Conference Agenda

Purpose: Access to essential medicines remains a major challenge in developing countries due to inefficient supply chains and weak Logistics Management Information Systems (LMIS) (Stevens & Huys, 2017; WHO, 2010). In Ethiopia, persistent problems related to data quality, infrastructure, workforce capacity, and system integration continue to hinder health commodities supply chain performance despite ongoing health sector reforms. This study employed a grounded theory approach to develop a conceptual model for evaluating LMIS performance in public healthcare facilities in Ethiopia. The model aims to strengthen digital logistics data management, improve supply chain efficiency, and support evidence-based healthcare delivery aligned with Ethiopia’s Health Information Revolution agenda.

Methodology: An institution-based concurrent exploratory mixed-methods design with a qualitative emphasis was employed. The study was conducted across 102 public healthcare facilities in the Amhara Region. Eighty-three key informants, including health facility heads, medical directors, pharmacy heads, supply chain coordinators, district coordinators, and chief executive officers, were purposively selected based on their experience in health commodities supply chain management. Data were collected through in-depth interviews using open-ended questions. Qualitative data were analyzed using grounded theory methodology, including open, axial, and selective coding based on the coding paradigm model of Corbin and Strauss (Corbin J & Strauss A, 2015). MAXQDA version 24.4.1 was used for coding and analysis. Ethical approval was obtained from Addis Ababa University, and participant confidentiality was maintained throughout the study.

Findings and Analysis: The analysis generated 1,427 coding segments, 187 first-level concepts, 64 second-level categories, and 21 high-level categories. The resulting conceptual model identified key factors influencing LMIS performance and health commodities supply chain efficiency. Major causal factors included budget limitations, poor data quality, inadequate workforce training, and LMIS implementation challenges. Contextual conditions such as government commitment, infrastructure availability, public health emergencies, and security concerns also significantly affected system performance. The model further highlighted strategic interventions including digital innovation, LMIS integration, workforce development, and enhanced implementation support. Overall, the model demonstrated that an effective LMIS improves healthcare delivery through timely, reliable, and high-quality data for informed decision-making.

Originality: This study contributes original evidence by developing a grounded theory-based conceptual model for evaluating health commodities LMIS performance in a low-resource healthcare setting. Unlike previous studies that primarily focused on isolated operational challenges, this study integrates causal, contextual, and strategic dimensions into a comprehensive logistics performance framework. The model provides a practical foundation for strengthening digital logistics platforms, improving supply chain data systems, and supporting evidence-based decision-making in healthcare logistics management.

Conclusion and Recommendations: The developed conceptual model provides a practical framework for evaluating and improving LMIS performance in Ethiopian healthcare facilities. The study recommends piloting the model prior to large-scale implementation and conducting further quantitative validation studies to establish measurable key performance indicators for routine LMIS monitoring and evaluation.

Abstract:

Purpose: The purpose of this study is to operationalize Digital Twin (DT) technologies into real-life project-based supply network environments by investigating how practical implementation can improve visibility, enable real-time decision support, and enhance coordination across operations (Ivanov, 2021) (Zaidi, 2024). The study emerges from recurring operational issues in the supply network, such as missing or incorrect materials, and poor adherence to schedules, all of which lead to significant resource waste and delays. The study therefore aims to identify and structure a set of relevant KPIs that capture lean performance, e.g., value creation, waste reduction. These KPIs will be examined in terms of how they can be integrated into a Digital Shadow artifact, enabling real-time monitoring, analytics, and decision support.

Methodology/approach: The study adopted a qualitative approach, and the Design Science Research (DSR) is considered a research methodology which is appropriate for research aimed at creating and evaluating innovative artifacts. DSR methodology combines real-life challenges (relevance) and scientific knowledge (rigor) or expertise in design, development, and evaluation (design) of the artifact (Hevner et al. 2004).

Findings: The results of this study indicate that digital shadow can significantly enhance visualization and real-time situational awareness of material location over supply network. The developed digital shadow enables both real-time material location tracking and detailed pallet-level traceability, thereby improving transparency and control over supply network. However, the findings also highlight that the accuracy and precision requirements for material location may constrain freedom to select IoT solutions. These constraints are particularly influenced by the operational environment and surrounding conditions where artifact is deployed. During the design phase, the digital shadow was conceptualized and iteratively refined. Resulting artifact integrates multiple technologies, including IoT-sensors, wireless communications 5G/GPS networks, cloud-based data platforms, Geographical Information System (GIS), and visualization dashboards. Furthermore, the rigorously developed dashboard incorporates performance analytics, supporting both lean value stream analysis and the monitoring of key performance indicators (KPIs).

Originality: This study focuses on identified a lack of empirical implementation of network level material tracking and dashboard solutions within a digital twin (DT) of supply network. The managerial perspective, study introduces practically relevant KPIs to support lean waste reduction and supply networks operational targets in project business context and provides a dashboard level visibility with real-time situational awareness for decision making. At the network level, study contributes not only to the development of the artifact but also to broader knowledge creation regarding digital shadow applications in material tracking. From a scientific perspective, it contributes by empirically testing Digital Twin (DT) operationalization in a real-life supply network context, bridging the gap between conceptual DT models and their practical implementation.

Verifying engineering requirements is a critical step in railway Signalling projects because delayed or inconsistent compliance assessment can slow certification, postpone system handover, and affect operational readiness. Engineers use a Requirements Verification Matrix to compare stated requirements with submitted evidence and confirm whether each requirement has been satisfied. This process remains labor-intensive, error-prone, and difficult to scale across large projects. The challenge increases when expert-labeled compliance records are scarce, especially for non-compliant cases that require specialist judgment.

This study proposes a semi-supervised adversarial transformer framework for railway compliance verification when only a small number of records have been checked and labeled by compliance experts. The framework uses 636 confirmed compliant records from a railway signalling project and applies six controlled perturbation strategies to construct 636 synthetic non-compliant records. These perturbations include negation insertion, evidence mismatch, scope substitution, meaning inversion, numerical alteration, and standard reference swapping. The resulting 1,272-record dataset serves as a proof-of-concept evaluation setting for testing whether artificial intelligence can distinguish compliant evidence from non-compliant evidence under this controlled setting. Each record is represented in a natural language inference format that combines verification context, the stated requirement, and the submitted evidence.

The proposed model follows a GAN-BERT-inspired training strategy. A transformer encoder learns representations from real compliance records, while a generator produces artificial feature vectors. The discriminator learns from three types of training signals, including expert-labeled real records for compliance classification, real records without expert labels for distribution learning, and generated feature vectors for adversarial regularization. This design enables learning from both labeled and unlabeled records without requiring full expert annotation.

Five transformer architectures are evaluated within this pipeline, namely BERT-base, RoBERTa, DeBERTa, SciBERT, and DistilBERT. The experiment tests nine training settings, where only 8 to 64 records are provided with expert labels. The results show that BERT-base and SciBERT provide stable performance across different training settings, while SciBERT performs strongly when more expert-labeled records are available. DistilBERT reaches a practical safety-oriented threshold with only 14 expert-labeled records, achieving non-compliant recall of 0.901 without training collapse. Within the controlled perturbation-based setting, DistilBERT achieved very high internal performance at 48 expert-labeled records, with macro F1 of 0.998 and non-compliant recall of 1.000. RoBERTa and DeBERTa show higher sensitivity to instability under extreme expert-label scarcity.

The findings suggest that semi-supervised adversarial transformer learning can reduce the number of records that compliance experts need to label while supporting digital assurance workflows for railway Signalling projects. The study contributes an AI-based verification pipeline for scarce-label environments and highlights the importance of matching model capacity to the number of available expert-labeled records. Because the non-compliant cases in this experiment are synthetically constructed, future validation with real project non-compliance records is needed before operational deployment.