PRISMA-Guided Survey

• PRISMA-guided survey is a systematic review method that ensures transparency and reproducibility via defined identification, screening, eligibility, and inclusion phases.
• The approach adapts to various domains such as medical imaging, V2X safety, and AI-driven literature reviews by incorporating domain-specific extensions and rigorous audit practices.
• Comprehensive documentation, including search strategies and flowcharts, minimizes bias and supports reproducible and auditable evidence synthesis.

A PRISMA-guided survey is a systematic review or mapping study that explicitly employs the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) reporting standard to ensure transparency, reproducibility, and rigor throughout the process of literature identification, screening, synthesis, and reporting. Originating in evidence-based medicine, PRISMA’s structured methodology has become foundational across fields for constructing auditable corpora, minimizing bias, and documenting methodological choices in large-scale evidence syntheses. In advanced surveys—including those targeting remote sensing, V2X safety, medical imaging, assurance cases, and literature review automation—PRISMA’s principles are adapted and often extended to address domain-specific requirements and enable auditability, particularly in high-stakes or data-rich research domains.

1. PRISMA: Foundations and Standard Workflow

At its core, the PRISMA standard prescribes a staged protocol for transparent literature retrieval and selection:

1. Identification: Exhaustive database and supplemental searches are performed using well-documented Boolean strings. Scope is defined by topic, date interval, and publication type.
2. Screening: Titles, abstracts, and keywords are reviewed; nonrelevant, duplicate, or non-English items are removed.
3. Eligibility: Full-text assessment eliminates studies that do not satisfy inclusion/exclusion criteria (e.g., absence of AI components or methodological detail).
4. Inclusion: The final corpus is explicitly documented—often with a PRISMA flow chart tracing records from identification to inclusion.

This staged approach is accompanied by comprehensive documentation of search parameters, database versions, reviewer assignments, and curation steps. Selection decisions and conflicts are resolved with transparent logs, enabling downstream reproducibility (Rahman et al., 29 Sep 2025, Shahandashti et al., 2023).

2. PRISMA-guided Survey Methodology in Contemporary Research

Beyond the standard four-step process, research domains adapt and extend the PRISMA protocol in several ways:

• Database Coverage: Surveys extend the identification phase across discipline-specific databases (e.g., PubMed, IEEE Xplore, Web of Science for medical imaging (Rahman et al., 29 Sep 2025); IEEE Xplore, ACM DL, Elsevier, Web of Science for V2X (Zhang et al., 29 Nov 2025); Scopus, Engineering Village for assurance cases (Shahandashti et al., 2023)).
• Structured Query Design: Search expressions integrate domain ontologies and systematically combine core topic concepts using logical operators and wildcards, maximizing coverage while maintaining specificity.
• Screening Documentation: Extensive checklists and systematic logging are employed for inclusion/exclusion, with reasons for exclusion typically reported at each screening phase.
• Snowball Sampling: Backward and forward snowballing is iteratively applied (e.g., with connected-papers.com) to ensure saturation and prevent omission of outlier studies (Shahandashti et al., 2023, Rahman et al., 29 Sep 2025).

Selection flows are typically summarized in a PRISMA diagram or in tabular form. For example, Rahman and Lee’s osteoporosis AI survey documents 289 records identified, 97 duplicates removed, 192 screened, 86 full-text assessed, and 40 studies included, detailing stepwise culling (Rahman et al., 29 Sep 2025).

3. Domain-Specific PRISMA Extensions and Innovations

Researchers regularly extend PRISMA to address domain- and technology-specific challenges:

• PRISMA-DFLLM: For systematic literature reviews augmented by LLMs, the PRISMA-DFLLM framework extends standard reporting with items on finetuning dataset construction, LLM hyperparameters, reproducibility, ethical/legal considerations, and living/incremental review processes. Parameter-Efficient Fine-Tuning (PEFT) techniques such as LoRA and QLoRA are documented in explicit checklist items, and measurable improvements in extraction precision and recall are benchmarked pre- and post-finetuning. New sections (16–18, 31, 26e, 30) capture aspects crucial to AI-driven, scalable, and living SLRs, including dataset curation, LLM specification, evaluation methodology, and resource sharing (Susnjak, 2023).
• SPD Taxonomy in V2X Surveys: Systematic PRISMA-guided surveys are combined with unified analytical frameworks—for example, mapping the V2X safety literature onto a Sensor–Perception–Decision (SPD) taxonomy, allowing bibliometric and technical synthesis not only by provenance, but by conceptual layer and cooperation scope. This enables identification of research trends, citation stratification, and evaluation gaps (e.g., few studies report timing, calibration, or human-centered metrics) (Zhang et al., 29 Nov 2025).
• Quality and Risk Assessment: Checklists are adapted to extract extended metadata (dataset public/private, internal/external validation, model architecture, evaluation metrics). While no single risk-of-bias tool is universally adopted, key indicators are systematically tracked (Rahman et al., 29 Sep 2025).
• AI as an Enabler: For emerging SLR automation, PRISMA-driven pipelines directly support LLM fine-tuning, enabling automated screening, extraction, and synthesis under auditable protocols and supporting living review updates, as described by Susnjak et al. (Susnjak, 2023).

4. Representative Applications and Case Studies

Several high-impact applications illustrate the operationalization and impact of PRISMA-guided surveys:

Domain	Survey Examples	Core PRISMA Extensions
Medical Imaging	AI for Osteoporosis (Rahman et al., 29 Sep 2025)	Modality-task-method taxonomy, robust flowcharting, evaluation stratification
V2X Transportation	Cooperative Safety Intelligence (Zhang et al., 29 Nov 2025)	SPD layered mapping, search strategy innovation, audit logs
Assurance Cases	System Assurance Weakeners (Shahandashti et al., 2023)	Formal taxonomy of weakeners, modeling-level mapping, SEGRESS integration
Literature Review Automation	PRISMA-DFLLM (Susnjak, 2023)	LLM-specific checklist, living review, dataset/model card availability

In each, the PRISMA workflow is customized to yield domain-appropriate inclusion/exclusion, synthesis, and framework mapping, with special attention to dataset transparency, workflow reproducibility, and evaluation metric justification.

5. Advantages and Systemic Challenges

Strengths:

• Transparency and Auditability: Every methodological decision, from database choice to reviewer adjudication, is explicitly logged and traceable.
• Reproducibility: Structured flow diagrams and stepwise logging facilitate third-party verification and updates, which is especially critical for living reviews (Rahman et al., 29 Sep 2025, Susnjak, 2023).
• Bias Minimization: Dual screening, explicit inclusion/exclusion, and standardized metadata extraction reduce the risk of selection and publication bias.

Challenges:

• Manual Bottlenecks: Traditional SLRs require intensive human input across screening, extraction, and synthesis, often leading to static snapshots that rapidly become outdated (Susnjak, 2023).
• Heterogeneity Handling: Variability in paper structure, data reporting, and terminology complicates automation and quality scoring.
• Living Reviews: Updating corpora necessitates robust versioning, incremental data extraction, and explicit tracking of evidence changes (as with “Checklist Item 26e” in PRISMA-DFLLM).
• Coverage vs. Specificity: Balancing recall and precision in search strategy design remains a methodological challenge; complex Boolean logic is often required.

6. Emerging Directions and Open Questions

Several developments and future paths are charted by recent studies:

• LLM-Driven Systematic Reviews: The incorporation of domain-specific LLMs, fine-tuned on gold-standard PRISMA-curated corpora, is rapidly advancing automation and scalability in the SLR workflow. Open questions concern optimal representation of non-textual data for finetuning, instruction-pair sufficiency for LLM expertise, and bias quantification under automation (Susnjak, 2023).
• Framework and Checklist Evolution: Expansion of PRISMA standards to capture living review updates, dataset/model card availability, and legal/ethical dimensions is ongoing.
• Formal Ontologies and Taxonomies: The use of formal models (SPD, tri-axial modality-task-method maps, weakeners taxonomies) aids in unifying highly heterogeneous literatures and illuminating research gaps (Zhang et al., 29 Nov 2025, Shahandashti et al., 2023).
• AI and Tool Support for Modeling: Automated detection and mitigation of logical, epistemic, and ontological weakeners in assurance cases, as well as defect identification in other engineering domains, are targeted for AI-assisted tooling built atop PRISMA-curated corpora (Shahandashti et al., 2023).

A plausible implication is the ongoing convergence of methodological rigor (PRISMA), computational scalability (LLMs, automated extraction), and domain formalization (taxonomies and ontologies), yielding a reproducible, living knowledge synthesis culture across technical disciplines.

7. Summary and Impact

A PRISMA-guided survey constitutes a rigorously defined, auditable, and often extensible methodology for capturing, analyzing, and synthesizing large bodies of technical literature. Its protocols ensure transparent reporting and reproducibility while supporting innovation through domain-specific adaptation and technological augmentation (e.g., LLM-based automation). Key applications in medical imaging, intelligent transportation, engineering assurance, and systematic evidence synthesis demonstrate its centrality for high-confidence, scalable, and continually updated research practices. Open challenges persist in automation, heterogeneity management, and incremental update workflows, but ongoing methodological and computational innovations continue to extend PRISMA’s impact across the scientific landscape (Rahman et al., 29 Sep 2025, Zhang et al., 29 Nov 2025, Shahandashti et al., 2023, Susnjak, 2023, Ferrari et al., 14 Nov 2025).