Original Research

Definition & Purpose

Original Research articles report the findings of primary empirical investigations that generate new knowledge relevant to artificial intelligence in healthcare. These submissions must present novel data, methods, or analyses not previously published, and must make a clear and direct contribution to the evidence base for AI in clinical or public health settings.

Scope

NJAIH welcomes Original Research across all methodological paradigms relevant to AI in healthcare, including:

•       Development, training, and validation of machine learning or deep learning models for clinical applications

•       Evaluation of large language models (LLMs) and retrieval-augmented generation (RAG) systems in healthcare

•       Computer vision and medical image analysis studies

•       Natural language processing applied to clinical notes, radiology reports, or public health data

•       AI-assisted diagnostic and prognostic tool evaluation

•       Predictive modeling for disease surveillance and outbreak detection

•       Human factors studies examining clinician or patient interaction with AI systems

•       Implementation science studies of AI tool deployment in real-world health settings

Manuscript Structure

Manuscripts must follow the IMRaD structure unless a compelling methodological justification is provided:

•       Title — concise, specific, and descriptive of the study design and main finding

•       Abstract — structured (Background, Methods, Results, Conclusion); maximum 300 words

•       Introduction — background, knowledge gap, and specific objectives or hypotheses

•       Methods — study design, setting, participants/data, AI model description, evaluation metrics, statistical methods, ethical approvals

•       Results — reported objectively with appropriate tables and figures; no interpretation

•       Discussion — interpretation, comparison with existing literature, limitations, implications

•       Conclusion — brief, grounded in findings; no unsupported claims

•       Data Availability Statement

•       Conflicts of Interest and Funding Declarations

•       References

Reporting Standards

Authors must adhere to applicable reporting guidelines. The following are mandatory where relevant:

•       CONSORT — for randomized controlled trials

•       STROBE — for observational studies

•       TRIPOD — for prediction model development and validation

•       SPIRIT — for study protocols embedded within research reports

•       STARD — for diagnostic accuracy studies

•       CLAIM — Checklist for AI in Medical Imaging studies

Authors must submit the completed reporting checklist as a supplementary file.

Submission Specifications

Ethical Requirements

All Original Research involving human participants or patient data must include a statement of ethical approval from a recognized Institutional Review Board (IRB) or Research Ethics Committee (REC). Studies involving AI models trained on patient data must describe the data governance framework applied. Authors must declare any use of proprietary datasets and their access conditions.

Technical Reports

Definition & Purpose

Technical Reports provide detailed documentation of AI systems, software tools, datasets, benchmarks, or computational pipelines developed for healthcare applications. These articles serve as permanent, citable records of technical innovations and enable reproducibility, replication, and further development by the research community. Technical Reports emphasize the specification and validation of a system or resource rather than the testing of a scientific hypothesis.

Scope

NJAIH accepts Technical Reports describing:

•       Novel AI model architectures or training pipelines for clinical or public health applications

•       Healthcare-specific datasets, benchmarks, or annotated corpora released for community use

•       Software libraries, APIs, or platforms designed to support AI in healthcare workflows

•       Federated learning frameworks and privacy-preserving AI systems for health data

•       Interoperability frameworks and data standards for AI-ready health information systems

•       Evaluation frameworks, scoring rubrics, or metrics developed for healthcare AI assessment

Manuscript Structure

•       Title — should clearly identify the system or resource being described

•       Abstract — unstructured or structured; maximum 250 words

•       Introduction — motivation, existing gap, and contribution of the described system or resource

•       System / Dataset Description — detailed technical specification including architecture, data sources, preprocessing, training, and validation

•       Technical Validation — performance benchmarks, stress tests, or user acceptance testing

•       Usage Notes — access instructions, licensing, dependencies, known limitations

•       Future Development — planned enhancements and maintenance commitments

•       Availability & Access — repository URLs, DOIs, licensing terms

•       Conflicts of Interest and Funding

•       References

Reproducibility Requirements

Technical Reports must meet NJAIH's reproducibility standards:

•       Source code must be deposited in a public repository (GitHub, Zenodo, or equivalent) with a DOI

•       Datasets must include a data dictionary, provenance statement, and access pathway

•       Model weights and configuration files should be shared where clinically and legally permissible

•       Authors must specify the software environment, hardware configuration, and package versions used

Submission Specifications

AI Model Cards & Datasheets

Authors describing AI models are strongly encouraged to include a model card (Mitchell et al., 2019) and, where applicable, a datasheet for datasets (Gebru et al., 2021) as supplementary documents. These transparency artifacts will be published alongside the article.

Systematic Reviews

Definition & Purpose

Systematic Reviews provide comprehensive, reproducible syntheses of the published literature on AI in healthcare. These articles follow pre-specified, transparent methodologies to identify, appraise, and synthesize evidence, minimizing bias and enabling evidence-informed decision making. NJAIH also welcomes Scoping Reviews, Umbrella Reviews, and Living Systematic Reviews.

Types Accepted

•       Systematic Review — with or without pooled meta-analysis

•       Scoping Review — mapping breadth of literature on an emerging AI health topic

•       Umbrella Review (Review of Reviews) — synthesis of existing systematic reviews

•       Rapid Review — abbreviated systematic review with acknowledged methodological compromises

•       Living Systematic Review — continuously updated as new evidence emerges

Pre-registration Requirement

NJAIH strongly recommends and preferentially considers Systematic Reviews pre-registered before data extraction in an approved registry:

•       PROSPERO — required for clinical and health-related reviews

•       Open Science Framework (OSF) — acceptable for reviews outside PROSPERO scope

Authors must provide the registration number in the manuscript. Reviews without pre-registration must provide a justification confirming that data extraction had not commenced prior to submission.

Mandatory Reporting Standards

•       PRISMA 2020 — mandatory for all systematic reviews

•       PRISMA-P — for systematic review protocols

•       MOOSE — for meta-analyses of observational studies

•       PRISMA-ScR — for scoping reviews

The completed PRISMA 2020 checklist and PRISMA flow diagram must be submitted as mandatory supplementary files.

Manuscript Structure

•       Title — must include the study design (e.g., 'A Systematic Review and Meta-analysis')

•       Structured Abstract — Background, Methods (databases and date range), Results, Conclusion; maximum 350 words

•       Introduction — rationale, objectives, and PICO/PICOS framework

•       Methods — eligibility criteria, information sources, search strategy, study selection, data extraction, quality assessment tools, synthesis methods

•       Results — study selection (PRISMA flow diagram), study characteristics, risk of bias, synthesis results

•       Discussion — summary of evidence, limitations, and implications for practice and policy

•       PROSPERO Registration Number

•       Full Search Strategy — mandatory appendix for all databases searched

•       PRISMA Checklist — mandatory supplementary file

Submission Specifications