Y. Machine learning and multi-agent systems in oil and gas industry applications: A survey. Comput. Sci. Rev. 2019, 34, 100191. [Google Scholar] [CrossRef]
  25. Odimarha, A.C.; Ayodeji, S.A.; Abaku, E.A. Machine learning’s influence on supply chain and logistics optimization in the oil and gas sector: A comprehensive analysis. Comput. Sci. IT Res. J. 2024, 5, 725–740. [Google Scholar] [CrossRef]
  26. Kruke, B.I.; Auestad, A.C. Emergency preparedness and rescue in Arctic waters. Saf. Sci. 2021, 136, 105163. [Google Scholar] [CrossRef]
  27. American Petroleum Institute; National Oceanic & Atmospheric Administration; U.S. Coast Guard; U.S. Environmental Protection Agency. A Guide for Spill Response Planning in Marine Environments; U.S. Coast Guard: Washington, DC, USA, 2001. Available online: https://homeport.uscg.mil/Lists/Content/Attachments/75984/Guide%20for%20Spill%20Response%20Planning%20in%20Marine%20Environment.pdf (accessed on 20 August 2024).
  28. Wright, S.K.; Wilkin, S.M.; Jensen, A.S.; Rowles, T.K.; Dushane, J.; Ziccardi, M. Arctic Marine Mammal Disaster Response Guidelines: National Marine Fisheries Service Guidance Report; National Oceanic and Atmospheric Administration: Washington, DC, USA, 2017. Available online: https://repository.library.noaa.gov/view/noaa/16986/noaa_16986_DS3.pdf (accessed on 4 November 2024).
  29. Roud, E.; Gausdal, A.H. Trust and emergency management: Experiences from the Arctic Sea region. J. Trust Res. 2019, 9, 203–225. [Google Scholar] [CrossRef]
  30. Ocean Conservancy. Arctic Alaska: Oil Spill Prevention and Response Requirements Primer; Nuka Research & Planning Group: Seldovia, Alaska, 2020; Available online: https://oceanconservancy-org.webpkgcache.com/doc/-/s/oceanconservancy.org/wp-content/uploads/2020/07/200702-OC-Arctic-AK-Guide-SCREEN-vf.pdf (accessed on 20 August 2024).
  31. Grabowski, M.R.; Martelli, P.F.; Roberts, K.H. Reliability-Seeking virtual organizations at the margins of systems, resources and capacity. Saf. Sci. 2023, 168, 106327. [Google Scholar] [CrossRef]
  32. Farshadfar, Z.; Mucha, T.; Tanskanen, K. Leveraging Machine Learning for Advancing Circular Supply Chains: A Systematic Literature Review. Logistics 2024, 8, 108. [Google Scholar] [CrossRef]
  33. Ranran, L.; Lv, Z.; Dang, S.; Su, T.; Li, X. Application of machine learning in ocean data. Multimed. Syst. 2023, 29, 1815–1824. [Google Scholar] [CrossRef]
  34. Schneider, C.; Barker, A.; Dobson, S. A survey of self-healing systems frameworks. Softw. Pract. Exp. 2015, 45, 1375–1398. [Google Scholar]
  35. Mozaffari, M.; Dignös, A.; Gamper, J.; Störl, U. Self-tuning Database Systems: A Systematic Literature Review of Automatic Database Schema Design and Tuning. ACM Comput. Surv. 2024, 56, 277. [Google Scholar] [CrossRef]
  36. Panwar, V. AI-driven query optimization: Revolutionizing database performance and efficiency. Int. J. Comput. Trends Technol. 2024, 72, 18–26. [Google Scholar] [CrossRef]
  37. Al-Jumeily, D.; Hussain, A.; Fergus, P. Using adaptive neural networks to provide self-healing autonomic software. Int. J. Space-Based Situated Comput. 2015, 5, 129–140. [Google Scholar] [CrossRef]
  38. Ahmed, S.; Ahamed, S.I.; Sharmin, M.; Haque, M.M. Self-healing for autonomic pervasive computing. In Proceedings of the 2007 ACM Symposium on Applied Computing, Seoul, Republic of Korea, 11–15 March 2007; pp. 110–111. [Google Scholar] [CrossRef]
  39. Shukla, S.K.; Pant, B.; Viriyasitavat, W.; Verma, D.; Kautish, S.; Dhiman, G.; Kaur, A.; Srihari, K.; Mohanty, S.N. An integration of autonomic computing with multicore systems for performance optimization in Industrial Internet of Things. IET Commun. 2022. early view. [Google Scholar] [CrossRef]
  40. Mohammadiun, S.; Hu, G.; Gharahbagh, A.A.; Li, J.; Hewage, K.; Sadiq, R. Intelligent computational techniques in marine oil spill management: A critical review. J. Hazard. Mater. 2021, 419, 126425. [Google Scholar] [CrossRef]
  41. Comfort, L.K. Self-organization in complex systems. J. Public Adm. Res. Theory J-PART 1994, 3, 393–410. [Google Scholar]
  42. Shatnawi, A.; Faye, E.; Rima, B.; Al Shara, Z.; Seriai, A.-D. The State of the Art of Emergent Software Systems. IEEE Access 2024, 12, 31808–31823. [Google Scholar] [CrossRef]
  43. Patil, A.; Rane, M. Convolutional neural networks: An overview and its applications in pattern recognition. In Proceedings of the Information and Communication Technology for Intelligent Systems: Proceedings of ICTIS 2020, Ahmedabad, India, 15–16 May 2020; Volume 1, pp. 21–30. [Google Scholar] [CrossRef]
  44. Sekar, J.; Aquilanz, L.L.C. Autonomous Cloud Management Using AI: Techniques for Self-Healing and Self-Optimization. J. Emerg. Technol. Innov. Res. 2023, 11, 571–580. [Google Scholar]
  45. Zawish, M.; Dharejo, F.A.; Khowaja, S.A.; Raza, S.; Davy, S.; Dev, K.; Bellavista, P. AI and 6G into the metaverse: Fundamentals, challenges and future research trends. IEEE Open J. Commun. Soc. 2024, 5, 730–778. [Google Scholar] [CrossRef]
  46. Ghahremani, S.; Giese, H. Evaluation of self-healing systems: An analysis of the state-of-the-art and required improvements. Computers 2020, 9, 16. [Google Scholar] [CrossRef]
  47. U.S. Congress. Oil Pollution Act of 1990 [PDF File]. 1990. Available online: https://www.bsee.gov/sites/bsee.gov/files/federal-register-notice/presentations/opa90.pdf (accessed on 29 February 2024).
  48. Mathew, R. 30 Years Removed, Oil-Spill Liability Insurance’s Evolution Since the 1989 Exxon Valdez Incident. Ocean Coast. Law J. 2024, 29, 25. [Google Scholar]
  49. Alaska Regional Response Team (RRT). Alaska Regional Contingency Plan (Version 2) [PDF]. 2022. Available online: https://nrt.org/sites/176/files/Alaska_RCP_V2_2022FEB.pdf (accessed on 7 November 2024).
  50. Ivshina, I.B.; Kuyukina, M.S.; Krivoruchko, A.V.; Elkin, A.A.; Makarov, S.O.; Cunningham, C.J.; Peshkur, T.A.; Atlas, R.M.; Philp, J.C.; 2015. Oil spill problems and sustainable response strategies through new technologies. Environ. Sci. Process. Impacts 2015, 17, 1201–1219. [Google Scholar] [CrossRef]
  51. Lee, G.Y.; Alzamil, L.; Doskenov, B.; Termehchy, A. A survey on data cleaning methods for improved machine learning model performance. arXiv 2021, arXiv:2109.07127. [Google Scholar]
  52. Ash, J. Flying against the clock–risk management and resilience in Arctic search and rescue and casualty evacuation flights. Saf. Extrem. Environ. 2023, 5, 79–89. [Google Scholar] [CrossRef]
  53. Wilkinson, J.; Beegle-Krause, C.J.; Evers, K.U.; Hughes, N.; Lewis, A.; Reed, M.; Wadhams, P. Oil spill response capabilities and technologies for ice-covered Arctic marine waters: A review of recent developments and established practices. Ambio 2017, 46 (Suppl. S3), S423–S441. [Google Scholar] [CrossRef] [PubMed]
  54. Clark, D.G.; Ford, J.D.; Tabish, T. What role can unmanned aerial vehicles play in emergency response in the Arctic: A case study from Canada. PLoS ONE 2018, 13, e0205299. [Google Scholar] [CrossRef]
  55. Knol, M.; Arbo, P. Oil spill response in the Arctic: Norwegian experiences and future perspectives. Mar. Policy 2014, 50, 171–177. [Google Scholar] [CrossRef]
  56. Turoff, M.; Chumer, M.; de Walle, B.V.; Yao, X. The design of a dynamic emergency response management information system (DERMIS). J. Inf. Technol. Theory Appl. (JITTA) 2004, 5, 3. [Google Scholar]
  57. Hristidis, V.; Chen, S.C.; Li, T.; Luis, S.; Deng, Y. Survey of data management and analysis in disaster situations. J. Syst. Softw. 2010, 83, 1701–1714. [Google Scholar] [CrossRef]
  58. McMillan, L.; Varga, L. Towards self-healing in water infrastructure systems. Proc. Inst. Civ. Eng.-Smart Infrastruct. Constr. 2022, 176, 53–61. [Google Scholar] [CrossRef]
  59. Hazel, W.E.B. Leveraging Advanced Information Technologies and Artificial Intelligence Applications to Enhance Situational Awareness: New and Future Models for Oil Spill Prevention, Preparedness and Response. In Proceedings of the International Oil Spill Conference Proceedings, New Orleans, LA, USA, 13–16 May 2024; Allen Press: Lawrence, KS, USA, 2024. [Google Scholar]
  60. Remil, Y.; Bendimerad, A.; Mathonat, R.; Kaytoue, M. AIops solutions for incident management: Technical guidelines and a comprehensive literature review. arXiv 2024, arXiv:2404.01363. [Google Scholar]
  61. Edwards, P.N.; Mayernik, M.S.; Batcheller, A.L.; Bowker, G.C.; Borgman, C.L. Science Friction: Data, Metadata, and Collaboration. Soc. Stud. Sci. 2011, 41, 667–690. [Google Scholar] [CrossRef]
  62. Cucinelli, J.; Goerlandt, F.; Pelot, R. Exploring Risk Governance Deficits for Marine Oil Spill Preparedness and Response in Canada. In Area-Based Management of Shipping: Canadian and Comparative Perspectives; Springer Nature: Cham, Switzerland, 2024; pp. 227–260. [Google Scholar] [CrossRef]
  63. Deng, F.; Tao, X.; Wei, P.; Shiyin Wei, S. A robust deep learning-based damage identification approach for SHM considering missing data. Appl. Sci. 2023, 13, 5421. [Google Scholar] [CrossRef]
  64. Jerez, J.M.; Molina, I.; García-Laencina, P.J.; Alba, E.; Ribelles, N.; Martín, M.; Franco, L. Missing data imputation using statistical and machine learning methods in a real breast cancer problem. Artif. Intell. Med. 2010, 50, 105–115. [Google Scholar] [CrossRef]
  65. Angelopoulos, A.; Michailidis, E.T.; Nomikos, N.; Trakadas, P.; Hatziefremidis, A.; Voliotis, S.; Zahariadis, T. Tackling faults in the industry 4.0 era—A survey of machine-learning solutions and key aspects. Sensors 2019, 20, 109. [Google Scholar] [CrossRef]
  66. Lakshminarayan, K.; Harp, S.A.; Samad, T. Imputation of missing data in industrial databases. Appl. Intell. 1999, 11, 259–275. [Google Scholar] [CrossRef]
Logistics 09 00023 g001
Figure 1. U.S. Coast Guard Arctic Area of Responsibility [23].
Figure 1. U.S. Coast Guard Arctic Area of Responsibility [23].
Logistics 09 00023 g001
Table 1. Data attributes.
Table 1. Data attributes.
Data IdentificationData Attributes
1Vessel ID (MMIC)
2Vessel Name
3Last port of call (LPOC)
4Next port of call (NPOC)
5Intended route
6Estimated date of arrival
7Estimated time of arrival
8Fuel oil type
9Fuel oil quantity
10Lube oil type
11Lube oil quantity
12Cargo on board (type)
13Cargo volume
14Location of last fuel received
15Exhaust scrubber installed? (yes/no)
16Vessel contact information—Email
17Vessel contact information—Phone
18Intended route deviating from routing measures? (yes/no)
19On board Automated Identification System functioning? (yes/no)
20On board Automated Identification System tested? (yes/no)
21On board Automated Identification System date tested
Table 2. Number and types of vessels.
Table 2. Number and types of vessels.
Vessel TypeNumber of Vessels (2023)Percentage
Break Bulk50.2%
Bulk Carrier98147.9%
Bulk/Container592.9%
Container68433.4%
Heavy Lift30.1%
LNG/LPG361.8%
Passenger854.2%
Refrigerated Cargo361.8%
RoRo90.4%
Tug/Offshore Supply Vessel10.0%
Vehicle Carrier1497.3%
Total2048
Table 3. Automated data cleaning results—Oil Spill Response Organization Records, 2023.
Table 3. Automated data cleaning results—Oil Spill Response Organization Records, 2023.
Types of RecordsNumber of RecordsPercentage of Original Dataset
Initial number of records11,568
Duplicative records649056.10%
Incomplete records1181.02%
Incomplete oil records289124.99%
Final dataset (eliminating duplicate, incomplete records)206917.89%
Table 4. Database analysis process.
Table 4. Database analysis process.
PhasesAutomated Anomaly Detection Hours (Scripts)Self Healing Hours
(Neural Nets)		Process Improvement (% Change in Hours)
Data Collection00-
Data Cleaning1202678.33%
Data Analysis1616-
Total1364269.1%
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

McGarvey, J.; Grabowski, M.R.; Custard, B.; Gabelein, S. Self-Healing Databases for Emergency Response Logistics in Remote and Infrastructure-Poor Settings. Logistics 2025, 9, 23. https://doi.org/10.3390/logistics9010023

AMA Style

McGarvey J, Grabowski MR, Custard B, Gabelein S. Self-Healing Databases for Emergency Response Logistics in Remote and Infrastructure-Poor Settings. Logistics. 2025; 9(1):23. https://doi.org/10.3390/logistics9010023

Chicago/Turabian Style

McGarvey, James, Martha R. Grabowski, Buddy Custard, and Steven Gabelein. 2025. "Self-Healing Databases for Emergency Response Logistics in Remote and Infrastructure-Poor Settings" Logistics 9, no. 1: 23. https://doi.org/10.3390/logistics9010023

APA Style

McGarvey, J., Grabowski, M. R., Custard, B., & Gabelein, S. (2025). Self-Healing Databases for Emergency Response Logistics in Remote and Infrastructure-Poor Settings. Logistics, 9(1), 23. https://doi.org/10.3390/logistics9010023

Article Metrics

Back to TopTop