Year: 2026
Type: preprint
Abstract: Background Systematic review software must coordinate independent human decisions, protocol amendments and heterogeneous information sources. Synchronizing reviewers at every stage can create waiting time unrelated to the scientific assessment of a record. We describe R-SYS, a workflow that links decisions to versioned protocols and allows reviewers to progress independently before a final reconciliation step. Methods The reference implementation combines an eleven-phase guarded lifecycle, immutable locked protocol versions, deterministic rendering of a canonical search strategy, per-reviewer working sets and decision provenance. Each primary reviewer advances through title and abstract screening, full-text assessment, extraction and risk-of-bias appraisal using their own preceding decisions, subject to human adjudication. A final convergence checkpoint resolves outstanding work before synthesis. Artificial intelligence (AI) supplies advisory recommendations that are excluded from human reconciliation. We analyse completion times under explicit scheduling assumptions and describe implementation limitations. Results With identical task sets, continuous reviewer availability and no external waiting, independent lanes achieve a completion time equal to the largest reviewer workload. This is a lower bound for schedules that impose additional synchronization constraints. Disagreement can introduce superseded work and delays; a sufficient condition for retaining an advantage over stage barriers is stated. The implementation provides mechanisms for protocol change control, reviewer coordination and traceable reporting. No empirical measurements of accuracy, efficiency, usability or cost are reported. Conclusions R-SYS provides a reference design for separating reviewer independence from workflow synchronization. Its theoretical scheduling properties do not establish practical time savings, search completeness or AI accuracy. A separate, prospectively specified validation study is planned after completed reviews using the platform become available.
Source: Research Square
Authors: Yehya Tlaiss, Mohamad Ismail
Language: English
Subfield: Statistics, Probability and Uncertainty
Field: Decision Sciences
Domain: Social Sciences
Open Access status: green
OpenAlex ID: W7235918264
