Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

From Chaos to Clarity: Schema-Constrained AI for Auditable Biomedical Evidence Extraction from Full-Text PDFs

Pouria Mortezaagha, Joseph Shaw, Bowen Sun, Arya Rahgozar

TL;DR

The paper tackles the challenge of scalable, auditable extraction of synthesis-critical variables from full-text biomedical PDFs. It presents a schema-constrained OCR pipeline that processes documents in page-level chunks, binds outputs to typed fields with closed vocabularies, and attaches sentence-level provenance to key decisions, enabling post-hoc auditing. Evaluated on a DOAC level-measurement corpus (734 PDFs, 7,228 pages), the approach achieves corpus-scale throughput with deterministic merging, high internal consistency, and substantial fidelity gains after iterative schema refinement. The resulting reproducible CSV/Parquet datasets and multimodal markdown reconstructions support transparent, scalable biomedical evidence synthesis, demonstrating that constrained inference and provenance-aware extraction can meet the transparency, reliability, and auditability requirements of modern systematic reviews.

Abstract

Biomedical evidence synthesis relies on accurate extraction of methodological, laboratory, and outcome variables from full-text research articles, yet these variables are embedded in complex scientific PDFs that make manual abstraction time-consuming and difficult to scale. Existing document AI systems remain limited by OCR errors, long-document fragmentation, constrained throughput, and insufficient auditability for high-stakes synthesis. We present a schema-constrained AI extraction system that transforms full-text biomedical PDFs into structured, analysis-ready records by explicitly restricting model inference through typed schemas, controlled vocabularies, and evidence-gated decisions. Documents are ingested using resume-aware hashing, partitioned into caption-aware page-level chunks, and processed asynchronously under explicit concurrency controls. Chunk-level outputs are deterministically merged into study-level records using conflict-aware consolidation, set-based aggregation, and sentence-level provenance to support traceability and post-hoc audit. Evaluated on a corpus of studies on direct oral anticoagulant level measurement, the pipeline processed all documents without manual intervention, maintained stable throughput under service constraints, and exhibited strong internal consistency across document chunks. Iterative schema refinement substantially improved extraction fidelity for synthesis-critical variables, including assay classification, outcome definitions, follow-up duration, and timing of measurement. These results demonstrate that schema-constrained, provenance-aware extraction enables scalable and auditable transformation of heterogeneous scientific PDFs into structured evidence, aligning modern document AI with the transparency and reliability requirements of biomedical evidence synthesis.

From Chaos to Clarity: Schema-Constrained AI for Auditable Biomedical Evidence Extraction from Full-Text PDFs

TL;DR

The paper tackles the challenge of scalable, auditable extraction of synthesis-critical variables from full-text biomedical PDFs. It presents a schema-constrained OCR pipeline that processes documents in page-level chunks, binds outputs to typed fields with closed vocabularies, and attaches sentence-level provenance to key decisions, enabling post-hoc auditing. Evaluated on a DOAC level-measurement corpus (734 PDFs, 7,228 pages), the approach achieves corpus-scale throughput with deterministic merging, high internal consistency, and substantial fidelity gains after iterative schema refinement. The resulting reproducible CSV/Parquet datasets and multimodal markdown reconstructions support transparent, scalable biomedical evidence synthesis, demonstrating that constrained inference and provenance-aware extraction can meet the transparency, reliability, and auditability requirements of modern systematic reviews.

Abstract

Biomedical evidence synthesis relies on accurate extraction of methodological, laboratory, and outcome variables from full-text research articles, yet these variables are embedded in complex scientific PDFs that make manual abstraction time-consuming and difficult to scale. Existing document AI systems remain limited by OCR errors, long-document fragmentation, constrained throughput, and insufficient auditability for high-stakes synthesis. We present a schema-constrained AI extraction system that transforms full-text biomedical PDFs into structured, analysis-ready records by explicitly restricting model inference through typed schemas, controlled vocabularies, and evidence-gated decisions. Documents are ingested using resume-aware hashing, partitioned into caption-aware page-level chunks, and processed asynchronously under explicit concurrency controls. Chunk-level outputs are deterministically merged into study-level records using conflict-aware consolidation, set-based aggregation, and sentence-level provenance to support traceability and post-hoc audit. Evaluated on a corpus of studies on direct oral anticoagulant level measurement, the pipeline processed all documents without manual intervention, maintained stable throughput under service constraints, and exhibited strong internal consistency across document chunks. Iterative schema refinement substantially improved extraction fidelity for synthesis-critical variables, including assay classification, outcome definitions, follow-up duration, and timing of measurement. These results demonstrate that schema-constrained, provenance-aware extraction enables scalable and auditable transformation of heterogeneous scientific PDFs into structured evidence, aligning modern document AI with the transparency and reliability requirements of biomedical evidence synthesis.
Paper Structure (49 sections, 1 equation, 9 figures, 4 tables, 2 algorithms)

This paper contains 49 sections, 1 equation, 9 figures, 4 tables, 2 algorithms.

Table of Contents

  1. Related Work
  2. Structural reconstruction and early scholarly PDF extraction
  3. From heuristic rules to feature-based learning
  4. Deep learning for layout and multimodal structure
  5. Layout-aware transformers and relational document modeling
  6. Semantic extraction with scientific language models
  7. OCR-driven and end-to-end transcription approaches
  8. Schema-constrained and evidence-centric extraction
  9. Introduction
  10. Methods
  11. Overview of the computational pipeline
  12. Document corpus and ingestion
  13. Schema–constrained AI inference design
  14. Payload blocks and target constructs
  15. Sentence–level evidence for auditability
  16. ...and 34 more sections

Figures (9)

  • Figure 1: From unstructured scientific PDFs to structured, provenance-linked evidence.
  • Figure 2: Structural challenges of scientific PDFs for evidence extraction.
  • Figure 3: Overview of the schema-constrained OCR pipeline. PDFs are ingested and chunked, processed asynchronously with schema-constrained OCR, merged into study-level records, and exported as structured tables and reviewable outputs.
  • Figure 4: Overview of the ingestion and resume logic. The system discovers PDF files, computes file identifiers and page counts, encodes documents, and uses a persistent index to determine whether each file requires processing. Documents already present in the index are skipped to ensure efficient, incremental execution.
  • Figure 5: Schema-constrained OCR extraction. Page-level OCR calls operate under rate-limiting with retry logic, producing structured outputs aligned with predefined extraction schemas. The system identifies captions, extracts text and figures, and compiles chunk-level annotations for downstream merging.
  • ...and 4 more figures