AI-Enhanced Point-of-Care Diagnostics for Infectious Diseases in Resource-Limited Settings: A Scoping Review

Preprint v3 (May 2026). Version 3 renumbers the seven added figures from R1–R7 (in versions 1 and 2) to Figures 10–16, extending the main-text figure numbering (Figures 1–9) introduced in v1. All other content (analyses, text, references, supplementary materials) is identical to v2 (https://doi.org/10.5281/zenodo.20278223, posted 19 May 2026). Versions of this preprint to date: v1 (https://doi.org/10.5281/zenodo.19446485, 7 April 2026) — original release v2 (https://doi.org/10.5281/zenodo.20278223, 19 May 2026) — added four post-hoc explanatory cross-tabulations on the same 237-study extraction set v3 (this version, 19 May 2026) — figure renumbering only The four post-hoc cross-tabulations added in v2 convert the synthesis from a descriptive map of the field to a diagnostic analysis of why this literature produces studies prolifically but rarely converts them into prospective field-validated tools. Four findings: The conventional "HIC develops, LMIC tests" framing is not supported in this corpus — high-income-country-affiliated studies skew toward later-stage validation, and low-income-country-affiliated studies report the highest prospective-field share (44%). The top 50 studies by citation count contain no lightweight architectures, and heavy architectures (ResNet, EfficientNet, Vision Transformer family) reach 0% prospective field validation across the full corpus, while the three lightweight-architecture studies reach 33%. Cost reporting is flat at 37–38% across all validation maturity stages — it does not mature with the validation pipeline. Disease coverage figures are partly an artefact of reporting completeness, with HIV over-represented and tuberculosis under-represented in the complete-case subset by 12 percentage points each. This scoping review systematically maps 237 studies on AI-enhanced point-of-care diagnostics for infectious diseases in resource-limited settings, published between 2015 and early 2026, following JBI methodology and PRISMA-ScR guidelines with a registered protocol (OSF: https://doi.org/10.17605/OSF.IO/KV8MP). The manuscript includes 16 figures (9 original + 7 added in v2), 54 references, and is accompanied by 5 supplementary materials deposited at OSF.

Authors

Institutions

Publication Details

Journal
Zenodo (CERN European Organization for Nuclear Research)
Published
2026-05-19
DOI
https://doi.org/10.5281/zenodo.20278758
Primary Topic
COVID-19 diagnosis using AI
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
preprint

AI-Enhanced Point-of-Care Diagnostics for Infectious Diseases in Resource-Limited Settings: A Scoping Review

Hayden Farquhar
Zenodo (CERN European Organization for Nuclear Research)
COVID-19 diagnosis using AI
preprint

AI-Enhanced Point-of-Care Diagnostics for Infectious Diseases in Resource-Limited Settings: A Scoping Review

Hayden Farquhar
preprint en

Abstract

Preprint v3 (May 2026). Version 3 renumbers the seven added figures from R1–R7 (in versions 1 and 2) to Figures 10–16, extending the main-text figure numbering (Figures 1–9) introduced in v1. All other content (analyses, text, references, supplementary materials) is identical to v2 (https://doi.org/10.5281/zenodo.20278223, posted 19 May 2026). Versions of this preprint to date: v1 (https://doi.org/10.5281/zenodo.19446485, 7 April 2026) — original release v2 (https://doi.org/10.5281/zenodo.20278223, 19 May 2026) — added four post-hoc explanatory cross-tabulations on the same 237-study extraction set v3 (this version, 19 May 2026) — figure renumbering only The four post-hoc cross-tabulations added in v2 convert the synthesis from a descriptive map of the field to a diagnostic analysis of why this literature produces studies prolifically but rarely converts them into prospective field-validated tools. Four findings: The conventional "HIC develops, LMIC tests" framing is not supported in this corpus — high-income-country-affiliated studies skew toward later-stage validation, and low-income-country-affiliated studies report the highest prospective-field share (44%). The top 50 studies by citation count contain no lightweight architectures, and heavy architectures (ResNet, EfficientNet, Vision Transformer family) reach 0% prospective field validation across the full corpus, while the three lightweight-architecture studies reach 33%. Cost reporting is flat at 37–38% across all validation maturity stages — it does not mature with the validation pipeline. Disease coverage figures are partly an artefact of reporting completeness, with HIV over-represented and tuberculosis under-represented in the complete-case subset by 12 percentage points each. This scoping review systematically maps 237 studies on AI-enhanced point-of-care diagnostics for infectious diseases in resource-limited settings, published between 2015 and early 2026, following JBI methodology and PRISMA-ScR guidelines with a registered protocol (OSF: https://doi.org/10.17605/OSF.IO/KV8MP). The manuscript includes 16 figures (9 original + 7 added in v2), 54 references, and is accompanied by 5 supplementary materials deposited at OSF.

Zenodo (CERN European Organization for Nuclear Research)
IP Australia (AU)
No poverty
COVID-19 diagnosis using AI
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.